diff --git a/ADL/metrics/alderlake_metrics_goldencove_core.json b/ADL/metrics/alderlake_metrics_goldencove_core.json index 96014476..d72eec5d 100644 --- a/ADL/metrics/alderlake_metrics_goldencove_core.json +++ b/ADL/metrics/alderlake_metrics_goldencove_core.json @@ -2,8 +2,8 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for 12th and 13th Generation Intel(R) Core(TM) Processor0", - "DatePublished": "11/12/2024", - "Version": "0", + "DatePublished": "11/15/2024", + "Version": "1.0", "Legend": "", "TmaVersion": "5.01", "TmaFlavor": "Full" @@ -111,7 +111,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Mispredictions > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Mispredictions" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Mispredictions > 20", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP", @@ -199,7 +207,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Big_Code > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Big_Code" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Big_Code > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB", @@ -356,7 +372,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Instruction_Fetch_BW" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", @@ -545,7 +569,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore", @@ -770,7 +802,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Latency" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;Mem;MemoryLat;Offcore", @@ -955,7 +995,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Data_TLBs" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore", @@ -1140,7 +1188,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Synchronization > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Synchronization" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Synchronization > 10", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;LockCont;Mem;Offcore", @@ -1228,7 +1284,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Compute_Bound_Est" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20", + "ThresholdIssues": "$issueComp" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Cor", @@ -1437,7 +1501,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Irregular_Overhead > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Irregular_Overhead" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Irregular_Overhead > 10", + "ThresholdIssues": "$issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BvIO;Cor;Ret", @@ -1830,7 +1902,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Other_Bottlenecks" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;Cor;Offcore", @@ -1866,7 +1946,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Branching_Overhead > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Branching_Overhead" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Bottleneck_Branching_Overhead > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Ret", @@ -1930,7 +2018,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Useful_Work > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Useful_Work" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Useful_Work > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;Ret", @@ -1974,7 +2070,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;BvIO;TmaL1;PGO", @@ -2023,7 +2127,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Frontend;TmaL2", @@ -2052,7 +2168,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss", @@ -2085,7 +2217,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -2114,7 +2266,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -2143,7 +2315,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB", @@ -2176,7 +2364,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2205,7 +2413,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2242,7 +2470,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2279,7 +2531,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2312,7 +2588,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -2369,7 +2661,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Mispredicts_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP", @@ -2426,7 +2738,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Clears_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;MachineClears", @@ -2455,7 +2787,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Unknown_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat", @@ -2492,7 +2844,23 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC, $issueMS, $issueMV, $issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MicroSeq", @@ -2521,7 +2889,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LCP(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -2550,7 +2934,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchLat", @@ -2599,7 +2999,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;Frontend;TmaL2", @@ -2645,7 +3053,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -2691,7 +3111,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Decoder0_Alone(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -2737,7 +3173,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;FetchBW", @@ -2783,7 +3231,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....LSD(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LSD(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....LSD(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;LSD", @@ -2837,7 +3297,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 5 & b > 20", + "BaseFormula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -2881,7 +3353,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", @@ -2922,7 +3402,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2", @@ -2975,7 +3467,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;BrMispredicts", @@ -3024,7 +3532,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueMC, $issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2", @@ -3081,7 +3601,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Nukes(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;Machine_Clears", @@ -3117,7 +3653,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;TmaL1", @@ -3158,7 +3702,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20", + "BaseFormula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2", @@ -3191,7 +3747,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1, $issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -3232,7 +3804,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -3273,7 +3865,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3302,7 +3918,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3343,7 +3983,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3384,7 +4052,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3425,7 +4121,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3454,7 +4178,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Fwd_Blk(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3504,7 +4248,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L1_Latency_Dependency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -3549,7 +4313,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Lock_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Offcore", @@ -3586,7 +4370,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......Split_Loads(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Loads(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......Split_Loads(%) > 30", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3615,7 +4407,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......FB_Full(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FB_Full(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......FB_Full(%) > 30", + "ThresholdIssues": "$issueBW, $issueSL, $issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW", @@ -3648,7 +4448,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem", @@ -3698,7 +4514,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L2_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryLat", @@ -3731,7 +4567,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -3793,7 +4645,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Contested_Accesses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -3855,7 +4727,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Data_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;Offcore;Snoop", @@ -3905,7 +4797,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L3_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -3938,7 +4850,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......SQ_Full(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 30 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -3967,7 +4899,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -3996,7 +4944,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Bandwidth(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -4029,7 +4997,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat;Offcore", @@ -4058,7 +5046,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -4099,7 +5103,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO, $issueSL, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore", @@ -4141,7 +5165,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......False_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -4183,7 +5227,27 @@ "Category": "TMA", "CountDomain": "Core_Utilization", "Threshold": { - "Formula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4212,7 +5276,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Streaming_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBW;Offcore", @@ -4258,7 +5342,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -4304,7 +5408,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4346,7 +5474,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4400,7 +5552,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4454,7 +5634,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4508,7 +5716,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4549,7 +5785,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2;Compute", @@ -4578,7 +5826,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB", @@ -4607,7 +5871,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4640,7 +5924,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......INT_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4673,7 +5977,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;PortsUtil", @@ -4702,7 +6022,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Slow_Pause(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4731,7 +6071,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......C01_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......C01_Wait(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......C01_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", @@ -4760,7 +6120,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......C02_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......C02_Wait(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......C02_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", @@ -4789,7 +6169,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Memory_Fence(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Memory_Fence(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Memory_Fence(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4862,7 +6262,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4907,7 +6323,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_0(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4936,7 +6372,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_........Mixing_Vectors(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Mixing_Vectors(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Mixing_Vectors(%) > 5", + "ThresholdIssues": "$issueMV" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4965,7 +6409,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_1(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4994,7 +6458,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_2(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -5023,7 +6507,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_3m(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 40 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;PortsUtil", @@ -5077,7 +6581,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........ALU_Op_Utilization(%) > 40" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........ALU_Op_Utilization(%)" + } + ], + "Formula": "a > 40", + "BaseFormula": "metric_TMA_........ALU_Op_Utilization(%) > 40", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5119,7 +6631,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_0(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_0(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_0(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -5161,7 +6681,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_1(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_1(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_1(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5203,7 +6731,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_6(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_6(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_6(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5245,7 +6781,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Load_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Load_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5291,7 +6835,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Store_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Store_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5327,7 +6879,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Retiring(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 70 | b > 10", + "BaseFormula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;TmaL1", @@ -5368,7 +6932,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -5425,7 +6997,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 20 & b > 60", + "BaseFormula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -5470,7 +7054,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......X87_Use(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -5515,7 +7115,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Scalar(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5560,7 +7176,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5609,7 +7241,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5658,7 +7310,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5719,7 +7391,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -5768,7 +7452,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......Int_Vector_128b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Int_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 60", + "BaseFormula": "metric_TMA_......Int_Vector_128b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;IntVector;Pipeline", @@ -5821,7 +7521,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......Int_Vector_256b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Int_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 60", + "BaseFormula": "metric_TMA_......Int_Vector_256b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;IntVector;Pipeline", @@ -5870,7 +7586,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Memory_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -5919,7 +7647,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Fused_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -5972,7 +7712,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Non_Fused_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -6065,7 +7817,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 30 & b > 60", + "BaseFormula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -6114,7 +7878,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Nop_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Pipeline", @@ -6163,7 +7943,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Shuffles_256b(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Shuffles_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Shuffles_256b(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC;Pipeline", @@ -6204,7 +8000,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -6253,7 +8057,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Few_Uops_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6282,7 +8098,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueMC, $issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -6311,7 +8139,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Assists(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO", @@ -6340,7 +8184,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........Page_Faults(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Page_Faults(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Page_Faults(%) > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6369,7 +8221,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........FP_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........FP_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -6398,7 +8258,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........AVX_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........AVX_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........AVX_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -6431,7 +8299,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......CISC(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6524,7 +8408,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_Core_Bound_Likely > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_Core_Bound_Likely" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Botlnk_Core_Bound_Likely > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;SMT", @@ -6551,6 +8443,11 @@ "BaseFormula": " inst_retired.any / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Summary", "LocateWith": "" @@ -6593,7 +8490,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UopPI > 1.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UopPI" + } + ], + "Formula": "a > 1.05", + "BaseFormula": "metric_TMA_Info_Thread_UopPI > 1.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret;Retire", @@ -6637,7 +8542,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UpTB < 6 * 1.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UpTB" + } + ], + "Formula": "a < 6 * 1.5", + "BaseFormula": "metric_TMA_Info_Thread_UpTB < 6 * 1.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW", @@ -6664,6 +8577,11 @@ "BaseFormula": " 1 / tma_info_thread_ipc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Mem", "LocateWith": "" @@ -6685,6 +8603,11 @@ "BaseFormula": " cpu_clk_unhalted.thread", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", "LocateWith": "" @@ -6706,6 +8629,11 @@ "BaseFormula": " topdown.slots:perf_metrics", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", "LocateWith": "" @@ -6740,6 +8668,11 @@ "BaseFormula": " tma_info_thread_slots / ( topdown.slots:percore / 2 ) if smt_on else 1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "SMT;TmaL1", "LocateWith": "" @@ -6765,6 +8698,11 @@ "BaseFormula": " uops_executed.thread / uops_issued.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline", "LocateWith": "" @@ -6803,6 +8741,11 @@ "BaseFormula": " inst_retired.any / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;SMT;TmaL1", "LocateWith": "" @@ -6853,6 +8796,11 @@ "BaseFormula": " ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.256b_packed_single ) / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Flops", "LocateWith": "" @@ -6899,6 +8847,11 @@ "BaseFormula": " ( fp_arith_dispatched.port_0 + fp_arith_dispatched.port_1 + fp_arith_dispatched.port_5 ) / ( 2 * tma_info_core_core_clks )", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -6924,6 +8877,11 @@ "BaseFormula": " uops_executed.thread / uops_executed.thread:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "LocateWith": "" @@ -6949,6 +8907,11 @@ "BaseFormula": " uops_executed.thread / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -6983,6 +8946,11 @@ "BaseFormula": " cpu_clk_unhalted.distributed if smt_on else tma_info_thread_clks", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "SMT", "LocateWith": "" @@ -7009,7 +8977,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpLoad < 3" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpLoad" + } + ], + "Formula": "a < 3", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpLoad < 3", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -7037,7 +9013,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpStore < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpStore" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpStore < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -7065,7 +9049,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpBranch < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpBranch" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpBranch < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;InsType", @@ -7093,7 +9085,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpCall < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpCall" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpCall < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", @@ -7121,7 +9121,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpTB < 6 * 2 + 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpTB" + } + ], + "Formula": "a < 6 * 2 + 1", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpTB < 6 * 2 + 1", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO", @@ -7148,6 +9156,11 @@ "BaseFormula": " br_inst_retired.all_branches / br_inst_retired.near_taken", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", "LocateWith": "" @@ -7186,7 +9199,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpFLOP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -7218,7 +9239,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -7246,7 +9275,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -7274,7 +9311,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -7306,7 +9351,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX128" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -7338,7 +9391,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX256" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -7365,6 +9426,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / cpu_clk_unhalted.pause_inst", "Category": "TMA", "CountDomain": "Inst_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", "LocateWith": "" @@ -7391,7 +9457,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpSWPF" + } + ], + "Formula": "a < 100", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -7414,6 +9488,11 @@ "BaseFormula": " inst_retired.any", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;TmaL1", "LocateWith": "INST_RETIRED.PREC_DIST" @@ -7455,6 +9534,11 @@ "BaseFormula": " ( tma_retiring * tma_info_thread_slots ) / uops_retired.slots:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret", "LocateWith": "" @@ -7481,7 +9565,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Pipeline_Strings_Cycles > 0.1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Pipeline_Strings_Cycles" + } + ], + "Formula": "a > 0.1", + "BaseFormula": "metric_TMA_Info_Pipeline_Strings_Cycles > 0.1", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq;Pipeline;Ret", @@ -7509,7 +9601,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Pipeline_IpAssist < 100000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Pipeline_IpAssist" + } + ], + "Formula": "a < 100000", + "BaseFormula": "metric_TMA_Info_Pipeline_IpAssist < 100000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", @@ -7519,7 +9619,7 @@ "MetricName": "Info_Pipeline_Execute", "LegacyName": "metric_TMA_Info_Pipeline_Execute", "Level": 1, - "BriefDescription": "pub/v3.5", + "BriefDescription": "", "UnitOfMeasure": "", "Events": [ { @@ -7549,6 +9649,11 @@ "BaseFormula": " uops_executed.thread / ( ( uops_executed.core_cycles_ge_1 / 2 ) if smt_on else uops_executed.thread:c1 )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "LocateWith": "" @@ -7574,6 +9679,11 @@ "BaseFormula": " lsd.uops / lsd.cycles_active", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7599,6 +9709,11 @@ "BaseFormula": " idq.dsb_uops / idq.dsb_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7624,6 +9739,11 @@ "BaseFormula": " idq.mite_uops / idq.mite_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7649,6 +9769,11 @@ "BaseFormula": " uops_issued.any / uops_issued.any:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7674,6 +9799,11 @@ "BaseFormula": " lsd.uops / ( uops_issued.any )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;LSD", "LocateWith": "" @@ -7700,7 +9830,19 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 6 > 0.35" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_DSB_Coverage" + }, + { + "Alias": "b", + "Value": "IPC" + } + ], + "Formula": "a < 0.7 & b / 6 > 0.35", + "BaseFormula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 6 > 0.35", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -7727,6 +9869,11 @@ "BaseFormula": " int_misc.unknown_branch_cycles / int_misc.unknown_branch_cycles:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed", "LocateWith": "" @@ -7752,6 +9899,11 @@ "BaseFormula": " dsb2mite_switches.penalty_cycles / dsb2mite_switches.penalty_cycles:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss", "LocateWith": "" @@ -7777,6 +9929,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -7802,6 +9959,11 @@ "BaseFormula": " icache_data.stalls / icache_data.stalls:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", "LocateWith": "" @@ -7828,7 +9990,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret" + } + ], + "Formula": "a < 50", + "BaseFormula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -7855,6 +10025,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / baclears.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed", "LocateWith": "" @@ -7880,6 +10055,11 @@ "BaseFormula": " 1000 * frontend_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -7905,6 +10085,11 @@ "BaseFormula": " 1000 * l2_rqsts.code_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -8036,7 +10221,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Misses > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Misses" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Misses > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -8141,7 +10334,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Bandwidth" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -8229,7 +10430,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_IC_Misses > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_IC_Misses" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Info_Botlnk_IC_Misses > 5", + "ThresholdIssues": "$issueFL" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", @@ -8257,7 +10466,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMispredict" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts", @@ -8285,7 +10502,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8313,7 +10538,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8341,7 +10574,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Ret" + } + ], + "Formula": "a < 500", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8369,7 +10610,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect" + } + ], + "Formula": "a < 1000", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8476,6 +10725,11 @@ "BaseFormula": " tma_mispredictions * tma_info_thread_slots / ( 6 ) / br_misp_retired.all_branches / 100", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "$issueBM" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", "LocateWith": "" @@ -8505,6 +10759,11 @@ "BaseFormula": " int_misc.clears_count / ( br_misp_retired.all_branches + machine_clears.count )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", "LocateWith": "" @@ -8530,6 +10789,11 @@ "BaseFormula": " br_inst_retired.cond_ntaken / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", "LocateWith": "" @@ -8555,6 +10819,11 @@ "BaseFormula": " br_inst_retired.cond_taken / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", "LocateWith": "" @@ -8584,6 +10853,11 @@ "BaseFormula": " ( br_inst_retired.near_call + br_inst_retired.near_return ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8617,6 +10891,11 @@ "BaseFormula": " ( br_inst_retired.near_taken - br_inst_retired.cond_taken - 2 * br_inst_retired.near_call ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8658,6 +10937,11 @@ "BaseFormula": " 1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8683,6 +10967,11 @@ "BaseFormula": " l1d_pend_miss.pending / mem_load_completed.l1_miss_any", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryLat", "LocateWith": "" @@ -8708,6 +10997,11 @@ "BaseFormula": " l1d_pend_miss.pending / l1d_pend_miss.pending_cycles", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryBW", "LocateWith": "" @@ -8733,6 +11027,11 @@ "BaseFormula": " 1000 * mem_load_retired.l1_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8758,6 +11057,11 @@ "BaseFormula": " 1000 * l2_rqsts.all_demand_data_rd / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8783,6 +11087,11 @@ "BaseFormula": " 1000 * mem_load_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;Backend;CacheHits", "LocateWith": "" @@ -8808,6 +11117,11 @@ "BaseFormula": " 1000 * l2_rqsts.miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem;Offcore", "LocateWith": "" @@ -8833,6 +11147,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8858,6 +11177,11 @@ "BaseFormula": " 1000 * l2_rqsts.rfo_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheMisses;Offcore", "LocateWith": "" @@ -8887,6 +11211,11 @@ "BaseFormula": " 1000 * ( l2_rqsts.references - l2_rqsts.miss ) / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8912,6 +11241,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8937,6 +11271,11 @@ "BaseFormula": " 1000 * mem_load_retired.l3_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -8962,6 +11301,11 @@ "BaseFormula": " 1000 * mem_load_retired.fb_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8988,6 +11332,11 @@ "BaseFormula": " 64 * l1d.replacement / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9014,6 +11363,11 @@ "BaseFormula": " 64 * l2_lines_in.all / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9040,6 +11394,11 @@ "BaseFormula": " 64 * longest_lat_cache.miss / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9066,6 +11425,11 @@ "BaseFormula": " 64 * offcore_requests.all_requests / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW;Offcore", "LocateWith": "" @@ -9113,7 +11477,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Page_Walks_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -9140,6 +11512,11 @@ "BaseFormula": " 1000 * itlb_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;MemoryTLB", "LocateWith": "" @@ -9165,6 +11542,11 @@ "BaseFormula": " 1000 * dtlb_load_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -9190,6 +11572,11 @@ "BaseFormula": " 1000 * dtlb_store_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -9216,6 +11603,11 @@ "BaseFormula": " tma_info_memory_l1d_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9242,6 +11634,11 @@ "BaseFormula": " tma_info_memory_l2_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9268,6 +11665,11 @@ "BaseFormula": " tma_info_memory_l3_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9294,6 +11696,11 @@ "BaseFormula": " tma_info_memory_l3_cache_access_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW;Offcore", "LocateWith": "" @@ -9324,7 +11731,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Useless_HWPF" + } + ], + "Formula": "a > 0.15", + "BaseFormula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -9351,6 +11766,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests.demand_data_rd", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Memory_Lat;Offcore", "LocateWith": "" @@ -9376,6 +11796,11 @@ "BaseFormula": " offcore_requests_outstanding.l3_miss_demand_data_rd / offcore_requests.l3_miss_demand_data_rd", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_Lat;Offcore", "LocateWith": "" @@ -9401,6 +11826,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests_outstanding.demand_data_rd:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -9426,6 +11856,11 @@ "BaseFormula": " offcore_requests_outstanding.data_rd / offcore_requests_outstanding.cycles_with_data_rd", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -9451,6 +11886,11 @@ "BaseFormula": " 1000 * mem_load_misc_retired.uc / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -9476,6 +11916,11 @@ "BaseFormula": " 1000 * sq_misc.bus_lock / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -9506,6 +11951,11 @@ "BaseFormula": " tma_info_system_cpus_utilized / num_cpus", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC;Summary", "LocateWith": "" @@ -9532,6 +11982,11 @@ "BaseFormula": " cpu_clk_unhalted.ref_tsc / tsc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", "LocateWith": "" @@ -9566,6 +12021,11 @@ "BaseFormula": " tma_info_system_turbo_utilization * tsc / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "SYSTEM", "MetricGroup": "Summary;Power", "LocateWith": "" @@ -9604,6 +12064,11 @@ "BaseFormula": " ( ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.256b_packed_single ) / ( 1000000000 ) ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -9629,6 +12094,11 @@ "BaseFormula": " tma_info_thread_clks / cpu_clk_unhalted.ref_tsc", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -9663,6 +12133,11 @@ "BaseFormula": " 1 - cpu_clk_unhalted.one_thread_active / cpu_clk_unhalted.ref_distributed if smt_on else 0", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "SMT", "LocateWith": "" @@ -9689,7 +12164,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Kernel_Utilization > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Kernel_Utilization" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_System_Kernel_Utilization > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "OS", @@ -9716,6 +12199,11 @@ "BaseFormula": " cpu_clk_unhalted.thread_p:sup / inst_retired.any_p:sup", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "OS", "LocateWith": "" @@ -9742,7 +12230,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_C0_Wait > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_C0_Wait" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_System_C0_Wait > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", @@ -9774,6 +12270,11 @@ "BaseFormula": " 64 * ( unc_arb_trk_requests.all + unc_arb_coh_trk_requests.all ) / ( 1000000 ) / tma_info_system_time / 1000", "Category": "TMA", "CountDomain": "GB/sec", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "$issueBW" + }, "ResolutionLevels": "ARB", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC", "LocateWith": "" @@ -9800,6 +12301,11 @@ "BaseFormula": " unc_pkg_energy_status * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "PKG", "MetricGroup": "Power;SoC", "LocateWith": "" @@ -9822,7 +12328,15 @@ "Category": "TMA", "CountDomain": "Seconds", "Threshold": { - "Formula": "metric_TMA_Info_System_Time < 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Time" + } + ], + "Formula": "a < 1", + "BaseFormula": "metric_TMA_Info_System_Time < 1", + "ThresholdIssues": "" }, "ResolutionLevels": "SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -9850,7 +12364,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_MUX" + } + ], + "Formula": "a > 1.1 | a < 0.9", + "BaseFormula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -9873,6 +12395,11 @@ "BaseFormula": " unc_clock.socket", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CLOCK", "MetricGroup": "SoC", "LocateWith": "" @@ -9899,7 +12426,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_IpFarBranch < 1000000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_IpFarBranch" + } + ], + "Formula": "a < 1000000", + "BaseFormula": "metric_TMA_Info_System_IpFarBranch < 1000000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;OS", diff --git a/ADL/metrics/perf/alderlake_metrics_goldencove_core_perf.json b/ADL/metrics/perf/alderlake_metrics_goldencove_core_perf.json index 7469d87c..b2bdf6c4 100644 --- a/ADL/metrics/perf/alderlake_metrics_goldencove_core_perf.json +++ b/ADL/metrics/perf/alderlake_metrics_goldencove_core_perf.json @@ -1,1635 +1,1806 @@ [ { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", + "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", + "MetricThreshold": "tma_bottleneck_mispredictions > 20", + "PublicDescription": "Total pipeline cost of Branch Misprediction related bottlenecks. Related metrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", + "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_big_code", + "MetricThreshold": "tma_bottleneck_big_code > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", + "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) ) ) - tma_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_instruction_fetch_bw", + "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) )", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", + "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", + "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks. Related metrics: tma_l3_hit_latency, tma_mem_latency.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", + "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs)", "MetricExpr": "100 * ( tma_memory_bound * ( tma_l1_bound / max( tma_memory_bound , ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) ) * ( tma_dtlb_load / max( tma_l1_bound , ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", + "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", + "PublicDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs). Related metrics: tma_dtlb_load, tma_dtlb_store.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", + "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors)", "MetricExpr": "100 * ( tma_memory_bound * ( ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", - "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", + "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", + "PublicDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", + "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation", "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", + "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", "PublicDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", + "BriefDescription": "Total pipeline cost of irregular execution (e.g", "MetricExpr": "100 * ( ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + RS.EMPTY_RESOURCE / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", - "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments).", + "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", + "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments). Related metrics: tma_microcode_sequencer, tma_ms_switches.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end.", + "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end", "MetricExpr": "100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", "MetricGroup": "BvOB;Cor;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_other_bottlenecks", + "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", "PublicDescription": "Total pipeline cost of remaining bottlenecks in the back-end. Examples include data-dependencies (Core Bound when Low ILP) and other unlisted memory-related stalls.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", + "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA", "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_branching_overhead", + "MetricThreshold": "tma_bottleneck_branching_overhead > 5", "PublicDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound).", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", + "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead", "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_useful_work", + "MetricThreshold": "tma_bottleneck_useful_work > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "topdown\\-fe\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_frontend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_frontend_bound > 0.15", "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound. Sample with: FRONTEND_RETIRED.LATENCY_GE_4.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( topdown\\-fetch\\-lat / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots )", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", "MetricName": "tma_fetch_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_16, FRONTEND_RETIRED.LATENCY_GE_8.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache", "MetricExpr": "max( 0 , tma_icache_misses - tma_code_l2_miss )", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_CODE_RD / tma_info_thread_clks", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_itlb_misses - tma_code_stlb_miss )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + tma_unknown_branches", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "( tma_branch_mispredicts / tma_bad_speculation ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "( 1 - ( tma_branch_mispredicts / tma_bad_speculation ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 3 ) * cpu@UOPS_RETIRED.MS\\,cmask\\=0x1\\,edge\\=0x1@ / ( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) / tma_info_thread_clks", - "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated", + "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: FRONTEND_RETIRED.MS_FLOWS." + "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: FRONTEND_RETIRED.MS_FLOWS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "DECODE.LCP / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." + "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", - "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS." + "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "max( 0 , tma_frontend_bound - tma_fetch_latency )", - "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", + "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2.", + "MetricThreshold": "tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.MITE_CYCLES_ANY - IDQ.MITE_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", "ScaleUnit": "100%", + "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder.", + "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder", "MetricExpr": "( cpu@INST_DECODED.DECODERS\\,cmask\\=0x1@ - cpu@INST_DECODED.DECODERS\\,cmask\\=0x2@ ) / tma_info_core_core_clks / 2", - "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated", + "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated;tma_issueD0", "MetricName": "tma_decoder0_alone", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder. Related metrics: tma_few_uops_instructions." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.DSB_CYCLES_ANY - IDQ.DSB_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", "ScaleUnit": "100%", + "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit", "MetricExpr": "( LSD.CYCLES_ACTIVE - LSD.CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "FetchBW;LSD;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_lsd", "ScaleUnit": "100%", + "MetricThreshold": "tma_lsd > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit. LSD typically does well sustaining Uop supply. However; in some rare cases; optimal uop-delivery could not be reached for small loops whose size (in terms of number of uops) does not suit well the LSD structure." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details", "MetricExpr": "max( IDQ.MS_CYCLES_ANY , cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ / ( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) ) / tma_info_core_core_clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_ms", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "max( 1 - ( tma_frontend_bound + tma_backend_bound + tma_retiring ) , 0 )", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_bad_speculation", "ScaleUnit": "100%", + "MetricThreshold": "tma_bad_speculation > 0.15", "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( topdown\\-br\\-mispredict / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS.", + "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", + "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types)", "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Slots", "MetricName": "tma_other_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "max( 0 , tma_bad_speculation - tma_branch_mispredicts )", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots", + "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT.", + "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering", "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_machine_clears_group;Slots", "MetricName": "tma_other_nukes", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "( topdown\\-be\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_backend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_backend_bound > 0.2", "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( topdown\\-mem\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", "MetricName": "tma_memory_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two).", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( EXE_ACTIVITY.BOUND_ON_LOADS - MEMORY_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", + "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT." + "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "min( ( 7 ) * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - MEMORY_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS." + "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", "MetricName": "tma_load_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", "MetricName": "tma_load_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", + "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache", "MetricExpr": "min( 2 * ( MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETIRED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS ) * 20 / 100 , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - MEMORY_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_l1_latency_dependency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( 16 * max( 0 , MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS.ALL_RFO ) + ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) * ( ( 10 ) * L2_RQSTS.RFO_HIT + ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) ) / tma_info_thread_clks", - "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks", + "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS." + "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "L1D_PEND_MISS.FB_FULL / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", + "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory)." + "MetricThreshold": "tma_fb_full > 0.3", + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L1D_MISS - MEMORY_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", + "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited)", "MetricExpr": "( 3 * tma_info_system_core_frequency ) * MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", "MetricName": "tma_l2_hit_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L2_MISS - MEMORY_ACTIVITY.STALLS_L3_MISS ) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( ( 28 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) + ( ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS." + "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD + MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 1 - ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD." + "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( ( 12 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_RETIRED.L3_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT." + "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_bottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "( XQ.FULL_CYCLES + L1D_PEND_MISS.L2_STALLS ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clks )", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that)." + "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that)." + "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( MEM_STORE_RETIRED.L2_HIT * ( 10 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)." + "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( 28 * tma_info_system_core_frequency ) * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM." + "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization", + "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores.", + "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores", "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread_clks", - "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSmSt", "MetricName": "tma_streaming_stores", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE." + "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma_fb_full." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( ( 7 ) * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_STORE_MISSES.WALK_ACTIVE ) / tma_info_core_core_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES." + "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_load." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", "MetricName": "tma_store_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", "MetricName": "tma_store_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "max( 0 , tma_backend_bound - tma_memory_bound )", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", "MetricName": "tma_core_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations).", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.DIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", "ScaleUnit": "100%", + "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIVIDER_ACTIVE." }, { - "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active", "MetricExpr": "ARITH.FPDIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_fp_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_int_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations", "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks + tma_c02_wait", - "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", + "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks;tma_issueSO", "MetricName": "tma_serializing_operation", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD." + "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions", "MetricExpr": "CPU_CLK_UNHALTED.PAUSE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_slow_pause", "ScaleUnit": "100%", + "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.PAUSE_INST." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings).", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings)", "MetricExpr": "CPU_CLK_UNHALTED.C01 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_c01_wait", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_c01_wait > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings).", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings)", "MetricExpr": "CPU_CLK_UNHALTED.C02 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_c02_wait", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_c02_wait > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions", "MetricExpr": "13 * MISC2_RETIRED.LFENCE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_memory_fence", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_fence > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( tma_ports_utilized_0 * tma_info_thread_clks + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIV_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - EXE_ACTIVITY.BOUND_ON_LOADS ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( EXE_ACTIVITY.EXE_BOUND_0_PORTS + max( RS.EMPTY_RESOURCE - RESOURCE_STALLS.SCOREBOARD , 0 ) ) / tma_info_thread_clks * ( CYCLE_ACTIVITY.STALLS_TOTAL - EXE_ACTIVITY.BOUND_ON_LOADS ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", + "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles)", "MetricExpr": "160 * ASSISTS.SSE_AVX_MIX / tma_info_thread_clks", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks;tma_issueMV", "MetricName": "tma_mixing_vectors", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic." + "MetricThreshold": "tma_mixing_vectors > 0.05", + "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Sample with: UOPS_EXECUTED.CYCLES_GE_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + UOPS_DISPATCHED.PORT_5_11 + UOPS_DISPATCHED.PORT_6 ) / ( 5 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED.PORT_0 / tma_info_core_core_clks", - "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0." + "MetricThreshold": "tma_port_0 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_1 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1." + "MetricThreshold": "tma_port_1 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_6 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1." + "MetricThreshold": "tma_port_6 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "UOPS_DISPATCHED.PORT_2_3_10 / ( 3 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_load_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations. Sample with: UOPS_DISPATCHED.PORT_2_3_10." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_4_9 + UOPS_DISPATCHED.PORT_7_8 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations. Sample with: UOPS_DISPATCHED.PORT_7_8." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( topdown\\-retiring / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_retiring", "ScaleUnit": "100%", + "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.1", "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "max( 0 , tma_retiring - tma_heavy_operations )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", "MetricName": "tma_light_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", "ScaleUnit": "100%", + "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired)", "MetricExpr": "tma_int_vector_128b + tma_int_vector_256b", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_int_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_int_operations > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired). Vector/Matrix Int operations and shuffles are counted. Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain." }, { - "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", + "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired", "MetricExpr": "( INT_VEC_RETIRED.ADD_128 + INT_VEC_RETIRED.VNNI_128 ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops", + "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops;tma_issue2P", "MetricName": "tma_int_vector_128b", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_vector_128b > 0.1 & tma_int_operations > 0.1 & tma_light_operations > 0.6", + "PublicDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", + "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired", "MetricExpr": "( INT_VEC_RETIRED.ADD_256 + INT_VEC_RETIRED.MUL_256 + INT_VEC_RETIRED.VNNI_256 ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops", + "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops;tma_issue2P", "MetricName": "tma_int_vector_256b", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_vector_256b > 0.1 & tma_int_operations > 0.1 & tma_light_operations > 0.6", + "PublicDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses", "MetricExpr": "tma_light_operations * MEM_UOP_RETIRED.ANY / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_memory_operations", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_operations > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.MACRO_FUSED / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_fused_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Other_Light_Ops in MTL!)}." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused", "MetricExpr": "tma_light_operations * ( BR_INST_RETIRED.ALL_BRANCHES - INST_RETIRED.MACRO_FUSED ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_non_fused_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused. Non-conditional branches like direct JMP or CALL would count here. Can be used to examine fusible conditional jumps that were not fused." }, { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", + "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes", "MetricExpr": "max( 0 , tma_light_operations - ( tma_fp_arith + tma_int_operations + tma_memory_operations + tma_fused_instructions + tma_non_fused_branches ) )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_other_light_ops", "ScaleUnit": "100%", + "MetricThreshold": "tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_light_ops_group;Slots", "MetricName": "tma_nop_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer)", "MetricExpr": "tma_light_operations * INT_VEC_RETIRED.SHUFFLES / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_other_light_ops_group;Slots", "MetricName": "tma_shuffles_256b", "ScaleUnit": "100%", + "MetricThreshold": "tma_shuffles_256b > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer). Shuffles may incur slow cross \"vector lane\" data transfers." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "( topdown\\-heavy\\-ops / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", "MetricName": "tma_heavy_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+]). Sample with: UOPS_RETIRED.HEAVY.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops", "MetricExpr": "max( 0 , tma_heavy_operations - tma_microcode_sequencer )", - "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueD0", "MetricName": "tma_few_uops_instructions", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions." + "MetricThreshold": "tma_few_uops_instructions > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions. Related metrics: tma_decoder0_alone." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "UOPS_RETIRED.MS / tma_info_thread_slots", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: UOPS_RETIRED.MS." + "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: UOPS_RETIRED.MS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( ( 99 *3 + 63 + 30 ) / 5 ) * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults", "MetricExpr": "99 * ASSISTS.PAGE_FAULT / tma_info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_page_faults", "ScaleUnit": "100%", + "MetricThreshold": "tma_page_faults > 0.05", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults. A Page Fault may apply on first application access to a memory page. Note operating system handling of page faults accounts for the majority of its cost." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists", "MetricExpr": "30 * ASSISTS.FP / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_fp_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_assists > 0.1", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists", "MetricExpr": "63 * ASSISTS.SSE_AVX_MIX / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_avx_assists", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_avx_assists > 0.1" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", "ScaleUnit": "100%", + "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources. Sample with: FRONTEND_RETIRED.MS_FLOWS." }, { - "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts.", + "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts", "MetricExpr": "100 * ( 1 - tma_core_bound / tma_ports_utilization if tma_core_bound < tma_ports_utilization else 1 ) if tma_info_system_smt_2t_utilization > 0.5 else 0", "MetricGroup": "Cor;SMT;Metric", - "MetricName": "tma_info_botlnk_l0_core_bound_likely" + "MetricName": "tma_info_botlnk_l0_core_bound_likely", + "MetricThreshold": "tma_info_botlnk_core_bound_likely > 0.5" }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", - "MetricName": "tma_info_thread_uoppi" + "MetricName": "tma_info_thread_uoppi", + "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", - "MetricName": "tma_info_thread_uptb" + "MetricName": "tma_info_thread_uptb", + "MetricThreshold": "tma_info_thread_uptb < 6 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor.", + "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor", "MetricExpr": "tma_info_thread_slots / ( TOPDOWN.SLOTS / 2 ) if #SMT_on else 1", "MetricGroup": "SMT;TmaL1;TopdownL1;tma_L1_group;Metric", "MetricName": "tma_info_thread_slots_utilization" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / tma_info_core_core_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( FP_ARITH_DISPATCHED.PORT_0 + FP_ARITH_DISPATCHED.PORT_1 + FP_ARITH_DISPATCHED.PORT_5 ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "uops Executed per Cycle.", + "BriefDescription": "uops Executed per Cycle", "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", "MetricGroup": "Power;Metric", "MetricName": "tma_info_core_epc" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipload" + "MetricName": "tma_info_inst_mix_ipload", + "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipstore" + "MetricName": "tma_info_inst_mix_ipstore", + "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipbranch" + "MetricName": "tma_info_inst_mix_ipbranch", + "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipcall" + "MetricName": "tma_info_inst_mix_ipcall", + "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", - "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_iptb" + "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", + "MetricName": "tma_info_inst_mix_iptb", + "MetricThreshold": "tma_info_inst_mix_iptb < 6 * 2 + 1", + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipflop" + "MetricName": "tma_info_inst_mix_ipflop", + "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", + "MetricThreshold": "tma_info_inst_mix_iparith < 10", "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", + "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", + "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / CPU_CLK_UNHALTED.PAUSE_INST", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ippause" }, { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / SW_PREFETCH_ACCESS.ANY", "MetricGroup": "Prefetches;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipswpf" + "MetricName": "tma_info_inst_mix_ipswpf", + "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions.", + "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions", "MetricExpr": "INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "MicroSeq;Pipeline;Ret;Metric", - "MetricName": "tma_info_pipeline_strings_cycles" + "MetricName": "tma_info_pipeline_strings_cycles", + "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1" }, { - "BriefDescription": "Instructions per a microcode Assist invocation.", + "BriefDescription": "Instructions per a microcode Assist invocation", "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", "MetricName": "tma_info_pipeline_ipassist", + "MetricThreshold": "tma_info_pipeline_ipassist < 100000", "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." }, { - "BriefDescription": "pub/v3.5.", + "BriefDescription": "", "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@ )", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Average number of uops fetched from LSD per cycle.", + "BriefDescription": "Average number of uops fetched from LSD per cycle", "MetricExpr": "LSD.UOPS / LSD.CYCLES_ACTIVE", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_lsd" }, { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", + "BriefDescription": "Average number of uops fetched from DSB per cycle", "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_dsb" }, { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", + "BriefDescription": "Average number of uops fetched from MITE per cycle", "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_mite" }, { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", + "BriefDescription": "Average number of Uops issued by front-end when it issued something", "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_frontend_fetch_upc" }, { - "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache).", + "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache)", "MetricExpr": "LSD.UOPS / ( UOPS_ISSUED.ANY )", "MetricGroup": "Fed;LSD;Metric", "MetricName": "tma_info_frontend_lsd_coverage" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( UOPS_ISSUED.ANY )", - "MetricGroup": "DSB;Fed;FetchBW;Metric", - "MetricName": "tma_info_frontend_dsb_coverage" + "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", + "MetricName": "tma_info_frontend_dsb_coverage", + "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 6 > 0.35", + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection.", + "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection", "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / cpu@INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_unknown_branch_cost", "PublicDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection. See Unknown_Branches node." }, { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", + "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "DSBmiss;Metric", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Average Latency for L1 instruction cache misses.", + "BriefDescription": "Average Latency for L1 instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / cpu@ICACHE_DATA.STALLS\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "Fed;FetchLat;IcMiss;Metric", "MetricName": "tma_info_frontend_icache_miss_latency" }, { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", "MetricGroup": "DSBmiss;Fed;Inst_Metric", - "MetricName": "tma_info_frontend_ipdsb_miss_ret" + "MetricName": "tma_info_frontend_ipdsb_miss_ret", + "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache true code cacheline misses per kilo instruction", "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code" }, { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction", "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code_all" }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) )", - "MetricGroup": "DSBmiss;Fed;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_misses" + "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_misses", + "MetricThreshold": "tma_info_botlnk_dsb_misses > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) )", - "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_bandwidth" + "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", + "MetricThreshold": "tma_info_botlnk_dsb_bandwidth > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", + "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_ic_misses" + "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", + "MetricName": "tma_info_botlnk_l2_ic_misses", + "MetricThreshold": "tma_info_botlnk_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmispredict" + "MetricName": "tma_info_bad_spec_ipmispredict", + "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken" + "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", + "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_cond_taken" + "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", + "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_ret" + "MetricName": "tma_info_bad_spec_ipmisp_ret", + "MetricThreshold": "tma_info_bad_spec_ipmisp_ret < 500" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_indirect" + "MetricName": "tma_info_bad_spec_ipmisp_indirect", + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 6 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", - "MetricGroup": "Bad;BrMispredicts;Core_Metric", - "MetricName": "tma_info_bad_spec_branch_misprediction_cost" + "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", + "MetricName": "tma_info_bad_spec_branch_misprediction_cost", + "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_mispredicts_resteers." }, { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", + "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", "MetricGroup": "BrMispredicts;Metric", "MetricName": "tma_info_bad_spec_spec_clears_ratio" }, { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", + "BriefDescription": "Fraction of branches that are non-taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_nt" }, { - "BriefDescription": "Fraction of branches that are taken conditionals.", + "BriefDescription": "Fraction of branches that are taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_tk" }, { - "BriefDescription": "Fraction of branches that are CALL or RET.", + "BriefDescription": "Fraction of branches that are CALL or RET", "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_callret" }, { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", + "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_jump" }, { - "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group).", + "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group)", "MetricExpr": "1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_other_branches" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / MEM_LOAD_COMPLETED.L1_MISS_ANY", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki_load" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * L2_RQSTS.RFO_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_all" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", + "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_fb_hpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Offcore;Metric", "MetricName": "tma_info_memory_l3_cache_access_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", - "MetricName": "tma_info_memory_tlb_page_walks_utilization" + "MetricName": "tma_info_memory_tlb_page_walks_utilization", + "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Fed;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_code_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_load_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_access_bw", "MetricGroup": "Mem;MemoryBW;Offcore;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_access_bw_2t" }, { - "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses.", + "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses", "MetricExpr": "L2_LINES_OUT.USELESS_HWPF / ( L2_LINES_OUT.SILENT + L2_LINES_OUT.NON_SILENT )", "MetricGroup": "Prefetches;Metric", - "MetricName": "tma_info_memory_prefetches_useless_hwpf" + "MetricName": "tma_info_memory_prefetches_useless_hwpf", + "MetricThreshold": "tma_info_memory_useless_hwpf > 0.15" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Latency for L3 cache miss demand Loads.", + "BriefDescription": "Average Latency for L3 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.L3_MISS_DEMAND_DATA_RD / OFFCORE_REQUESTS.L3_MISS_DEMAND_DATA_RD", "MetricGroup": "Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l3_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=0x1@", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", + "BriefDescription": "Un-cacheable retired load per kilo instruction", "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_uc_load_pki" }, { - "BriefDescription": "\"Bus lock\" per kilo instruction.", + "BriefDescription": "\"Bus lock\" per kilo instruction", "MetricExpr": "1000 * SQ_MISC.BUS_LOCK / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_bus_lock_pki" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", - "MetricName": "tma_info_system_kernel_utilization" + "MetricName": "tma_info_system_kernel_utilization", + "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states.", + "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states", "MetricExpr": "CPU_CLK_UNHALTED.C0_WAIT / tma_info_thread_clks", "MetricGroup": "C0Wait;Metric", - "MetricName": "tma_info_system_c0_wait" + "MetricName": "tma_info_system_c0_wait", + "MetricThreshold": "tma_info_system_c0_wait > 0.05" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "64 * ( UNC_ARB_TRK_REQUESTS.ALL + UNC_ARB_COH_TRK_REQUESTS.ALL ) / ( 1000000 ) / tma_info_system_time / 1000", - "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec", - "MetricName": "tma_info_system_dram_bw_use" + "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", + "MetricName": "tma_info_system_dram_bw_use", + "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "power@energy\\-pkg@ * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", - "MetricName": "tma_info_system_mux" + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "UNC_CLOCK.SOCKET", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", - "MetricName": "tma_info_system_ipfarbranch" + "MetricName": "tma_info_system_ipfarbranch", + "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" } ] \ No newline at end of file diff --git a/ARL/metrics/arrowlake_metrics_lioncove_core.json b/ARL/metrics/arrowlake_metrics_lioncove_core.json index 1dddd8c8..4edca423 100644 --- a/ARL/metrics/arrowlake_metrics_lioncove_core.json +++ b/ARL/metrics/arrowlake_metrics_lioncove_core.json @@ -2,8 +2,8 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for Intel(R) Core(TM) processors based on Arrow Lake performance hybrid architecture0", - "DatePublished": "11/12/2024", - "Version": "0", + "DatePublished": "11/15/2024", + "Version": "1.0", "Legend": "", "TmaVersion": "5.01", "TmaFlavor": "Full" @@ -99,7 +99,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Mispredictions > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Mispredictions" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Mispredictions > 20", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP", @@ -171,7 +179,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Big_Code > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Big_Code" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Big_Code > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB", @@ -307,7 +323,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Instruction_Fetch_BW" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", @@ -504,7 +528,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore", @@ -588,7 +620,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Compute_Bound_Est" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20", + "ThresholdIssues": "$issueComp" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Cor", @@ -776,7 +816,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Irregular_Overhead > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Irregular_Overhead" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Irregular_Overhead > 10", + "ThresholdIssues": "$issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BvIO;Cor;Ret", @@ -812,7 +860,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Branching_Overhead > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Branching_Overhead" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Bottleneck_Branching_Overhead > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Ret", @@ -876,7 +932,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Useful_Work > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Useful_Work" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Useful_Work > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;Ret", @@ -912,7 +976,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;BvIO;TmaL1;PGO", @@ -953,7 +1025,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Frontend;TmaL2", @@ -982,7 +1066,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss", @@ -1023,7 +1123,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -1056,7 +1176,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -1085,7 +1225,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB", @@ -1126,7 +1282,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -1159,7 +1335,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -1196,7 +1392,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -1233,7 +1453,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -1266,7 +1510,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -1315,7 +1575,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Mispredicts_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP", @@ -1364,7 +1644,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Clears_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;MachineClears", @@ -1393,7 +1693,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Unknown_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat", @@ -1422,7 +1742,23 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC, $issueMS, $issueMV, $issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MicroSeq", @@ -1451,7 +1787,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LCP(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -1480,7 +1832,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchLat", @@ -1521,7 +1889,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;Frontend;TmaL2", @@ -1566,7 +1942,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -1611,7 +1999,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;FetchBW", @@ -1640,7 +2040,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....LSD(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LSD(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....LSD(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;LSD", @@ -1669,7 +2081,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 5 & b > 20", + "BaseFormula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -1705,7 +2129,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", @@ -1746,7 +2178,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2", @@ -1779,7 +2223,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Cond_NT_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Cond_NT_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Cond_NT_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -1812,7 +2272,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Cond_TK_Bwd_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Cond_TK_Bwd_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Cond_TK_Bwd_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -1845,7 +2321,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Cond_TK_Fwd_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Cond_TK_Fwd_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Cond_TK_Fwd_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -1878,7 +2370,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Ind_Call_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ind_Call_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Ind_Call_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -1919,7 +2427,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Ind_Jump_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ind_Jump_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Ind_Jump_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -1952,7 +2476,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Ret_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ret_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Ret_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -2005,7 +2545,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;BrMispredicts", @@ -2046,7 +2602,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueMC, $issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2", @@ -2095,7 +2663,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Nukes(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;Machine_Clears", @@ -2131,7 +2715,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;TmaL1", @@ -2172,7 +2764,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20", + "BaseFormula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2", @@ -2201,7 +2805,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1, $issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -2238,7 +2858,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -2275,7 +2915,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -2304,7 +2968,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -2345,7 +3033,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -2386,7 +3102,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -2427,7 +3171,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -2456,7 +3228,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Fwd_Blk(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -2485,7 +3277,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L1_Latency_Dependency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -2518,7 +3330,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Capacity(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L1_Latency_Capacity(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L1_Latency_Capacity(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -2551,7 +3383,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Lock_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Offcore", @@ -2592,7 +3444,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......Split_Loads(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Loads(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......Split_Loads(%) > 30", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -2621,7 +3481,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......FB_Full(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FB_Full(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......FB_Full(%) > 30", + "ThresholdIssues": "$issueBW, $issueSL, $issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW", @@ -2650,7 +3518,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem", @@ -2704,7 +3588,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L2_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryLat", @@ -2733,7 +3637,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -2795,7 +3715,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Contested_Accesses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -2857,7 +3797,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Data_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;Offcore;Snoop", @@ -2911,7 +3871,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L3_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -2944,7 +3924,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......SQ_Full(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 30 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -2973,7 +3973,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -3002,7 +4018,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Bandwidth(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -3035,7 +4071,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat;Offcore", @@ -3064,7 +4120,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -3105,7 +4177,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO, $issueSL, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore", @@ -3138,7 +4230,27 @@ "Category": "TMA", "CountDomain": "Core_Utilization", "Threshold": { - "Formula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3167,7 +4279,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Streaming_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBW;Offcore", @@ -3204,7 +4336,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -3241,7 +4393,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3270,7 +4446,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3311,7 +4511,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3352,7 +4580,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3393,7 +4649,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3434,7 +4718,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2;Compute", @@ -3463,7 +4759,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB", @@ -3492,7 +4804,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3525,7 +4857,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......INT_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3558,7 +4910,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;PortsUtil", @@ -3587,7 +4955,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Slow_Pause(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3616,7 +5004,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......C01_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......C01_Wait(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......C01_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", @@ -3645,7 +5053,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......C02_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......C02_Wait(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......C02_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", @@ -3674,7 +5102,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Memory_Fence(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Memory_Fence(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Memory_Fence(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3739,7 +5187,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -3768,7 +5232,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_0(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -3797,7 +5281,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_........Mixing_Vectors(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Mixing_Vectors(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Mixing_Vectors(%) > 5", + "ThresholdIssues": "$issueMV" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3826,7 +5318,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_1(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -3855,7 +5367,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_2(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -3884,7 +5416,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_3m(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 40 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;PortsUtil", @@ -3913,7 +5465,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........ALU_Op_Utilization(%) > 40" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........ALU_Op_Utilization(%)" + } + ], + "Formula": "a > 40", + "BaseFormula": "metric_TMA_........ALU_Op_Utilization(%) > 40", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3942,7 +5502,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Load_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Load_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3975,7 +5543,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Store_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Store_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4011,7 +5587,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Retiring(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 70 | b > 10", + "BaseFormula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;TmaL1", @@ -4052,7 +5640,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -4109,7 +5705,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 20 & b > 60", + "BaseFormula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -4154,7 +5762,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......X87_Use(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -4199,7 +5823,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Scalar(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -4244,7 +5884,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -4293,7 +5949,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -4338,7 +6014,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -4387,7 +6083,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -4432,7 +6140,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......Int_Vector_128b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Int_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 60", + "BaseFormula": "metric_TMA_......Int_Vector_128b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;IntVector;Pipeline", @@ -4477,7 +6201,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......Int_Vector_256b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Int_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 60", + "BaseFormula": "metric_TMA_......Int_Vector_256b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;IntVector;Pipeline", @@ -4526,7 +6266,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Memory_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -4575,7 +6327,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Fused_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -4628,7 +6392,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Non_Fused_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -4725,7 +6501,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 30 & b > 60", + "BaseFormula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -4774,7 +6562,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Nop_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Pipeline", @@ -4823,7 +6627,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Shuffles_256b(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Shuffles_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Shuffles_256b(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC;Pipeline", @@ -4864,7 +6684,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -4913,7 +6741,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Few_Uops_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4942,7 +6782,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueMC, $issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -4971,7 +6823,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Assists(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO", @@ -5000,7 +6868,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........Page_Faults(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Page_Faults(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Page_Faults(%) > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5029,7 +6905,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........FP_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........FP_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -5058,7 +6942,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........AVX_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........AVX_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........AVX_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -5091,7 +6983,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......CISC(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5118,6 +7026,11 @@ "BaseFormula": " inst_retired.any / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Summary", "LocateWith": "" @@ -5160,7 +7073,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UopPI > 1.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UopPI" + } + ], + "Formula": "a > 1.05", + "BaseFormula": "metric_TMA_Info_Thread_UopPI > 1.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret;Retire", @@ -5204,7 +7125,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UpTB < 8 * 1.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UpTB" + } + ], + "Formula": "a < 8 * 1.5", + "BaseFormula": "metric_TMA_Info_Thread_UpTB < 8 * 1.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW", @@ -5231,6 +7160,11 @@ "BaseFormula": " 1 / tma_info_thread_ipc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Mem", "LocateWith": "" @@ -5252,6 +7186,11 @@ "BaseFormula": " cpu_clk_unhalted.thread", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", "LocateWith": "" @@ -5273,6 +7212,11 @@ "BaseFormula": " topdown.slots:perf_metrics", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", "LocateWith": "" @@ -5298,6 +7242,11 @@ "BaseFormula": " uops_executed.thread / uops_issued.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline", "LocateWith": "" @@ -5335,6 +7284,11 @@ "BaseFormula": " ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.256b_packed_single ) / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Flops", "LocateWith": "" @@ -5372,6 +7326,11 @@ "BaseFormula": " ( fp_arith_dispatched.v0 + fp_arith_dispatched.v1 + fp_arith_dispatched.v2 + fp_arith_dispatched.v3 ) / ( 4 * tma_info_thread_clks )", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -5397,6 +7356,11 @@ "BaseFormula": " uops_executed.thread / uops_executed.thread:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "LocateWith": "" @@ -5422,6 +7386,11 @@ "BaseFormula": " uops_executed.thread / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -5448,7 +7417,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpLoad < 3" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpLoad" + } + ], + "Formula": "a < 3", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpLoad < 3", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -5476,7 +7453,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpStore < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpStore" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpStore < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -5504,7 +7489,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpBranch < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpBranch" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpBranch < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;InsType", @@ -5532,7 +7525,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpCall < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpCall" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpCall < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", @@ -5560,7 +7561,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpTB < 8 * 2 + 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpTB" + } + ], + "Formula": "a < 8 * 2 + 1", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpTB < 8 * 2 + 1", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO", @@ -5587,6 +7596,11 @@ "BaseFormula": " br_inst_retired.all_branches / br_inst_retired.near_taken", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", "LocateWith": "" @@ -5625,7 +7639,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpFLOP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -5657,7 +7679,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -5685,7 +7715,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -5713,7 +7751,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -5745,7 +7791,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX128" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -5777,7 +7831,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX256" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -5804,6 +7866,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / cpu_clk_unhalted.pause_inst", "Category": "TMA", "CountDomain": "Inst_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", "LocateWith": "" @@ -5830,7 +7897,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpSWPF" + } + ], + "Formula": "a < 100", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -5853,6 +7928,11 @@ "BaseFormula": " inst_retired.any", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;TmaL1", "LocateWith": "INST_RETIRED.PREC_DIST" @@ -5894,6 +7974,11 @@ "BaseFormula": " ( tma_retiring * tma_info_thread_slots ) / uops_retired.slots:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret", "LocateWith": "" @@ -5920,7 +8005,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Pipeline_Strings_Cycles > 0.1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Pipeline_Strings_Cycles" + } + ], + "Formula": "a > 0.1", + "BaseFormula": "metric_TMA_Info_Pipeline_Strings_Cycles > 0.1", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq;Pipeline;Ret", @@ -5948,7 +8041,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Pipeline_IpAssist < 100000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Pipeline_IpAssist" + } + ], + "Formula": "a < 100000", + "BaseFormula": "metric_TMA_Info_Pipeline_IpAssist < 100000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", @@ -5975,6 +8076,11 @@ "BaseFormula": " lsd.uops / lsd.cycles_active", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -6000,6 +8106,11 @@ "BaseFormula": " idq.dsb_uops / idq.dsb_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -6025,6 +8136,11 @@ "BaseFormula": " idq.mite_uops / idq.mite_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -6050,6 +8166,11 @@ "BaseFormula": " uops_issued.any / uops_issued.any:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -6075,6 +8196,11 @@ "BaseFormula": " lsd.uops / ( uops_issued.any )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;LSD", "LocateWith": "" @@ -6101,7 +8227,19 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 8 > 0.35" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_DSB_Coverage" + }, + { + "Alias": "b", + "Value": "IPC" + } + ], + "Formula": "a < 0.7 & b / 8 > 0.35", + "BaseFormula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 8 > 0.35", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -6128,6 +8266,11 @@ "BaseFormula": " int_misc.unknown_branch_cycles / int_misc.unknown_branch_cycles:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed", "LocateWith": "" @@ -6153,6 +8296,11 @@ "BaseFormula": " dsb2mite_switches.penalty_cycles / dsb2mite_switches.penalty_cycles:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss", "LocateWith": "" @@ -6178,6 +8326,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -6208,7 +8361,15 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_DSB_Switches_Ret > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_DSB_Switches_Ret" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_Frontend_DSB_Switches_Ret > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed;FetchLat", @@ -6240,7 +8401,15 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_MS_Latency_Ret > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_MS_Latency_Ret" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_Frontend_MS_Latency_Ret > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;MicroSeq", @@ -6271,6 +8440,11 @@ "BaseFormula": " ( frontend_retired.unknown_branch * frontend_retired.unknown_branch:retire_latency ) / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Clocks_Retired", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat", "LocateWith": "" @@ -6296,6 +8470,11 @@ "BaseFormula": " icache_data.stalls / icache_data.stall_periods", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", "LocateWith": "" @@ -6322,7 +8501,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret" + } + ], + "Formula": "a < 50", + "BaseFormula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -6349,6 +8536,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / baclears.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed", "LocateWith": "" @@ -6374,6 +8566,11 @@ "BaseFormula": " 1000 * frontend_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -6399,6 +8596,11 @@ "BaseFormula": " 1000 * l2_rqsts.code_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -6501,7 +8703,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Misses > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Misses" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Misses > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -6577,7 +8787,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Bandwidth" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -6649,7 +8867,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_IC_Misses > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_IC_Misses" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Info_Botlnk_IC_Misses > 5", + "ThresholdIssues": "$issueFL" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", @@ -6677,7 +8903,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMispredict" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts", @@ -6705,7 +8939,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -6732,6 +8974,11 @@ "BaseFormula": " inst_retired.any / br_misp_retired.cond_taken_bwd", "Category": "TMA", "CountDomain": "", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", "LocateWith": "" @@ -6757,6 +9004,11 @@ "BaseFormula": " inst_retired.any / br_misp_retired.cond_taken_fwd", "Category": "TMA", "CountDomain": "", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", "LocateWith": "" @@ -6783,7 +9035,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Ret" + } + ], + "Formula": "a < 500", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -6811,7 +9071,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect" + } + ], + "Formula": "a < 1000", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -6906,6 +9174,11 @@ "BaseFormula": " tma_mispredictions * tma_info_thread_slots / ( 8 ) / br_misp_retired.all_branches / 100", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "$issueBM" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", "LocateWith": "" @@ -6935,6 +9208,11 @@ "BaseFormula": " int_misc.clears_count / ( br_misp_retired.all_branches + machine_clears.count )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", "LocateWith": "" @@ -6960,6 +9238,11 @@ "BaseFormula": " br_inst_retired.cond_ntaken / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", "LocateWith": "" @@ -6986,7 +9269,15 @@ "Category": "TMA", "CountDomain": "Fraction", "Threshold": { - "Formula": "metric_TMA_Info_Branches_Cond_TK_Bwd > 0.3" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Branches_Cond_TK_Bwd" + } + ], + "Formula": "a > 0.3", + "BaseFormula": "metric_TMA_Info_Branches_Cond_TK_Bwd > 0.3", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", @@ -7014,7 +9305,15 @@ "Category": "TMA", "CountDomain": "Fraction", "Threshold": { - "Formula": "metric_TMA_Info_Branches_Cond_TK_Fwd > 0.2" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Branches_Cond_TK_Fwd" + } + ], + "Formula": "a > 0.2", + "BaseFormula": "metric_TMA_Info_Branches_Cond_TK_Fwd > 0.2", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", @@ -7045,6 +9344,11 @@ "BaseFormula": " ( br_inst_retired.near_call + br_inst_retired.near_return ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -7082,6 +9386,11 @@ "BaseFormula": " ( br_inst_retired.near_taken - br_inst_retired.cond_taken_bwd - br_inst_retired.cond_taken_fwd - 2 * br_inst_retired.near_call ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -7127,6 +9436,11 @@ "BaseFormula": " 1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk_bwd + tma_info_branches_cond_tk_fwd + tma_info_branches_callret + tma_info_branches_jump )", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -7152,6 +9466,11 @@ "BaseFormula": " l1d_pending.load / l1d_miss.load", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryLat", "LocateWith": "" @@ -7177,6 +9496,11 @@ "BaseFormula": " l1d_pending.load / l1d_pending.load_cycles", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryBW", "LocateWith": "" @@ -7202,6 +9526,11 @@ "BaseFormula": " 1000 * mem_load_retired.l1_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -7227,6 +9556,11 @@ "BaseFormula": " 1000 * l2_rqsts.all_demand_data_rd / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -7252,6 +9586,11 @@ "BaseFormula": " 1000 * mem_load_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;Backend;CacheHits", "LocateWith": "" @@ -7277,6 +9616,11 @@ "BaseFormula": " 1000 * l2_rqsts.miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem;Offcore", "LocateWith": "" @@ -7302,6 +9646,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -7327,6 +9676,11 @@ "BaseFormula": " 1000 * l2_rqsts.rfo_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheMisses;Offcore", "LocateWith": "" @@ -7356,6 +9710,11 @@ "BaseFormula": " 1000 * ( l2_rqsts.references - l2_rqsts.miss ) / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -7381,6 +9740,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -7406,6 +9770,11 @@ "BaseFormula": " 1000 * mem_load_retired.l3_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -7431,6 +9800,11 @@ "BaseFormula": " 1000 * mem_load_retired.fb_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -7457,6 +9831,11 @@ "BaseFormula": " 64 * l1d.l0_replacement / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -7483,6 +9862,11 @@ "BaseFormula": " 64 * l2_lines_in.all / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -7509,6 +9893,11 @@ "BaseFormula": " 64 * longest_lat_cache.miss / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -7535,6 +9924,11 @@ "BaseFormula": " 64 * offcore_requests.all_requests / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW;Offcore", "LocateWith": "" @@ -7569,7 +9963,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Page_Walks_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -7596,6 +9998,11 @@ "BaseFormula": " 1000 * itlb_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;MemoryTLB", "LocateWith": "" @@ -7621,6 +10028,11 @@ "BaseFormula": " 1000 * dtlb_load_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -7646,6 +10058,11 @@ "BaseFormula": " 1000 * dtlb_store_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -7676,7 +10093,15 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Load_STLB_Miss_Ret > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Load_STLB_Miss_Ret" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_Memory_Load_STLB_Miss_Ret > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -7708,7 +10133,15 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Store_STLB_Miss_Ret > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Store_STLB_Miss_Ret" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_Memory_Store_STLB_Miss_Ret > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -7740,7 +10173,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Useless_HWPF" + } + ], + "Formula": "a > 0.15", + "BaseFormula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -7767,6 +10208,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests.demand_data_rd", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Memory_Lat;Offcore", "LocateWith": "" @@ -7792,6 +10238,11 @@ "BaseFormula": " offcore_requests_outstanding.l3_miss_demand_data_rd / offcore_requests.l3_miss_demand_data_rd", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_Lat;Offcore", "LocateWith": "" @@ -7817,6 +10268,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests_outstanding.demand_data_rd:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -7842,6 +10298,11 @@ "BaseFormula": " offcore_requests_outstanding.data_rd / offcore_requests_outstanding.cycles_with_data_rd", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -7867,6 +10328,11 @@ "BaseFormula": " 1000 * mem_load_misc_retired.uc / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -7892,6 +10358,11 @@ "BaseFormula": " 1000 * sq_misc.bus_lock / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -7922,6 +10393,11 @@ "BaseFormula": " tma_info_system_cpus_utilized / num_cpus", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC;Summary", "LocateWith": "" @@ -7948,6 +10424,11 @@ "BaseFormula": " cpu_clk_unhalted.ref_tsc / tsc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", "LocateWith": "" @@ -7982,6 +10463,11 @@ "BaseFormula": " tma_info_system_turbo_utilization * tsc / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;Power", "LocateWith": "" @@ -8020,6 +10506,11 @@ "BaseFormula": " ( ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.256b_packed_single ) / ( 1000000000 ) ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -8045,6 +10536,11 @@ "BaseFormula": " tma_info_thread_clks / cpu_clk_unhalted.ref_tsc", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -8071,7 +10567,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Kernel_Utilization > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Kernel_Utilization" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_System_Kernel_Utilization > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "OS", @@ -8098,6 +10602,11 @@ "BaseFormula": " cpu_clk_unhalted.thread_p:sup / inst_retired.any_p:sup", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "OS", "LocateWith": "" @@ -8124,7 +10633,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_C0_Wait > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_C0_Wait" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_System_C0_Wait > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", @@ -8152,6 +10669,11 @@ "BaseFormula": " unc_pkg_energy_status * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "PKG", "MetricGroup": "Power;SoC", "LocateWith": "" @@ -8174,7 +10696,15 @@ "Category": "TMA", "CountDomain": "Seconds", "Threshold": { - "Formula": "metric_TMA_Info_System_Time < 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Time" + } + ], + "Formula": "a < 1", + "BaseFormula": "metric_TMA_Info_System_Time < 1", + "ThresholdIssues": "" }, "ResolutionLevels": "SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -8202,7 +10732,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_MUX" + } + ], + "Formula": "a > 1.1 | a < 0.9", + "BaseFormula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -8225,6 +10763,11 @@ "BaseFormula": " unc_clock.socket", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CLOCK", "MetricGroup": "SoC", "LocateWith": "" @@ -8251,7 +10794,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_IpFarBranch < 1000000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_IpFarBranch" + } + ], + "Formula": "a < 1000000", + "BaseFormula": "metric_TMA_Info_System_IpFarBranch < 1000000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;OS", diff --git a/ARL/metrics/perf/arrowlake_metrics_lioncove_core_perf.json b/ARL/metrics/perf/arrowlake_metrics_lioncove_core_perf.json index 32d8598a..b533b805 100644 --- a/ARL/metrics/perf/arrowlake_metrics_lioncove_core_perf.json +++ b/ARL/metrics/perf/arrowlake_metrics_lioncove_core_perf.json @@ -1,1588 +1,1756 @@ [ { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", + "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", + "MetricThreshold": "tma_bottleneck_mispredictions > 20", + "PublicDescription": "Total pipeline cost of Branch Misprediction related bottlenecks. Related metrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", + "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_big_code", + "MetricThreshold": "tma_bottleneck_big_code > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", + "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) ) ) - tma_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_instruction_fetch_bw", + "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_l1_latency_capacity + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) )", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", + "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_mem_bandwidth, tma_sq_full.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", + "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation", "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", + "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", "PublicDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", + "BriefDescription": "Total pipeline cost of irregular execution (e.g", "MetricExpr": "100 * ( ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + RS.EMPTY_RESOURCE / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_microcode_sequencer + max( 0 , tma_heavy_operations - tma_microcode_sequencer ) ) ) * ( ( ( 99 *3 + 63 + 30 ) / 5 ) * ASSISTS.ANY / tma_info_thread_slots / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", - "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments).", + "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", + "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments). Related metrics: tma_microcode_sequencer, tma_ms_switches.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", + "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA", "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_branching_overhead", + "MetricThreshold": "tma_bottleneck_branching_overhead > 5", "PublicDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound).", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", + "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead", "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_microcode_sequencer + max( 0 , tma_heavy_operations - tma_microcode_sequencer ) ) ) * ( ( ( 99 *3 + 63 + 30 ) / 5 ) * ASSISTS.ANY / tma_info_thread_slots / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_useful_work", + "MetricThreshold": "tma_bottleneck_useful_work > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "( topdown\\-fe\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_frontend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_frontend_bound > 0.15", "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( topdown\\-fetch\\-lat / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", "MetricName": "tma_fetch_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_16, FRONTEND_RETIRED.LATENCY_GE_8.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache", "MetricExpr": "max( 0 , ( FRONTEND_RETIRED.L1I_MISS * cpu_core@FRONTEND_RETIRED.L1I_MISS@R ) / tma_info_thread_clks - tma_code_l2_miss )", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache", "MetricExpr": "( FRONTEND_RETIRED.L2_MISS * cpu_core@FRONTEND_RETIRED.L2_MISS@R ) / tma_info_thread_clks", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , ( FRONTEND_RETIRED.ITLB_MISS * cpu_core@FRONTEND_RETIRED.ITLB_MISS@R ) / tma_info_thread_clks - tma_code_stlb_miss )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", "MetricExpr": "( FRONTEND_RETIRED.STLB_MISS * cpu_core@FRONTEND_RETIRED.STLB_MISS@R ) / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + tma_unknown_branches", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "( tma_branch_mispredicts / tma_bad_speculation ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "( 1 - ( tma_branch_mispredicts / tma_bad_speculation ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 3 ) * IDQ.MS_SWITCHES / tma_info_thread_clks", - "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated", + "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES." + "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "DECODE.LCP / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." + "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", - "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS." + "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "max( 0 , tma_frontend_bound - tma_fetch_latency )", - "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", + "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend.", + "MetricThreshold": "tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Related metrics: tma_dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( cpu@IDQ.MITE_UOPS\\,cmask\\=0x8\\,inv\\=0x1@ / tma_info_thread_clks + IDQ.MITE_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS ) * ( IDQ_BUBBLES.CYCLES_0_UOPS_DELIV.CORE - IDQ_BUBBLES.FETCH_LATENCY ) ) / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", "ScaleUnit": "100%", + "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( cpu@IDQ.DSB_UOPS\\,cmask\\=0x8\\,inv\\=0x1@ + IDQ.DSB_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS ) * ( IDQ_BUBBLES.CYCLES_0_UOPS_DELIV.CORE - IDQ_BUBBLES.FETCH_LATENCY ) ) / tma_info_thread_clks", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", "ScaleUnit": "100%", + "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit", "MetricExpr": "cpu@LSD.UOPS\\,cmask\\=0x8\\,inv\\=0x1@ / tma_info_thread_clks", "MetricGroup": "FetchBW;LSD;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_lsd", "ScaleUnit": "100%", + "MetricThreshold": "tma_lsd > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit. LSD typically does well sustaining Uop supply. However; in some rare cases; optimal uop-delivery could not be reached for small loops whose size (in terms of number of uops) does not suit well the LSD structure." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details", "MetricExpr": "IDQ.MS_CYCLES_ANY / tma_info_thread_clks", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_ms", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "( topdown\\-bad\\-spec / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_bad_speculation", "ScaleUnit": "100%", + "MetricThreshold": "tma_bad_speculation > 0.15", "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( topdown\\-br\\-mispredict / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS.", + "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by non-taken conditional branches.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by non-taken conditional branches", "MetricExpr": "( BR_MISP_RETIRED.COND_NTAKEN_COST * cpu_core@BR_MISP_RETIRED.COND_NTAKEN_COST@R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_cond_nt_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_cond_nt_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to misprediction by backward-taken conditional branches.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to misprediction by backward-taken conditional branches", "MetricExpr": "( BR_MISP_RETIRED.COND_TAKEN_BWD_COST * cpu_core@BR_MISP_RETIRED.COND_TAKEN_BWD_COST@R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_cond_tk_bwd_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_cond_tk_bwd_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to misprediction by forward-taken conditional branches.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to misprediction by forward-taken conditional branches", "MetricExpr": "( BR_MISP_RETIRED.COND_TAKEN_FWD_COST * cpu_core@BR_MISP_RETIRED.COND_TAKEN_FWD_COST@R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_cond_tk_fwd_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_cond_tk_fwd_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect CALL instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect CALL instructions", "MetricExpr": "( BR_MISP_RETIRED.INDIRECT_CALL_COST * cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_ind_call_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ind_call_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect JMP instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect JMP instructions", "MetricExpr": "max( ( ( BR_MISP_RETIRED.INDIRECT_COST * cpu_core@BR_MISP_RETIRED.INDIRECT_COST@R ) - ( BR_MISP_RETIRED.INDIRECT_CALL_COST * cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@R ) ) / tma_info_thread_clks , 0 )", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_ind_jump_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ind_jump_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by (indirect) RET instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by (indirect) RET instructions", "MetricExpr": "( BR_MISP_RETIRED.RET_COST * cpu_core@BR_MISP_RETIRED.RET_COST@R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_ret_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ret_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", + "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types)", "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Slots", "MetricName": "tma_other_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "max( 0 , tma_bad_speculation - tma_branch_mispredicts )", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots", + "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT.", + "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering", "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_machine_clears_group;Slots", "MetricName": "tma_other_nukes", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "( topdown\\-be\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_backend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_backend_bound > 0.2", "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( topdown\\-mem\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", "MetricName": "tma_memory_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two).", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "MEMORY_STALLS.L1 / tma_info_thread_clks", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", + "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT." + "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "( min( ( MEM_INST_RETIRED.STLB_HIT_LOADS * cpu_core@MEM_INST_RETIRED.STLB_HIT_LOADS@R ) , MEM_INST_RETIRED.STLB_HIT_LOADS * ( 7 ) ) if ( cpu_core@MEM_INST_RETIRED.STLB_HIT_LOADS@R >= 0 ) else ( MEM_INST_RETIRED.STLB_HIT_LOADS * ( 7 ) ) ) / tma_info_thread_clks + tma_load_stlb_miss", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS." + "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_dtlb_load - tma_load_stlb_miss )", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", "MetricName": "tma_load_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", "MetricName": "tma_load_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", + "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache", "MetricExpr": "4 * DEPENDENT_LOADS.ANY / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_l1_latency_dependency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: DEPENDENT_LOADS.ANY." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit Level 1 after missing Level 0 within the L1D cache.", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit Level 1 after missing Level 0 within the L1D cache", "MetricExpr": "( min( ( MEM_LOAD_RETIRED.L1_HIT_L1 * cpu_core@MEM_LOAD_RETIRED.L1_HIT_L1@R ) , MEM_LOAD_RETIRED.L1_HIT_L1 * 9 ) if ( cpu_core@MEM_LOAD_RETIRED.L1_HIT_L1@R >= 0 ) else ( MEM_LOAD_RETIRED.L1_HIT_L1 * 9 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Retired", "MetricName": "tma_l1_latency_capacity", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_latency_capacity > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( MEM_INST_RETIRED.LOCK_LOADS * cpu_core@MEM_INST_RETIRED.LOCK_LOADS@R ) / tma_info_thread_clks", - "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks", + "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS." + "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "( min( ( MEM_INST_RETIRED.SPLIT_LOADS * cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@R ) , MEM_INST_RETIRED.SPLIT_LOADS * tma_info_memory_load_miss_real_latency ) if ( cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@R >= 0 ) else ( MEM_INST_RETIRED.SPLIT_LOADS * tma_info_memory_load_miss_real_latency ) ) / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "L1D_MISS.FB_FULL / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", + "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory)." + "MetricThreshold": "tma_fb_full > 0.3", + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "MEMORY_STALLS.L2 / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", + "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited)", "MetricExpr": "( min( ( MEM_LOAD_RETIRED.L2_HIT * cpu_core@MEM_LOAD_RETIRED.L2_HIT@R ) , MEM_LOAD_RETIRED.L2_HIT * ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_RETIRED.L2_HIT@R >= 0 ) else ( MEM_LOAD_RETIRED.L2_HIT * ( 3 * tma_info_system_core_frequency ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", "MetricName": "tma_l2_hit_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "MEMORY_STALLS.L3 / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) + ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM@R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * ( 28 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM@R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * ( 28 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS." + "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_data_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) + ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 28 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 28 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD." + "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_contested_accesses, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( min( ( MEM_LOAD_RETIRED.L3_HIT * cpu_core@MEM_LOAD_RETIRED.L3_HIT@R ) , MEM_LOAD_RETIRED.L3_HIT * ( 12 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_RETIRED.L3_HIT@R >= 0 ) else ( MEM_LOAD_RETIRED.L3_HIT * ( 12 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT." + "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "( XQ.FULL + L1D_MISS.L2_STALLS ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( MEMORY_STALLS.MEM / tma_info_thread_clks )", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that)." + "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that)." + "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( MEM_STORE_RETIRED.L2_HIT * ( 10 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)." + "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "( min( ( MEM_INST_RETIRED.SPLIT_STORES * cpu_core@MEM_INST_RETIRED.SPLIT_STORES@R ) , MEM_INST_RETIRED.SPLIT_STORES * 1 ) if ( cpu_core@MEM_INST_RETIRED.SPLIT_STORES@R >= 0 ) else ( MEM_INST_RETIRED.SPLIT_STORES * 1 ) ) / tma_info_thread_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization", + "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores.", + "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores", "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread_clks", - "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSmSt", "MetricName": "tma_streaming_stores", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE." + "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma_fb_full." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( min( ( MEM_INST_RETIRED.STLB_HIT_STORES * cpu_core@MEM_INST_RETIRED.STLB_HIT_STORES@R ) , MEM_INST_RETIRED.STLB_HIT_STORES * ( 7 ) ) if ( cpu_core@MEM_INST_RETIRED.STLB_HIT_STORES@R >= 0 ) else ( MEM_INST_RETIRED.STLB_HIT_STORES * ( 7 ) ) ) / tma_info_thread_clks + tma_store_stlb_miss", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES." + "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_dtlb_load." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_dtlb_store - tma_store_stlb_miss )", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", "MetricName": "tma_store_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", "MetricName": "tma_store_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "max( 0 , tma_backend_bound - tma_memory_bound )", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", "MetricName": "tma_core_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations).", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.DIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", "ScaleUnit": "100%", + "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIV_ACTIVE." }, { - "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active", "MetricExpr": "ARITH.FPDIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_fp_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_int_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations", "MetricExpr": "( BE_STALLS.SCOREBOARD + CPU_CLK_UNHALTED.C02 ) / tma_info_thread_clks", - "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", + "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks;tma_issueSO", "MetricName": "tma_serializing_operation", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: BE_STALLS.SCOREBOARD." + "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: BE_STALLS.SCOREBOARD. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions", "MetricExpr": "CPU_CLK_UNHALTED.PAUSE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_slow_pause", "ScaleUnit": "100%", + "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.PAUSE_INST." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings).", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings)", "MetricExpr": "CPU_CLK_UNHALTED.C01 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_c01_wait", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_c01_wait > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings).", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings)", "MetricExpr": "CPU_CLK_UNHALTED.C02 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_c02_wait", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_c02_wait > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions", "MetricExpr": "13 * MISC2_RETIRED.LFENCE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_memory_fence", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_fence > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( EXE_ACTIVITY.EXE_BOUND_0_PORTS + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIV_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - EXE_ACTIVITY.BOUND_ON_LOADS ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.EXE_BOUND_0_PORTS / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", + "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles)", "MetricExpr": "160 * ASSISTS.SSE_AVX_MIX / tma_info_thread_clks", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks;tma_issueMV", "MetricName": "tma_mixing_vectors", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic." + "MetricThreshold": "tma_mixing_vectors > 0.05", + "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Sample with: UOPS_EXECUTED.CYCLES_GE_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "UOPS_DISPATCHED.ALU / ( 6 * tma_info_thread_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "UOPS_DISPATCHED.LOAD / ( 3 * tma_info_thread_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_load_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations. Sample with: UOPS_DISPATCHED.LOAD." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "( UOPS_DISPATCHED.STD + UOPS_DISPATCHED.STA ) / ( 7 * tma_info_thread_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations. Sample with: UOPS_DISPATCHED.STD, UOPS_DISPATCHED.STA." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( topdown\\-retiring / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_retiring", "ScaleUnit": "100%", + "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.1", "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "max( 0 , tma_retiring - tma_heavy_operations )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", "MetricName": "tma_light_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", "ScaleUnit": "100%", + "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "cpu@FP_ARITH_INST_RETIRED.VECTOR\\,umask\\=0x30@ / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_int_vector_128b, tma_int_vector_256b, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired)", "MetricExpr": "tma_int_vector_128b + tma_int_vector_256b", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_int_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_int_operations > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired). Vector/Matrix Int operations and shuffles are counted. Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain." }, { - "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", + "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired", "MetricExpr": "INT_VEC_RETIRED.128BIT / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops", + "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops;tma_issue2P", "MetricName": "tma_int_vector_128b", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_vector_128b > 0.1 & tma_int_operations > 0.1 & tma_light_operations > 0.6", + "PublicDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_256b, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", + "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired", "MetricExpr": "INT_VEC_RETIRED.256BIT / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops", + "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops;tma_issue2P", "MetricName": "tma_int_vector_256b", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_vector_256b > 0.1 & tma_int_operations > 0.1 & tma_light_operations > 0.6", + "PublicDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses", "MetricExpr": "tma_light_operations * MEM_UOP_RETIRED.ANY / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_memory_operations", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_operations > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.MACRO_FUSED / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_fused_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Other_Light_Ops in MTL!)}." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused", "MetricExpr": "tma_light_operations * ( BR_INST_RETIRED.ALL_BRANCHES - INST_RETIRED.BR_FUSED ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_non_fused_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused. Non-conditional branches like direct JMP or CALL would count here. Can be used to examine fusible conditional jumps that were not fused." }, { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", + "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes", "MetricExpr": "max( 0 , tma_light_operations - ( tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD + ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) ) / ( tma_retiring * tma_info_thread_slots ) + ( INT_VEC_RETIRED.ADD_128 + INT_VEC_RETIRED.VNNI_128 + INT_VEC_RETIRED.ADD_256 + INT_VEC_RETIRED.MUL_256 + INT_VEC_RETIRED.VNNI_256 ) / ( tma_retiring * tma_info_thread_slots ) + tma_memory_operations + tma_fused_instructions + tma_non_fused_branches ) )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_other_light_ops", "ScaleUnit": "100%", + "MetricThreshold": "tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_light_ops_group;Slots", "MetricName": "tma_nop_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer)", "MetricExpr": "tma_light_operations * INT_VEC_RETIRED.SHUFFLES / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_other_light_ops_group;Slots", "MetricName": "tma_shuffles_256b", "ScaleUnit": "100%", + "MetricThreshold": "tma_shuffles_256b > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer). Shuffles may incur slow cross \"vector lane\" data transfers." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "( topdown\\-heavy\\-ops / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", "MetricName": "tma_heavy_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+]). Sample with: UOPS_RETIRED.HEAVY.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops", "MetricExpr": "max( 0 , tma_heavy_operations - tma_microcode_sequencer )", - "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueD0", "MetricName": "tma_few_uops_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_few_uops_instructions > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "UOPS_RETIRED.MS / tma_info_thread_slots", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: UOPS_RETIRED.MS." + "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: UOPS_RETIRED.MS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( ( 99 *3 + 63 + 30 ) / 5 ) * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults", "MetricExpr": "99 * ASSISTS.PAGE_FAULT / tma_info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_page_faults", "ScaleUnit": "100%", + "MetricThreshold": "tma_page_faults > 0.05", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults. A Page Fault may apply on first application access to a memory page. Note operating system handling of page faults accounts for the majority of its cost." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists", "MetricExpr": "30 * ASSISTS.FP / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_fp_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_assists > 0.1", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists", "MetricExpr": "63 * ASSISTS.SSE_AVX_MIX / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_avx_assists", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_avx_assists > 0.1" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", "ScaleUnit": "100%", + "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources." }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", - "MetricName": "tma_info_thread_uoppi" + "MetricName": "tma_info_thread_uoppi", + "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", - "MetricName": "tma_info_thread_uptb" + "MetricName": "tma_info_thread_uptb", + "MetricThreshold": "tma_info_thread_uptb < 8 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / tma_info_thread_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( FP_ARITH_DISPATCHED.V0 + FP_ARITH_DISPATCHED.V1 + FP_ARITH_DISPATCHED.V2 + FP_ARITH_DISPATCHED.V3 ) / ( 4 * tma_info_thread_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "uops Executed per Cycle.", + "BriefDescription": "uops Executed per Cycle", "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", "MetricGroup": "Power;Metric", "MetricName": "tma_info_core_epc" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipload" + "MetricName": "tma_info_inst_mix_ipload", + "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipstore" + "MetricName": "tma_info_inst_mix_ipstore", + "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipbranch" + "MetricName": "tma_info_inst_mix_ipbranch", + "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipcall" + "MetricName": "tma_info_inst_mix_ipcall", + "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", - "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_iptb" + "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", + "MetricName": "tma_info_inst_mix_iptb", + "MetricThreshold": "tma_info_inst_mix_iptb < 8 * 2 + 1", + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipflop" + "MetricName": "tma_info_inst_mix_ipflop", + "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", + "MetricThreshold": "tma_info_inst_mix_iparith < 10", "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", + "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", + "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / CPU_CLK_UNHALTED.PAUSE_INST", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ippause" }, { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_SWPF", "MetricGroup": "Prefetches;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipswpf" + "MetricName": "tma_info_inst_mix_ipswpf", + "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions.", + "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions", "MetricExpr": "INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "MicroSeq;Pipeline;Ret;Metric", - "MetricName": "tma_info_pipeline_strings_cycles" + "MetricName": "tma_info_pipeline_strings_cycles", + "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1" }, { - "BriefDescription": "Instructions per a microcode Assist invocation.", + "BriefDescription": "Instructions per a microcode Assist invocation", "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", "MetricName": "tma_info_pipeline_ipassist", + "MetricThreshold": "tma_info_pipeline_ipassist < 100000", "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." }, { - "BriefDescription": "Average number of uops fetched from LSD per cycle.", + "BriefDescription": "Average number of uops fetched from LSD per cycle", "MetricExpr": "LSD.UOPS / LSD.CYCLES_ACTIVE", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_lsd" }, { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", + "BriefDescription": "Average number of uops fetched from DSB per cycle", "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_dsb" }, { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", + "BriefDescription": "Average number of uops fetched from MITE per cycle", "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_mite" }, { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", + "BriefDescription": "Average number of Uops issued by front-end when it issued something", "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_frontend_fetch_upc" }, { - "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache).", + "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache)", "MetricExpr": "LSD.UOPS / ( UOPS_ISSUED.ANY )", "MetricGroup": "Fed;LSD;Metric", "MetricName": "tma_info_frontend_lsd_coverage" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( UOPS_ISSUED.ANY )", - "MetricGroup": "DSB;Fed;FetchBW;Metric", - "MetricName": "tma_info_frontend_dsb_coverage" + "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", + "MetricName": "tma_info_frontend_dsb_coverage", + "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 8 > 0.35", + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection.", + "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection", "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / cpu@INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_unknown_branch_cost", "PublicDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection. See Unknown_Branches node." }, { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", + "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "DSBmiss;Metric", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired DSB misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired DSB misses", "MetricExpr": "( FRONTEND_RETIRED.ANY_DSB_MISS * cpu_core@FRONTEND_RETIRED.ANY_DSB_MISS@R ) / tma_info_thread_clks", "MetricGroup": "DSBmiss;Fed;FetchLat;Clocks_Retired", - "MetricName": "tma_info_frontend_dsb_switches_ret" + "MetricName": "tma_info_frontend_dsb_switches_ret", + "MetricThreshold": "tma_info_frontend_dsb_switches_ret > 0.05" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired operations that invoke the Microcode Sequencer.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired operations that invoke the Microcode Sequencer", "MetricExpr": "( FRONTEND_RETIRED.MS_FLOWS * cpu_core@FRONTEND_RETIRED.MS_FLOWS@R ) / tma_info_thread_clks", "MetricGroup": "Fed;FetchLat;MicroSeq;Clocks_Retired", - "MetricName": "tma_info_frontend_ms_latency_ret" + "MetricName": "tma_info_frontend_ms_latency_ret", + "MetricThreshold": "tma_info_frontend_ms_latency_ret > 0.05" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired branches who got branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired branches who got branch address clears", "MetricExpr": "( FRONTEND_RETIRED.UNKNOWN_BRANCH * cpu_core@FRONTEND_RETIRED.UNKNOWN_BRANCH@R ) / tma_info_thread_clks", "MetricGroup": "Fed;FetchLat;Clocks_Retired", "MetricName": "tma_info_frontend_unknown_branches_ret" }, { - "BriefDescription": "Average Latency for L1 instruction cache misses.", + "BriefDescription": "Average Latency for L1 instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / ICACHE_DATA.STALL_PERIODS", "MetricGroup": "Fed;FetchLat;IcMiss;Metric", "MetricName": "tma_info_frontend_icache_miss_latency" }, { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", "MetricGroup": "DSBmiss;Fed;Inst_Metric", - "MetricName": "tma_info_frontend_ipdsb_miss_ret" + "MetricName": "tma_info_frontend_ipdsb_miss_ret", + "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache true code cacheline misses per kilo instruction", "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code" }, { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction", "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code_all" }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) )", - "MetricGroup": "DSBmiss;Fed;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_misses" + "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_misses", + "MetricThreshold": "tma_info_botlnk_dsb_misses > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) )", - "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_bandwidth" + "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", + "MetricThreshold": "tma_info_botlnk_dsb_bandwidth > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", + "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_ic_misses" + "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", + "MetricName": "tma_info_botlnk_l2_ic_misses", + "MetricThreshold": "tma_info_botlnk_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmispredict" + "MetricName": "tma_info_bad_spec_ipmispredict", + "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken" + "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", + "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional backward-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional backward-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN_BWD", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken_bwd" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional forward-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional forward-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN_FWD", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken_fwd" }, { - "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_ret" + "MetricName": "tma_info_bad_spec_ipmisp_ret", + "MetricThreshold": "tma_info_bad_spec_ipmisp_ret < 500" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_indirect" + "MetricName": "tma_info_bad_spec_ipmisp_indirect", + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 8 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", - "MetricGroup": "Bad;BrMispredicts;Core_Metric", - "MetricName": "tma_info_bad_spec_branch_misprediction_cost" + "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", + "MetricName": "tma_info_bad_spec_branch_misprediction_cost", + "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_mispredicts_resteers." }, { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", + "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", "MetricGroup": "BrMispredicts;Metric", "MetricName": "tma_info_bad_spec_spec_clears_ratio" }, { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", + "BriefDescription": "Fraction of branches that are non-taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_nt" }, { - "BriefDescription": "Fraction of branches that are forward taken conditionals.", + "BriefDescription": "Fraction of branches that are forward taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_TAKEN_BWD / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", - "MetricName": "tma_info_branches_cond_tk_bwd" + "MetricName": "tma_info_branches_cond_tk_bwd", + "MetricThreshold": "tma_info_branches_cond_tk_bwd > 0.3" }, { - "BriefDescription": "Fraction of branches that are forward taken conditionals.", + "BriefDescription": "Fraction of branches that are forward taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_TAKEN_FWD / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", - "MetricName": "tma_info_branches_cond_tk_fwd" + "MetricName": "tma_info_branches_cond_tk_fwd", + "MetricThreshold": "tma_info_branches_cond_tk_fwd > 0.2" }, { - "BriefDescription": "Fraction of branches that are CALL or RET.", + "BriefDescription": "Fraction of branches that are CALL or RET", "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_callret" }, { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", + "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN_BWD - BR_INST_RETIRED.COND_TAKEN_FWD - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_jump" }, { - "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group).", + "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group)", "MetricExpr": "1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk_bwd + tma_info_branches_cond_tk_fwd + tma_info_branches_callret + tma_info_branches_jump )", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_other_branches" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PENDING.LOAD / L1D_MISS.LOAD", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PENDING.LOAD / L1D_PENDING.LOAD_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki_load" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * L2_RQSTS.RFO_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_all" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", + "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_fb_hpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the Level 0 within L1D cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the Level 0 within L1D cache [GB / sec]", "MetricExpr": "64 * L1D.L0_REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1dl0_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Offcore;Metric", "MetricName": "tma_info_memory_l3_cache_access_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 4 * tma_info_thread_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", - "MetricName": "tma_info_memory_tlb_page_walks_utilization" + "MetricName": "tma_info_memory_tlb_page_walks_utilization", + "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Fed;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_code_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_load_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand loads.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand loads", "MetricExpr": "( MEM_INST_RETIRED.STLB_MISS_LOADS * cpu_core@MEM_INST_RETIRED.STLB_MISS_LOADS@R ) / tma_info_thread_clks", "MetricGroup": "Mem;MemoryTLB;Clocks_Retired", - "MetricName": "tma_info_memory_tlb_load_stlb_miss_ret" + "MetricName": "tma_info_memory_tlb_load_stlb_miss_ret", + "MetricThreshold": "tma_info_memory_load_stlb_miss_ret > 0.05" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand stores.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand stores", "MetricExpr": "( MEM_INST_RETIRED.STLB_MISS_STORES * cpu_core@MEM_INST_RETIRED.STLB_MISS_STORES@R ) / tma_info_thread_clks", "MetricGroup": "Mem;MemoryTLB;Clocks_Retired", - "MetricName": "tma_info_memory_tlb_store_stlb_miss_ret" + "MetricName": "tma_info_memory_tlb_store_stlb_miss_ret", + "MetricThreshold": "tma_info_memory_store_stlb_miss_ret > 0.05" }, { - "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses.", + "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses", "MetricExpr": "L2_LINES_OUT.USELESS_HWPF / ( L2_LINES_OUT.SILENT + L2_LINES_OUT.NON_SILENT )", "MetricGroup": "Prefetches;Metric", - "MetricName": "tma_info_memory_prefetches_useless_hwpf" + "MetricName": "tma_info_memory_prefetches_useless_hwpf", + "MetricThreshold": "tma_info_memory_useless_hwpf > 0.15" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Latency for L3 cache miss demand Loads.", + "BriefDescription": "Average Latency for L3 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.L3_MISS_DEMAND_DATA_RD / OFFCORE_REQUESTS.L3_MISS_DEMAND_DATA_RD", "MetricGroup": "Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l3_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=0x1@", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", + "BriefDescription": "Un-cacheable retired load per kilo instruction", "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_uc_load_pki" }, { - "BriefDescription": "\"Bus lock\" per kilo instruction.", + "BriefDescription": "\"Bus lock\" per kilo instruction", "MetricExpr": "1000 * SQ_MISC.BUS_LOCK / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_bus_lock_pki" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", - "MetricName": "tma_info_system_kernel_utilization" + "MetricName": "tma_info_system_kernel_utilization", + "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states.", + "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states", "MetricExpr": "CPU_CLK_UNHALTED.C0_WAIT / tma_info_thread_clks", "MetricGroup": "C0Wait;Metric", - "MetricName": "tma_info_system_c0_wait" + "MetricName": "tma_info_system_c0_wait", + "MetricThreshold": "tma_info_system_c0_wait > 0.05" }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "power@energy\\-pkg@ * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", - "MetricName": "tma_info_system_mux" + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "UNC_CLOCK.SOCKET", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", - "MetricName": "tma_info_system_ipfarbranch" + "MetricName": "tma_info_system_ipfarbranch", + "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" } ] \ No newline at end of file diff --git a/BDW/metrics/broadwell_metrics.json b/BDW/metrics/broadwell_metrics.json index bb82be1d..bd8d4e1d 100644 --- a/BDW/metrics/broadwell_metrics.json +++ b/BDW/metrics/broadwell_metrics.json @@ -2,7 +2,7 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for 5th Generation Intel(R) Core(TM) Processor0", - "DatePublished": "11/12/2024", + "DatePublished": "11/15/2024", "Version": "1.0", "Legend": "", "TmaVersion": "5.01", @@ -44,7 +44,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;BvIO;TmaL1;PGO", @@ -86,7 +94,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Frontend;TmaL2", @@ -115,7 +135,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss", @@ -152,7 +188,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB", @@ -189,7 +241,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "~overlap" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -226,7 +294,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Mispredicts_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP", @@ -263,7 +351,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Clears_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;MachineClears", @@ -300,7 +408,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Unknown_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat", @@ -329,7 +457,23 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC, $issueMS, $issueMV, $issueSO" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MicroSeq", @@ -358,7 +502,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LCP(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -387,7 +547,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchLat", @@ -433,7 +609,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;Frontend;TmaL2", @@ -479,7 +663,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -525,7 +721,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;FetchBW", @@ -578,7 +786,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", @@ -640,7 +856,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2", @@ -702,7 +930,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueMC, $issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2", @@ -759,7 +999,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;TmaL1", @@ -853,7 +1101,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20", + "BaseFormula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2", @@ -886,7 +1146,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1, $issueMC" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -923,7 +1199,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -952,7 +1248,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Fwd_Blk(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -989,7 +1305,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Lock_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Offcore", @@ -1030,7 +1366,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......Split_Loads(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Loads(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......Split_Loads(%) > 30", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -1059,7 +1403,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......4K_Aliasing(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......4K_Aliasing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......4K_Aliasing(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -1100,7 +1464,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......FB_Full(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FB_Full(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......FB_Full(%) > 30", + "ThresholdIssues": "$issueBW, $issueSL, $issueSmSt" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW", @@ -1133,7 +1505,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem", @@ -1170,7 +1558,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -1223,7 +1627,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Contested_Accesses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -1276,7 +1700,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Data_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;Offcore;Snoop", @@ -1329,7 +1773,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L3_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat, ~overlap" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -1371,7 +1835,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......SQ_Full(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 30 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -1408,7 +1892,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -1437,7 +1937,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Bandwidth(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -1470,7 +1990,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat;Offcore", @@ -1499,7 +2039,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -1540,7 +2096,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO, $issueSL, ~overlap" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore", @@ -1569,7 +2145,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......False_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -1611,7 +2207,27 @@ "Category": "TMA", "CountDomain": "Core_Utilization", "Threshold": { - "Formula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -1648,7 +2264,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -1742,7 +2378,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2;Compute", @@ -1784,7 +2432,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB", @@ -1858,7 +2522,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -1912,7 +2592,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_0(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -1966,7 +2666,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_1(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -2020,7 +2740,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_2(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -2066,7 +2806,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_3m(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 40 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;PortsUtil", @@ -2120,7 +2880,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........ALU_Op_Utilization(%) > 40" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........ALU_Op_Utilization(%)" + } + ], + "Formula": "a > 40", + "BaseFormula": "metric_TMA_........ALU_Op_Utilization(%) > 40", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -2162,7 +2930,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_0(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_0(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_0(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -2204,7 +2980,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_1(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_1(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_1(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -2246,7 +3030,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_5(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_5(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_5(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -2288,7 +3080,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_6(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_6(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_6(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -2342,7 +3142,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Load_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Load_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -2384,7 +3192,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_2(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_2(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_2(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -2426,7 +3242,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_3(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_3(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_3(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -2468,7 +3292,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Store_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Store_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -2510,7 +3342,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_4(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_4(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_4(%) > 60", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -2552,7 +3392,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_7(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_7(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_7(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -2593,7 +3441,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Retiring(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 70 | b > 10", + "BaseFormula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;TmaL1", @@ -2643,7 +3503,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -2684,7 +3552,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 20 & b > 60", + "BaseFormula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -2717,7 +3597,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......X87_Use(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -2746,7 +3642,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Scalar(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -2775,7 +3687,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -2808,7 +3736,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -2841,7 +3789,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -2891,7 +3859,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -2941,7 +3917,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueMC, $issueMS" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -2983,7 +3971,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Assists(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO", @@ -3037,7 +4041,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......CISC(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3064,6 +4084,11 @@ "BaseFormula": " inst_retired.any / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Summary", "LocateWith": "" @@ -3090,7 +4115,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UopPI > 1.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UopPI" + } + ], + "Formula": "a > 1.05", + "BaseFormula": "metric_TMA_Info_Thread_UopPI > 1.05", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret;Retire", @@ -3118,7 +4151,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UpTB < 4 * 1.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UpTB" + } + ], + "Formula": "a < 4 * 1.5", + "BaseFormula": "metric_TMA_Info_Thread_UpTB < 4 * 1.5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW", @@ -3145,6 +4186,11 @@ "BaseFormula": " 1 / tma_info_thread_ipc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Mem", "LocateWith": "" @@ -3166,6 +4212,11 @@ "BaseFormula": " cpu_clk_unhalted.thread", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", "LocateWith": "" @@ -3200,6 +4251,11 @@ "BaseFormula": " ( 4 ) * tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", "LocateWith": "" @@ -3225,6 +4281,11 @@ "BaseFormula": " uops_executed.thread / uops_issued.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline", "LocateWith": "" @@ -3263,6 +4324,11 @@ "BaseFormula": " inst_retired.any / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;SMT;TmaL1", "LocateWith": "" @@ -3313,6 +4379,11 @@ "BaseFormula": " ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.256b_packed_single ) / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Flops", "LocateWith": "" @@ -3355,6 +4426,11 @@ "BaseFormula": " ( ( fp_arith_inst_retired.scalar ) + ( fp_arith_inst_retired.vector ) ) / ( 2 * tma_info_core_core_clks )", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -3380,6 +4456,11 @@ "BaseFormula": " uops_executed.thread / uops_executed.thread:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "LocateWith": "" @@ -3414,6 +4495,11 @@ "BaseFormula": " ( cpu_clk_unhalted.thread_any / 2 ) if smt_on else tma_info_thread_clks", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "SMT", "LocateWith": "" @@ -3440,7 +4526,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpLoad < 3" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpLoad" + } + ], + "Formula": "a < 3", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpLoad < 3", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -3468,7 +4562,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpStore < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpStore" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpStore < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -3496,7 +4598,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpBranch < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpBranch" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpBranch < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;InsType", @@ -3524,7 +4634,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpCall < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpCall" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpCall < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", @@ -3552,7 +4670,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpTB < 4 * 2 + 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpTB" + } + ], + "Formula": "a < 4 * 2 + 1", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpTB < 4 * 2 + 1", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO", @@ -3579,6 +4705,11 @@ "BaseFormula": " br_inst_retired.all_branches / br_inst_retired.near_taken", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", "LocateWith": "" @@ -3617,7 +4748,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpFLOP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -3649,7 +4788,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -3677,7 +4824,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -3705,7 +4860,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -3737,7 +4900,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX128" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -3769,7 +4940,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX256" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -3792,6 +4971,11 @@ "BaseFormula": " inst_retired.any", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;TmaL1", "LocateWith": "INST_RETIRED.PREC_DIST" @@ -3817,6 +5001,11 @@ "BaseFormula": " ( uops_retired.retire_slots ) / uops_retired.retire_slots:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret", "LocateWith": "" @@ -3855,6 +5044,11 @@ "BaseFormula": " uops_executed.thread / ( ( uops_executed.core:c1 / 2 ) if smt_on else uops_executed.cycles_ge_1_uop_exec )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "LocateWith": "" @@ -3889,7 +5083,19 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 4 > 0.35" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_DSB_Coverage" + }, + { + "Alias": "b", + "Value": "IPC" + } + ], + "Formula": "a < 0.7 & b / 4 > 0.35", + "BaseFormula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 4 > 0.35", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -3916,6 +5122,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -3941,6 +5152,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / baclears.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Fed", "LocateWith": "" @@ -3967,7 +5183,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMispredict" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts", @@ -4003,7 +5227,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect" + } + ], + "Formula": "a < 1000", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -4034,6 +5266,11 @@ "BaseFormula": " l1d_pend_miss.pending / ( mem_load_uops_retired.l1_miss + mem_load_uops_retired.hit_lfb )", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryLat", "LocateWith": "" @@ -4059,6 +5296,11 @@ "BaseFormula": " l1d_pend_miss.pending / l1d_pend_miss.pending_cycles", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryBW", "LocateWith": "" @@ -4084,6 +5326,11 @@ "BaseFormula": " 1000 * mem_load_uops_retired.l1_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -4109,6 +5356,11 @@ "BaseFormula": " 1000 * mem_load_uops_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;Backend;CacheHits", "LocateWith": "" @@ -4134,6 +5386,11 @@ "BaseFormula": " 1000 * l2_rqsts.miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem;Offcore", "LocateWith": "" @@ -4159,6 +5416,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -4184,6 +5446,11 @@ "BaseFormula": " 1000 * offcore_requests.demand_rfo / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheMisses;Offcore", "LocateWith": "" @@ -4213,6 +5480,11 @@ "BaseFormula": " 1000 * ( l2_rqsts.references - l2_rqsts.miss ) / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -4238,6 +5510,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -4263,6 +5540,11 @@ "BaseFormula": " 1000 * mem_load_uops_retired.l3_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -4289,6 +5571,11 @@ "BaseFormula": " 64 * l1d.replacement / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -4315,6 +5602,11 @@ "BaseFormula": " 64 * l2_lines_in.all / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -4341,6 +5633,11 @@ "BaseFormula": " 64 * longest_lat_cache.miss / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -4400,7 +5697,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Page_Walks_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -4428,6 +5733,11 @@ "BaseFormula": " tma_info_memory_l1d_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -4454,6 +5764,11 @@ "BaseFormula": " tma_info_memory_l2_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -4480,6 +5795,11 @@ "BaseFormula": " tma_info_memory_l3_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -4505,6 +5825,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests.demand_data_rd", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Memory_Lat;Offcore", "LocateWith": "" @@ -4530,6 +5855,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests_outstanding.cycles_with_demand_data_rd", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -4555,6 +5885,11 @@ "BaseFormula": " offcore_requests_outstanding.all_data_rd / offcore_requests_outstanding.cycles_with_data_rd", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -4585,6 +5920,11 @@ "BaseFormula": " tma_info_system_cpus_utilized / num_cpus", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "HPC;Summary", "LocateWith": "" @@ -4611,6 +5951,11 @@ "BaseFormula": " cpu_clk_unhalted.ref_tsc / tsc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", "LocateWith": "" @@ -4645,6 +5990,11 @@ "BaseFormula": " tma_info_system_turbo_utilization * tsc / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "SYSTEM", "MetricGroup": "Summary;Power", "LocateWith": "" @@ -4683,6 +6033,11 @@ "BaseFormula": " ( ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.256b_packed_single ) / ( 1000000000 ) ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -4708,6 +6063,11 @@ "BaseFormula": " tma_info_thread_clks / cpu_clk_unhalted.ref_tsc", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -4742,6 +6102,11 @@ "BaseFormula": " 1 - cpu_clk_unhalted.one_thread_active / ( cpu_clk_unhalted.ref_xclk_any / 2 ) if smt_on else 0", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "SMT", "LocateWith": "" @@ -4768,7 +6133,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Kernel_Utilization > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Kernel_Utilization" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_System_Kernel_Utilization > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "OS", @@ -4795,6 +6168,11 @@ "BaseFormula": " cpu_clk_unhalted.thread_p:sup / inst_retired.any_p:sup", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "OS", "LocateWith": "" @@ -4825,6 +6203,11 @@ "BaseFormula": " 64 * ( unc_arb_trk_requests.all + unc_arb_coh_trk_requests.all ) / ( 1000000 ) / tma_info_system_time / 1000", "Category": "TMA", "CountDomain": "GB/sec", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "$issueBW" + }, "ResolutionLevels": "ARB", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC", "LocateWith": "" @@ -4851,6 +6234,11 @@ "BaseFormula": " unc_pkg_energy_status * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "PKG", "MetricGroup": "Power;SoC", "LocateWith": "" @@ -4873,7 +6261,15 @@ "Category": "TMA", "CountDomain": "Seconds", "Threshold": { - "Formula": "metric_TMA_Info_System_Time < 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Time" + } + ], + "Formula": "a < 1", + "BaseFormula": "metric_TMA_Info_System_Time < 1", + "ThresholdIssues": "" }, "ResolutionLevels": "SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -4901,7 +6297,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_MUX" + } + ], + "Formula": "a > 1.1 | a < 0.9", + "BaseFormula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -4924,6 +6328,11 @@ "BaseFormula": " unc_clock.socket", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CLOCK", "MetricGroup": "SoC", "LocateWith": "" @@ -4950,7 +6359,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_IpFarBranch < 1000000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_IpFarBranch" + } + ], + "Formula": "a < 1000000", + "BaseFormula": "metric_TMA_Info_System_IpFarBranch < 1000000", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;OS", diff --git a/BDW/metrics/perf/broadwell_metrics_perf.json b/BDW/metrics/perf/broadwell_metrics_perf.json index 2e1dbaf5..1c0dcf51 100644 --- a/BDW/metrics/perf/broadwell_metrics_perf.json +++ b/BDW/metrics/perf/broadwell_metrics_perf.json @@ -1,962 +1,1059 @@ [ { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "IDQ_UOPS_NOT_DELIVERED.CORE / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_frontend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_frontend_bound > 0.15", "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( 4 ) * IDQ_UOPS_NOT_DELIVERED.CYCLES_0_UOPS_DELIV.CORE / tma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", "MetricName": "tma_fetch_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: RS_EVENTS.EMPTY_END." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE.IFDATA_STALL / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "( ( 14 * ITLB_MISSES.STLB_HIT + cpu@ITLB_MISSES.WALK_DURATION\\,cmask\\=0x1@ + 7 * ITLB_MISSES.WALK_COMPLETED ) ) / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: ITLB_MISSES.WALK_COMPLETED." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "( 12 ) * ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT + BACLEARS.ANY ) / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES * tma_branch_resteers / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT + BACLEARS.ANY )", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Related metrics: tma_branch_mispredicts." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "MACHINE_CLEARS.COUNT * tma_branch_resteers / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT + BACLEARS.ANY )", - "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "tma_branch_resteers - tma_mispredicts_resteers - tma_clears_resteers", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: BACLEARS.ANY." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 2 ) * IDQ.MS_SWITCHES / tma_info_thread_clks", - "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated", + "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES." + "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "ILD_STALL.LCP / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." + "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", - "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty." + "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Related metrics: tma_fetch_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "tma_frontend_bound - tma_fetch_latency", - "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", + "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend." + "MetricThreshold": "tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Related metrics: tma_dsb_switches, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.ALL_MITE_CYCLES_ANY_UOPS - IDQ.ALL_MITE_CYCLES_4_UOPS ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", "ScaleUnit": "100%", + "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.ALL_DSB_CYCLES_ANY_UOPS - IDQ.ALL_DSB_CYCLES_4_UOPS ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", "ScaleUnit": "100%", + "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "( UOPS_ISSUED.ANY - ( UOPS_RETIRED.RETIRE_SLOTS ) + ( 4 ) * ( ( INT_MISC.RECOVERY_CYCLES_ANY / 2 ) if #SMT_on else INT_MISC.RECOVERY_CYCLES ) ) / tma_info_thread_slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_bad_speculation", "ScaleUnit": "100%", + "MetricThreshold": "tma_bad_speculation > 0.15", "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example." }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * tma_bad_speculation", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_mispredicts_resteers." }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "tma_bad_speculation - tma_branch_mispredicts", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", + "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT." + "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "1 - ( tma_frontend_bound + tma_bad_speculation + tma_retiring )", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_backend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_backend_bound > 0.2", "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound." }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( ( CYCLE_ACTIVITY.STALLS_MEM_ANY + RESOURCE_STALLS.SB ) / ( ( CYCLE_ACTIVITY.STALLS_TOTAL + UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC - ( UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC if ( tma_info_thread_ipc > 1.8 ) else UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC ) - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) + RESOURCE_STALLS.SB ) ) ) * tma_backend_bound", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_memory_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two)." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", + "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT." + "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "( ( 8 ) * DTLB_LOAD_MISSES.STLB_HIT + cpu@DTLB_LOAD_MISSES.WALK_DURATION\\,cmask\\=0x1@ + 7 * DTLB_LOAD_MISSES.WALK_COMPLETED ) / tma_info_thread_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_UOPS_RETIRED.STLB_MISS_LOADS." + "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_UOPS_RETIRED.STLB_MISS_LOADS. Related metrics: tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_STORES ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) / tma_info_thread_clks", - "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks", + "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOADS." + "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_UOPS_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset.", + "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset", "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_4k_aliasing", "ScaleUnit": "100%", + "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset. False match is possible; which incur a few cycles load re-issue. However; the short re-issue duration is often hidden by the out-of-order core and HW optimizations; hence a user may safely ignore a high value of this metric unless it manages to propagate up into parent nodes of the hierarchy (e.g. to L1_Bound)." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PEND_MISS.FB_FULL\\,cmask\\=0x1@ / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", + "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory)." + "MetricThreshold": "tma_fb_full > 0.3", + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( MEM_LOAD_UOPS_RETIRED.L3_HIT / ( MEM_LOAD_UOPS_RETIRED.L3_HIT + ( 7 ) * MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) * CYCLE_ACTIVITY.STALLS_L2_MISS / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( 60 ) * ( MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) ) + ( 43 ) * ( MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS." + "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( 43 ) * ( MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT." + "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_contested_accesses, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( 29 ) * ( MEM_LOAD_UOPS_RETIRED.L3_HIT * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT." + "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT. Related metrics: tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "( ( OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 ) if #SMT_on else OFFCORE_REQUESTS_BUFFER.SQ_FULL ) / tma_info_core_core_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( 1 - ( MEM_LOAD_UOPS_RETIRED.L3_HIT / ( MEM_LOAD_UOPS_RETIRED.L3_HIT + ( 7 ) * MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) ) * CYCLE_ACTIVITY.STALLS_L2_MISS / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that)." + "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that)." + "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "RESOURCE_STALLS.SB / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_UOPS_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( L2_RQSTS.RFO_HIT * ( 9 ) * ( 1 - ( MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)." + "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( 60 ) * OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM." + "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related metrics: tma_contested_accesses, tma_data_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "2 * MEM_UOPS_RETIRED.SPLIT_STORES / tma_info_core_core_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization", + "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", "ScaleUnit": "100%", - "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_UOPS_RETIRED.SPLIT_STORES." + "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_UOPS_RETIRED.SPLIT_STORES. Related metrics: tma_port_4." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( ( 8 ) * DTLB_STORE_MISSES.STLB_HIT + cpu@DTLB_STORE_MISSES.WALK_DURATION\\,cmask\\=0x1@ + 7 * DTLB_STORE_MISSES.WALK_COMPLETED ) / tma_info_thread_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_UOPS_RETIRED.STLB_MISS_STORES." + "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_UOPS_RETIRED.STLB_MISS_STORES. Related metrics: tma_dtlb_load." }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "tma_backend_bound - tma_memory_bound", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_core_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations)." }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.FPU_DIV_ACTIVE / tma_info_core_core_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", "ScaleUnit": "100%", + "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.FPU_DIV_ACTIVE." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( ( ( CYCLE_ACTIVITY.STALLS_TOTAL + UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC - ( UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC if ( tma_info_thread_ipc > 1.8 ) else UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC ) - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) + RESOURCE_STALLS.SB ) ) - RESOURCE_STALLS.SB - CYCLE_ACTIVITY.STALLS_MEM_ANY ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,inv\\=0x1\\,cmask\\=0x1@ ) / 2 if #SMT_on else ( CYCLE_ACTIVITY.STALLS_TOTAL - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) ) ) / tma_info_core_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ ) / 2 if #SMT_on else ( UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC - UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC ) ) / tma_info_core_core_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful." + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ ) / 2 if #SMT_on else ( UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC - UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC ) ) / tma_info_core_core_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop." + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ / 2 ) if #SMT_on else UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC ) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT.PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_0 / tma_info_core_core_clks", - "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0." + "MetricThreshold": "tma_port_0 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_1 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1." + "MetricThreshold": "tma_port_1 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_5 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_5", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5." + "MetricThreshold": "tma_port_5 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_6 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1." + "MetricThreshold": "tma_port_6 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "( UOPS_DISPATCHED_PORT.PORT_2 + UOPS_DISPATCHED_PORT.PORT_3 + UOPS_DISPATCHED_PORT.PORT_7 - UOPS_DISPATCHED_PORT.PORT_4 ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_op_utilization > 0.6" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_2 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_group;Core_Clocks", "MetricName": "tma_port_2", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_2 > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads). Sample with: UOPS_DISPATCHED_PORT.PORT_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_3 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_group;Core_Clocks", "MetricName": "tma_port_3", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_3 > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads). Sample with: UOPS_DISPATCHED_PORT.PORT_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_op_utilization > 0.6" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks;tma_issueSpSt", "MetricName": "tma_port_4", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data). Sample with: UOPS_DISPATCHED_PORT.PORT_4." + "MetricThreshold": "tma_port_4 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data). Sample with: UOPS_DISPATCHED_PORT.PORT_4. Related metrics: tma_split_stores." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_7 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks", "MetricName": "tma_port_7", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_7 > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address). Sample with: UOPS_DISPATCHED_PORT.PORT_7." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / tma_info_thread_slots", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_retiring", "ScaleUnit": "100%", + "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.1", "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.RETIRE_SLOTS." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_light_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "INST_RETIRED.X87 * tma_info_thread_uoppi / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", "ScaleUnit": "100%", + "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "tma_microcode_sequencer", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_heavy_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+])." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "( ( UOPS_RETIRED.RETIRE_SLOTS ) / UOPS_ISSUED.ANY ) * IDQ.MS_UOPS / tma_info_thread_slots", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS." + "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( 66 ) * OTHER_ASSISTS.ANY_WB_ASSIST / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY_WB_ASSIST." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", "ScaleUnit": "100%", + "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources." }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", - "MetricName": "tma_info_thread_uoppi" + "MetricName": "tma_info_thread_uoppi", + "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", - "MetricName": "tma_info_thread_uptb" + "MetricName": "tma_info_thread_uptb", + "MetricThreshold": "tma_info_thread_uptb < 4 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "( 4 ) * tma_info_core_core_clks", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / tma_info_core_core_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "( CPU_CLK_UNHALTED.THREAD_ANY / 2 ) if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_UOPS_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipload" + "MetricName": "tma_info_inst_mix_ipload", + "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_UOPS_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipstore" + "MetricName": "tma_info_inst_mix_ipstore", + "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipbranch" + "MetricName": "tma_info_inst_mix_ipbranch", + "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipcall" + "MetricName": "tma_info_inst_mix_ipcall", + "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", - "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_iptb" + "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", + "MetricName": "tma_info_inst_mix_iptb", + "MetricThreshold": "tma_info_inst_mix_iptb < 4 * 2 + 1", + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipflop" + "MetricName": "tma_info_inst_mix_ipflop", + "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", + "MetricThreshold": "tma_info_inst_mix_iparith < 10", "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", + "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", + "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / cpu@UOPS_RETIRED.RETIRE_SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core.", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core", "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ / 2 ) if #SMT_on else UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC )", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( ( IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS ) )", - "MetricGroup": "DSB;Fed;FetchBW;Metric", - "MetricName": "tma_info_frontend_dsb_coverage" + "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", + "MetricName": "tma_info_frontend_dsb_coverage", + "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 4 > 0.35", + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmispredict" + "MetricName": "tma_info_bad_spec_ipmispredict", + "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / ( ( ( UOPS_RETIRED.RETIRE_SLOTS ) / UOPS_ISSUED.ANY ) * BR_MISP_EXEC.INDIRECT )", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_indirect" + "MetricName": "tma_info_bad_spec_ipmisp_indirect", + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_UOPS_RETIRED.L1_MISS + MEM_LOAD_UOPS_RETIRED.HIT_LFB )", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_UOPS_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_UOPS_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * OFFCORE_REQUESTS.DEMAND_RFO / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_all" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_UOPS_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( cpu@ITLB_MISSES.WALK_DURATION\\,cmask\\=0x1@ + cpu@DTLB_LOAD_MISSES.WALK_DURATION\\,cmask\\=0x1@ + cpu@DTLB_STORE_MISSES.WALK_DURATION\\,cmask\\=0x1@ + 7 * ( DTLB_STORE_MISSES.WALK_COMPLETED + DTLB_LOAD_MISSES.WALK_COMPLETED + ITLB_MISSES.WALK_COMPLETED ) ) / tma_info_core_core_clks", "MetricGroup": "Mem;MemoryTLB;Core_Metric", - "MetricName": "tma_info_memory_tlb_page_walks_utilization" + "MetricName": "tma_info_memory_tlb_page_walks_utilization", + "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / ( CPU_CLK_UNHALTED.REF_XCLK_ANY / 2 ) if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", - "MetricName": "tma_info_system_kernel_utilization" + "MetricName": "tma_info_system_kernel_utilization", + "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "64 * ( UNC_ARB_TRK_REQUESTS.ALL + UNC_ARB_COH_TRK_REQUESTS.ALL ) / ( 1000000 ) / tma_info_system_time / 1000", - "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec", - "MetricName": "tma_info_system_dram_bw_use" + "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", + "MetricName": "tma_info_system_dram_bw_use", + "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_fb_full, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "power@energy\\-pkg@ * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", - "MetricName": "tma_info_system_mux" + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "UNC_CLOCK.SOCKET", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", - "MetricName": "tma_info_system_ipfarbranch" + "MetricName": "tma_info_system_ipfarbranch", + "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" } ] \ No newline at end of file diff --git a/BDX/metrics/broadwellx_metrics.json b/BDX/metrics/broadwellx_metrics.json index b7113d4f..b6cf4348 100644 --- a/BDX/metrics/broadwellx_metrics.json +++ b/BDX/metrics/broadwellx_metrics.json @@ -2,7 +2,7 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for Intel(R) Xeon(R) Processor E5 v4 Family Based on the Broadwell Microarchitecture0", - "DatePublished": "11/05/2024", + "DatePublished": "11/15/2024", "Version": "1.0", "Legend": "", "TmaVersion": "5.01", @@ -6184,6 +6184,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -6503,6 +6508,11 @@ "BaseFormula": " 1000 * offcore_requests.demand_rfo / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheMisses;Offcore", "LocateWith": "" @@ -7417,7 +7427,15 @@ "Category": "TMA", "CountDomain": "Seconds", "Threshold": { - "Formula": "metric_TMA_Info_System_Time < 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Time" + } + ], + "Formula": "a < 1", + "BaseFormula": "metric_TMA_Info_System_Time < 1", + "ThresholdIssues": "" }, "ResolutionLevels": "SOCKET, SYSTEM", "MetricGroup": "Summary", diff --git a/BDX/metrics/perf/broadwellx_metrics_perf.json b/BDX/metrics/perf/broadwellx_metrics_perf.json index 9f6830e9..c9971d74 100644 --- a/BDX/metrics/perf/broadwellx_metrics_perf.json +++ b/BDX/metrics/perf/broadwellx_metrics_perf.json @@ -1,125 +1,125 @@ [ { - "BriefDescription": "CPU operating frequency (in GHz).", - "MetricExpr": "(( CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000)", + "BriefDescription": "CPU operating frequency (in GHz)", + "MetricExpr": "((CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000)", "MetricGroup": "", "MetricName": "cpu_operating_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Percentage of time spent in the active CPU power state C0.", + "BriefDescription": "Percentage of time spent in the active CPU power state C0", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "", "MetricName": "cpu_utilization", "ScaleUnit": "100%" }, { - "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles.", + "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles", "MetricExpr": "CPU_CLK_UNHALTED.THREAD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions", "MetricExpr": "MEM_UOPS_RETIRED.ALL_LOADS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "loads_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions", "MetricExpr": "MEM_UOPS_RETIRED.ALL_STORES / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "stores_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L1D.REPLACEMENT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_UOPS_RETIRED.L1_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions", "MetricExpr": "L2_RQSTS.ALL_CODE_RD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1_i_code_read_misses_with_prefetches_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_UOPS_RETIRED.L2_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L2_LINES_IN.ALL / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_UOPS_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions", "MetricExpr": "L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_code_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", - "MetricExpr": "( cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x182@ + cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x192@ ) / INST_RETIRED.ANY", + "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", + "MetricExpr": "(cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x182@ + cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x192@) / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_data_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", - "MetricExpr": "( cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x181@ + cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x191@ ) / INST_RETIRED.ANY", + "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", + "MetricExpr": "(cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x181@ + cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x191@) / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_code_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) in nano seconds.", - "MetricExpr": "( 1000000000 * ( cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x182@ ) / ( UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) in nano seconds", + "MetricExpr": "( 1000000000 * (cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x182@) / (UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_data_read_demand_plus_prefetch_miss_latency", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to local memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( cbox@UNC_C_TOR_OCCUPANCY.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ ) / ( UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to local memory in nano seconds", + "MetricExpr": "( 1000000000 * (cbox@UNC_C_TOR_OCCUPANCY.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@) / (UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_data_read_demand_plus_prefetch_miss_latency_for_local_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to remote memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( cbox@UNC_C_TOR_OCCUPANCY.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@ ) / ( UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to remote memory in nano seconds", + "MetricExpr": "( 1000000000 * (cbox@UNC_C_TOR_OCCUPANCY.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@) / (UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_data_read_demand_plus_prefetch_miss_latency_for_remote_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_mpi", @@ -127,7 +127,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the ITLB (Instruction TLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED_2M_4M / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_large_page_mpi", @@ -135,7 +135,7 @@ "PublicDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the Instruction Translation Lookaside Buffer (ITLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions", "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_load_mpi", @@ -143,7 +143,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions", "MetricExpr": "DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_store_mpi", @@ -151,98 +151,98 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ / ( cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ + cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@ )", + "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ / (cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ + cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_local_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@ / ( cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ + cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@ )", + "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@ / (cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ + cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_remote_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Uncore operating frequency in GHz.", - "MetricExpr": "( UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) / 1000000000) / duration_time", + "BriefDescription": "Uncore operating frequency in GHz", + "MetricExpr": "(UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) / 1000000000) / duration_time", "MetricGroup": "", "MetricName": "uncore_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Intel(R) Quick Path Interconnect (QPI) data transmit bandwidth (MB/sec).", - "MetricExpr": "( UNC_Q_TxL_FLITS_G0.DATA * 8 / 1000000) / duration_time", + "BriefDescription": "Intel(R) Quick Path Interconnect (QPI) data transmit bandwidth (MB/sec)", + "MetricExpr": "(UNC_Q_TxL_FLITS_G0.DATA * 8 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "qpi_data_transmit_bw", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory read bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory read bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory write bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory write bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory bandwidth (MB/sec)", + "MetricExpr": "((UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_total", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU.", - "MetricExpr": "( cbox@UNC_C_TOR_INSERTS.OPCODE\\,filter_opc\\=0x19e@ * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU", + "MetricExpr": "(cbox@UNC_C_TOR_INSERTS.OPCODE\\,filter_opc\\=0x19e@ * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU.", - "MetricExpr": "(( cbox@UNC_C_TOR_INSERTS.OPCODE\\,filter_opc\\=0x1c8\\,filter_tid\\=0x3e@ + cbox@UNC_C_TOR_INSERTS.OPCODE\\,filter_opc\\=0x180\\,filter_tid\\=0x3e@ ) * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU", + "MetricExpr": "((cbox@UNC_C_TOR_INSERTS.OPCODE\\,filter_opc\\=0x1c8\\,filter_tid\\=0x3e@ + cbox@UNC_C_TOR_INSERTS.OPCODE\\,filter_opc\\=0x180\\,filter_tid\\=0x3e@) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.DSB_UOPS / UOPS_ISSUED.ANY )", + "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.DSB_UOPS / UOPS_ISSUED.ANY)", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_decoded_icache", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MITE_UOPS / UOPS_ISSUED.ANY )", + "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MITE_UOPS / UOPS_ISSUED.ANY)", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_legacy_decode_pipeline", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MS_UOPS / UOPS_ISSUED.ANY )", + "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MS_UOPS / UOPS_ISSUED.ANY)", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_microcode_sequencer", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from loop stream detector(LSD) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( LSD.UOPS / UOPS_ISSUED.ANY )", + "BriefDescription": "Uops delivered from loop stream detector(LSD) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(LSD.UOPS / UOPS_ISSUED.ANY)", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_loop_stream_detector", "ScaleUnit": "100%" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "IDQ_UOPS_NOT_DELIVERED.CORE / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_frontend_bound", @@ -251,7 +251,7 @@ "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( 4 ) * IDQ_UOPS_NOT_DELIVERED.CYCLES_0_UOPS_DELIV.CORE / tma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", "MetricName": "tma_fetch_latency", @@ -260,7 +260,7 @@ "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: RS_EVENTS.EMPTY_END." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE.IFDATA_STALL / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", @@ -268,7 +268,7 @@ "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "( ( 14 * ITLB_MISSES.STLB_HIT + cpu@ITLB_MISSES.WALK_DURATION\\,cmask\\=0x1@ + 7 * ITLB_MISSES.WALK_COMPLETED ) ) / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", @@ -277,7 +277,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: ITLB_MISSES.WALK_COMPLETED." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "( 12 ) * ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT + BACLEARS.ANY ) / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", @@ -286,7 +286,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES * tma_branch_resteers / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT + BACLEARS.ANY )", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", @@ -295,16 +295,16 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Related metrics: tma_branch_mispredicts." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "MACHINE_CLEARS.COUNT * tma_branch_resteers / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT + BACLEARS.ANY )", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", "ScaleUnit": "100%", "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Related metrics: tma_l1_bound, tma_machine_clears, tma_ms_switches, tma_microcode_sequencer." + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "tma_branch_resteers - tma_mispredicts_resteers - tma_clears_resteers", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", @@ -313,43 +313,43 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: BACLEARS.ANY." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 2 ) * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_l1_bound, tma_machine_clears, tma_clears_resteers, tma_microcode_sequencer." + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "ILD_STALL.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_info_frontend_dsb_coverage, tma_fetch_bandwidth, tma_info_inst_mix_iptb." + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Related metrics: tma_lcp, tma_info_frontend_dsb_coverage, tma_fetch_bandwidth, tma_info_inst_mix_iptb." + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Related metrics: tma_fetch_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "tma_frontend_bound - tma_fetch_latency", "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", "MetricThreshold": "tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Related metrics: tma_dsb_switches, tma_lcp, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Related metrics: tma_dsb_switches, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.ALL_MITE_CYCLES_ANY_UOPS - IDQ.ALL_MITE_CYCLES_4_UOPS ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", @@ -358,7 +358,7 @@ "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.ALL_DSB_CYCLES_ANY_UOPS - IDQ.ALL_DSB_CYCLES_4_UOPS ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", @@ -367,7 +367,7 @@ "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "( UOPS_ISSUED.ANY - ( UOPS_RETIRED.RETIRE_SLOTS ) + ( 4 ) * ( ( INT_MISC.RECOVERY_CYCLES_ANY / 2 ) if #SMT_on else INT_MISC.RECOVERY_CYCLES ) ) / tma_info_thread_slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_bad_speculation", @@ -376,7 +376,7 @@ "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example." }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * tma_bad_speculation", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", @@ -385,16 +385,16 @@ "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_mispredicts_resteers." }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "tma_bad_speculation - tma_branch_mispredicts", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_ms_switches, tma_microcode_sequencer, tma_clears_resteers, tma_remote_cache, tma_false_sharing, tma_l1_bound, tma_data_sharing, tma_contested_accesses." + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches, tma_remote_cache." }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "1 - ( tma_frontend_bound + tma_bad_speculation + tma_retiring )", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_backend_bound", @@ -403,7 +403,7 @@ "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound." }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( ( CYCLE_ACTIVITY.STALLS_MEM_ANY + RESOURCE_STALLS.SB ) / ( ( CYCLE_ACTIVITY.STALLS_TOTAL + UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC - ( UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC if ( tma_info_thread_ipc > 1.8 ) else UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC ) - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) + RESOURCE_STALLS.SB ) ) ) * tma_backend_bound", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_memory_bound", @@ -412,16 +412,16 @@ "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two)." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT. Related metrics: tma_ms_switches, tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ports_utilized_1." + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "( ( 8 ) * DTLB_LOAD_MISSES.STLB_HIT + cpu@DTLB_LOAD_MISSES.WALK_DURATION\\,cmask\\=0x1@ + 7 * DTLB_LOAD_MISSES.WALK_COMPLETED ) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", @@ -430,7 +430,7 @@ "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_UOPS_RETIRED.STLB_MISS_LOADS. Related metrics: tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", @@ -439,7 +439,7 @@ "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_STORES ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", @@ -448,7 +448,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", @@ -457,7 +457,7 @@ "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_UOPS_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset.", + "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset", "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_4k_aliasing", @@ -466,16 +466,16 @@ "PublicDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset. False match is possible; which incur a few cycles load re-issue. However; the short re-issue duration is often hidden by the out-of-order core and HW optimizations; hence a user may safely ignore a high value of this metric unless it manages to propagate up into parent nodes of the hierarchy (e.g. to L1_Bound)." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PEND_MISS.FB_FULL\\,cmask\\=0x1@ / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", "MetricThreshold": "tma_fb_full > 0.3", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_info_system_dram_bw_use, tma_sq_full, tma_mem_bandwidth, tma_store_latency." + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", @@ -484,7 +484,7 @@ "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( MEM_LOAD_UOPS_RETIRED.L3_HIT / ( MEM_LOAD_UOPS_RETIRED.L3_HIT + ( 7 ) * MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) * CYCLE_ACTIVITY.STALLS_L2_MISS / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", @@ -493,43 +493,43 @@ "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( 60 ) * ( MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) + ( 43 ) * ( MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) ) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_machine_clears, tma_false_sharing, tma_remote_cache, tma_data_sharing." + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_data_sharing, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( 43 ) * ( MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_machine_clears, tma_false_sharing, tma_remote_cache, tma_contested_accesses." + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_contested_accesses, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( 41 ) * ( MEM_LOAD_UOPS_RETIRED.L3_HIT * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT. Related metrics: tma_store_latency, tma_mem_latency, tma_branch_resteers." + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT. Related metrics: tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "( ( OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 ) if #SMT_on else OFFCORE_REQUESTS_BUFFER.SQ_FULL ) / tma_info_core_core_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", "ScaleUnit": "100%", "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_info_system_dram_bw_use, tma_fb_full, tma_mem_bandwidth." + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( 1 - ( MEM_LOAD_UOPS_RETIRED.L3_HIT / ( MEM_LOAD_UOPS_RETIRED.L3_HIT + ( 7 ) * MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) ) * CYCLE_ACTIVITY.STALLS_L2_MISS / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", @@ -538,7 +538,7 @@ "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", @@ -547,7 +547,7 @@ "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", @@ -556,7 +556,7 @@ "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory", "MetricExpr": "( 200 ) * ( MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_local_mem", @@ -565,7 +565,7 @@ "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory. Caching will improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory", "MetricExpr": "( 310 ) * ( MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_remote_mem", @@ -574,16 +574,16 @@ "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues", "MetricExpr": "( ( 200 ) * ( MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) + ( 180 ) * ( MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) ) / tma_info_thread_clks", "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_remote_cache", "ScaleUnit": "100%", "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_machine_clears, tma_false_sharing, tma_data_sharing, tma_contested_accesses." + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "RESOURCE_STALLS.SB / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", @@ -592,25 +592,25 @@ "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_UOPS_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( L2_RQSTS.RFO_HIT * ( 9 ) * ( 1 - ( MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_l3_hit_latency, tma_fb_full, tma_lock_latency, tma_branch_resteers." + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( ( 200 ) * OFFCORE_RESPONSE.DEMAND_RFO.LLC_MISS.REMOTE_HITM + ( 60 ) * OFFCORE_RESPONSE.DEMAND_RFO.LLC_HIT.HITM_OTHER_CORE ) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM, OFFCORE_RESPONSE.DEMAND_RFO.LLC_HIT.HITM_OTHER_CORE, OFFCORE_RESPONSE.DEMAND_RFO.LLC_MISS.REMOTE_HITM. Related metrics: tma_machine_clears, tma_remote_cache, tma_data_sharing, tma_contested_accesses." + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM, OFFCORE_RESPONSE.DEMAND_RFO.LLC_HIT.HITM_OTHER_CORE, OFFCORE_RESPONSE.DEMAND_RFO.LLC_MISS.REMOTE_HITM. Related metrics: tma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "2 * MEM_UOPS_RETIRED.SPLIT_STORES / tma_info_core_core_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", @@ -619,7 +619,7 @@ "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_UOPS_RETIRED.SPLIT_STORES. Related metrics: tma_port_4." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( ( 8 ) * DTLB_STORE_MISSES.STLB_HIT + cpu@DTLB_STORE_MISSES.WALK_DURATION\\,cmask\\=0x1@ + 7 * DTLB_STORE_MISSES.WALK_COMPLETED ) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", @@ -628,7 +628,7 @@ "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_UOPS_RETIRED.STLB_MISS_STORES. Related metrics: tma_dtlb_load." }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "tma_backend_bound - tma_memory_bound", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_core_bound", @@ -637,7 +637,7 @@ "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations)." }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.FPU_DIV_ACTIVE / tma_info_core_core_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", @@ -646,7 +646,7 @@ "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.FPU_DIV_ACTIVE." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( ( ( CYCLE_ACTIVITY.STALLS_TOTAL + UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC - ( UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC if ( tma_info_thread_ipc > 1.8 ) else UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC ) - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) + RESOURCE_STALLS.SB ) ) - RESOURCE_STALLS.SB - CYCLE_ACTIVITY.STALLS_MEM_ANY ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", @@ -655,7 +655,7 @@ "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,inv\\=0x1\\,cmask\\=0x1@ ) / 2 if #SMT_on else ( CYCLE_ACTIVITY.STALLS_TOTAL - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) ) ) / tma_info_core_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", @@ -664,7 +664,7 @@ "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ ) / 2 if #SMT_on else ( UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC - UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC ) ) / tma_info_core_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", @@ -673,16 +673,16 @@ "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ ) / 2 if #SMT_on else ( UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC - UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC ) ) / tma_info_core_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Related metrics: tma_fp_vector_256b, tma_fp_scalar, tma_fp_vector_128b, tma_port_5, tma_fp_vector, tma_port_1, tma_port_0, tma_port_6." + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ / 2 ) if #SMT_on else UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC ) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", @@ -690,7 +690,7 @@ "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT.PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", @@ -698,43 +698,43 @@ "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_0 / tma_info_core_core_clks", "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", "ScaleUnit": "100%", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_ports_utilized_2, tma_fp_vector_256b, tma_fp_scalar, tma_fp_vector_128b, tma_port_5, tma_fp_vector, tma_port_1, tma_port_6." + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_1 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", "ScaleUnit": "100%", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_ports_utilized_2, tma_fp_vector_256b, tma_fp_scalar, tma_fp_vector_128b, tma_port_5, tma_fp_vector, tma_port_0, tma_port_6." + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_5 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_5", "ScaleUnit": "100%", "MetricThreshold": "tma_port_5 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_ports_utilized_2, tma_fp_vector_256b, tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector, tma_port_1, tma_port_0, tma_port_6." + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_6 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", "ScaleUnit": "100%", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_ports_utilized_2, tma_fp_vector_256b, tma_fp_scalar, tma_fp_vector_128b, tma_port_5, tma_fp_vector, tma_port_1, tma_port_0." + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "( UOPS_DISPATCHED_PORT.PORT_2 + UOPS_DISPATCHED_PORT.PORT_3 + UOPS_DISPATCHED_PORT.PORT_7 - UOPS_DISPATCHED_PORT.PORT_4 ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", @@ -742,7 +742,7 @@ "MetricThreshold": "tma_load_op_utilization > 0.6" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_2 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_group;Core_Clocks", "MetricName": "tma_port_2", @@ -751,7 +751,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads). Sample with: UOPS_DISPATCHED_PORT.PORT_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_3 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_group;Core_Clocks", "MetricName": "tma_port_3", @@ -760,7 +760,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads). Sample with: UOPS_DISPATCHED_PORT.PORT_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", @@ -768,7 +768,7 @@ "MetricThreshold": "tma_store_op_utilization > 0.6" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks;tma_issueSpSt", "MetricName": "tma_port_4", @@ -777,7 +777,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data). Sample with: UOPS_DISPATCHED_PORT.PORT_4. Related metrics: tma_split_stores." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_7 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks", "MetricName": "tma_port_7", @@ -786,7 +786,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address). Sample with: UOPS_DISPATCHED_PORT.PORT_7." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / tma_info_thread_slots", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_retiring", @@ -795,7 +795,7 @@ "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.RETIRE_SLOTS." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_light_operations", @@ -804,7 +804,7 @@ "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", @@ -813,7 +813,7 @@ "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "INST_RETIRED.X87 * tma_info_thread_uoppi / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", @@ -822,43 +822,43 @@ "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR ) / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", "ScaleUnit": "100%", "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_ports_utilized_2, tma_fp_vector_256b, tma_fp_vector_128b, tma_port_5, tma_fp_vector, tma_port_1, tma_port_0, tma_port_6." + "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR ) / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", "ScaleUnit": "100%", "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_ports_utilized_2, tma_fp_vector_256b, tma_fp_scalar, tma_fp_vector_128b, tma_port_5, tma_port_1, tma_port_0, tma_port_6." + "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE ) / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_ports_utilized_2, tma_fp_vector_256b, tma_fp_scalar, tma_port_5, tma_fp_vector, tma_port_1, tma_port_0, tma_port_6." + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", "ScaleUnit": "100%", "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_ports_utilized_2, tma_fp_scalar, tma_fp_vector_128b, tma_port_5, tma_fp_vector, tma_port_1, tma_port_0, tma_port_6." + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "tma_microcode_sequencer", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_heavy_operations", @@ -867,16 +867,16 @@ "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+])." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "( ( UOPS_RETIRED.RETIRE_SLOTS ) / UOPS_ISSUED.ANY ) * IDQ.MS_UOPS / tma_info_thread_slots", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_l1_bound, tma_machine_clears, tma_clears_resteers, tma_ms_switches." + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( 66 ) * OTHER_ASSISTS.ANY_WB_ASSIST / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", @@ -885,7 +885,7 @@ "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY_WB_ASSIST." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", @@ -894,133 +894,133 @@ "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources." }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", "MetricName": "tma_info_thread_uoppi", "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", "MetricName": "tma_info_thread_uptb", "MetricThreshold": "tma_info_thread_uptb < 4 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "( 4 ) * tma_info_core_core_clks", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / tma_info_core_core_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "( CPU_CLK_UNHALTED.THREAD_ANY / 2 ) if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_UOPS_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipload", "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_UOPS_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipstore", "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipbranch", "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", "MetricName": "tma_info_inst_mix_ipcall", "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", "MetricThreshold": "tma_info_inst_mix_iptb < 4 * 2 + 1", - "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_lcp, tma_info_frontend_dsb_coverage, tma_fetch_bandwidth." + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipflop", "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", @@ -1028,7 +1028,7 @@ "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", @@ -1036,7 +1036,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", @@ -1044,7 +1044,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", @@ -1052,7 +1052,7 @@ "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", @@ -1060,285 +1060,286 @@ "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / cpu@UOPS_RETIRED.RETIRE_SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core.", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core", "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ / 2 ) if #SMT_on else UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC )", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( ( IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS ) )", "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", "MetricName": "tma_info_frontend_dsb_coverage", "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 4 > 0.35", - "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_lcp, tma_fetch_bandwidth, tma_info_inst_mix_iptb." + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmispredict", "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / ( ( ( UOPS_RETIRED.RETIRE_SLOTS ) / UOPS_ISSUED.ANY ) * BR_MISP_EXEC.INDIRECT )", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_indirect", "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_UOPS_RETIRED.L1_MISS + MEM_LOAD_UOPS_RETIRED.HIT_LFB )", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_UOPS_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_UOPS_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * OFFCORE_REQUESTS.DEMAND_RFO / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_all" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_UOPS_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_DURATION + DTLB_LOAD_MISSES.WALK_DURATION + DTLB_STORE_MISSES.WALK_DURATION + 7 * ( DTLB_STORE_MISSES.WALK_COMPLETED + DTLB_LOAD_MISSES.WALK_COMPLETED + ITLB_MISSES.WALK_COMPLETED ) ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", "MetricName": "tma_info_memory_tlb_page_walks_utilization", "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz].", + "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz]", "MetricExpr": "tma_info_system_socket_clks / 1e9 / tma_info_system_time", "MetricGroup": "SoC;System_Metric", "MetricName": "tma_info_system_uncore_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / ( CPU_CLK_UNHALTED.REF_XCLK_ANY / 2 ) if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_utilization", "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "( 64 * ( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", "MetricName": "tma_info_system_dram_bw_use", "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_fb_full, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Average latency of data read request to external memory (in nanoseconds).", + "BriefDescription": "Average latency of data read request to external memory (in nanoseconds)", "MetricExpr": "( 1000000000 ) * ( cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x182@ ) / ( tma_info_system_socket_clks / tma_info_system_time )", "MetricGroup": "Mem;MemoryLat;SoC;NanoSeconds", "MetricName": "tma_info_system_mem_read_latency", "PublicDescription": "Average latency of data read request to external memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetches. ([RKL+]memory-controller only)." }, { - "BriefDescription": "Average number of parallel data read requests to external memory.", + "BriefDescription": "Average number of parallel data read requests to external memory", "MetricExpr": "cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filter_opc\\=0x182@", "MetricGroup": "Mem;MemoryBW;SoC;System_Metric", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches." }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "( power@energy\\-pkg@ * ( 61 ) + 15.6 * power@energy\\-ram@ ) / ( ( duration_time ) * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", "MetricName": "tma_info_system_mux", "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "cbox_0@event\\=0x0@", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", "MetricName": "tma_info_system_ipfarbranch", diff --git a/CLX/metrics/cascadelakex_metrics.json b/CLX/metrics/cascadelakex_metrics.json index 151034b2..13de24dc 100644 --- a/CLX/metrics/cascadelakex_metrics.json +++ b/CLX/metrics/cascadelakex_metrics.json @@ -2,7 +2,7 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for 2nd Generation Intel(R) Xeon(R) Processor Scalable Family based on Cascade Lake product0", - "DatePublished": "11/05/2024", + "DatePublished": "11/15/2024", "Version": "1.0", "Legend": "", "TmaVersion": "5.01", @@ -1039,7 +1039,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Mispredictions > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Mispredictions" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Mispredictions > 20", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP", @@ -1112,7 +1120,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Big_Code > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Big_Code" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Big_Code > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB", @@ -1237,7 +1253,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Instruction_Fetch_BW" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", @@ -1458,7 +1482,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore", @@ -1707,7 +1739,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Latency" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;Mem;MemoryLat;Offcore", @@ -1920,7 +1960,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Data_TLBs" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore", @@ -2145,7 +2193,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Synchronization > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Synchronization" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Synchronization > 10", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;LockCont;Mem;Offcore", @@ -2246,7 +2302,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Compute_Bound_Est" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20", + "ThresholdIssues": "$issueComp" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Cor", @@ -2411,7 +2475,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Irregular_Overhead > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Irregular_Overhead" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Irregular_Overhead > 10", + "ThresholdIssues": "$issueMS" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BvIO;Cor;Ret", @@ -2788,7 +2860,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Other_Bottlenecks" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;Cor;Offcore", @@ -2837,7 +2917,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Branching_Overhead > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Branching_Overhead" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Bottleneck_Branching_Overhead > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Ret", @@ -2914,7 +3002,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Useful_Work > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Useful_Work" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Useful_Work > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;Ret", @@ -2955,7 +3051,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;BvIO;TmaL1;PGO", @@ -2997,7 +3101,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Frontend;TmaL2", @@ -3030,7 +3146,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss", @@ -3059,7 +3191,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB", @@ -3090,9 +3238,29 @@ "Formula": "100 * ( max( 0 , ( a / ( b ) ) - ( c / ( b ) ) ) )", "BaseFormula": " max( 0 , tma_itlb_misses - tma_code_stlb_miss )", "Category": "TMA", - "CountDomain": "", + "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "> 5 & P" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3119,9 +3287,29 @@ "Formula": "100 * ( a / ( b ) )", "BaseFormula": " itlb_misses.walk_active / tma_info_thread_clks", "Category": "TMA", - "CountDomain": "", + "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "> 5 & P" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3158,7 +3346,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3195,7 +3407,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3228,7 +3464,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "~overlap" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -3265,7 +3517,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Mispredicts_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP", @@ -3302,7 +3574,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Clears_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;MachineClears", @@ -3331,7 +3623,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Unknown_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat", @@ -3360,7 +3672,23 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC, $issueMS, $issueMV, $issueSO" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MicroSeq", @@ -3389,7 +3717,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LCP(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -3418,7 +3762,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchLat", @@ -3464,7 +3824,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;Frontend;TmaL2", @@ -3510,12 +3878,24 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" - }, - "ResolutionLevels": "CORE, SOCKET, SYSTEM", - "MetricGroup": "DSBmiss;FetchBW", - "LocateWith": " FRONTEND_RETIRED.ANY_DSB_MISS" - }, + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" + }, + "ResolutionLevels": "CORE, SOCKET, SYSTEM", + "MetricGroup": "DSBmiss;FetchBW", + "LocateWith": " FRONTEND_RETIRED.ANY_DSB_MISS" + }, { "MetricName": "Decoder0_Alone", "LegacyName": "metric_TMA_......Decoder0_Alone(%)", @@ -3556,7 +3936,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Decoder0_Alone(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -3602,7 +3998,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;FetchBW", @@ -3655,7 +4063,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", @@ -3717,7 +4133,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2", @@ -3783,7 +4211,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;BrMispredicts", @@ -3845,7 +4289,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueMC, $issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2", @@ -3911,7 +4367,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Nukes(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;Machine_Clears", @@ -3964,7 +4436,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;TmaL1", @@ -4042,7 +4522,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20", + "BaseFormula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2", @@ -4075,7 +4567,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1, $issueMC" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -4116,7 +4624,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -4157,7 +4685,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4186,7 +4738,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4227,7 +4803,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4268,7 +4872,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4309,7 +4941,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4338,7 +4998,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Fwd_Blk(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4388,7 +5068,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L1_Latency_Dependency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -4433,7 +5133,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Lock_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Offcore", @@ -4474,7 +5194,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......Split_Loads(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Loads(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......Split_Loads(%) > 30", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4503,7 +5231,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......4K_Aliasing(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......4K_Aliasing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......4K_Aliasing(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4544,7 +5292,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......FB_Full(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FB_Full(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......FB_Full(%) > 30", + "ThresholdIssues": "$issueBW, $issueSL, $issueSmSt" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW", @@ -4593,7 +5349,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem", @@ -4643,7 +5415,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L2_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryLat", @@ -4676,7 +5468,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -4738,7 +5546,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Contested_Accesses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -4800,7 +5628,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Data_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;Offcore;Snoop", @@ -4850,7 +5698,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L3_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat, ~overlap" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -4892,7 +5760,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......SQ_Full(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 30 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -4945,7 +5833,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -4974,7 +5878,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Bandwidth(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -5007,7 +5931,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat;Offcore", @@ -5057,7 +6001,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Local_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Local_MEM(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Local_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Server", @@ -5107,7 +6075,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Remote_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Remote_MEM(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Remote_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Server;Snoop", @@ -5161,7 +6153,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Remote_Cache(%) > 5 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Remote_Cache(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Remote_Cache(%) > 5 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Offcore;Server;Snoop", @@ -5190,7 +6206,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -5231,7 +6263,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO, $issueSL, ~overlap" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore", @@ -5285,7 +6337,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......False_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -5327,7 +6399,27 @@ "Category": "TMA", "CountDomain": "Core_Utilization", "Threshold": { - "Formula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5373,7 +6465,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -5419,7 +6531,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5461,7 +6597,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5515,7 +6675,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5569,7 +6757,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5623,7 +6839,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5701,7 +6945,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2;Compute", @@ -5730,7 +6986,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB", @@ -5759,7 +7031,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSO" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;PortsUtil", @@ -5788,7 +7076,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Slow_Pause(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5854,7 +7162,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -5883,7 +7207,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_0(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -5912,7 +7256,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_........Mixing_Vectors(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Mixing_Vectors(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Mixing_Vectors(%) > 5", + "ThresholdIssues": "$issueMV" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5962,7 +7314,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_1(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -6012,7 +7384,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_2(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -6054,7 +7446,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_3m(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 40 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;PortsUtil", @@ -6108,7 +7520,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........ALU_Op_Utilization(%) > 40" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........ALU_Op_Utilization(%)" + } + ], + "Formula": "a > 40", + "BaseFormula": "metric_TMA_........ALU_Op_Utilization(%) > 40", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6150,7 +7570,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_0(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_0(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_0(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -6192,7 +7620,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_1(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_1(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_1(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6234,7 +7670,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_5(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_5(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_5(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6276,7 +7720,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_6(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_6(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_6(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6330,7 +7782,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Load_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Load_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6372,7 +7832,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_2(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_2(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_2(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6414,7 +7882,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_3(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_3(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_3(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6456,7 +7932,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Store_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Store_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6498,7 +7982,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_4(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_4(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_4(%) > 60", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6540,7 +8032,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_7(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_7(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_7(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6581,7 +8081,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Retiring(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 70 | b > 10", + "BaseFormula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;TmaL1", @@ -6631,7 +8143,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -6689,7 +8209,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 20 & b > 60", + "BaseFormula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -6739,7 +8271,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......X87_Use(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -6768,7 +8316,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Scalar(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -6797,7 +8361,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -6830,7 +8410,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -6863,7 +8463,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -6896,7 +8516,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_512b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -6950,7 +8590,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Memory_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -7000,7 +8652,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Fused_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -7054,7 +8718,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Non_Fused_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -7128,7 +8804,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 30 & b > 60", + "BaseFormula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -7182,7 +8870,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Nop_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Pipeline", @@ -7232,7 +8936,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -7290,7 +9002,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Few_Uops_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -7340,7 +9064,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueMC, $issueMS" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -7386,7 +9122,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Assists(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO", @@ -7428,7 +9180,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........FP_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........FP_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -7486,7 +9246,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......CISC(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -8788,6 +10564,11 @@ "BaseFormula": " idq.dsb_uops / idq.dsb_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -8813,6 +10594,11 @@ "BaseFormula": " idq.mite_uops / idq.mite_cycles", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -8942,6 +10728,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -9258,7 +11049,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Bandwidth" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -9941,6 +11740,11 @@ "BaseFormula": " 1000 * offcore_requests.demand_rfo / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheMisses;Offcore", "LocateWith": "" @@ -11384,7 +13188,15 @@ "Category": "TMA", "CountDomain": "Seconds", "Threshold": { - "Formula": "metric_TMA_Info_System_Time < 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Time" + } + ], + "Formula": "a < 1", + "BaseFormula": "metric_TMA_Info_System_Time < 1", + "ThresholdIssues": "" }, "ResolutionLevels": "SOCKET, SYSTEM", "MetricGroup": "Summary", diff --git a/CLX/metrics/perf/cascadelakex_metrics_perf.json b/CLX/metrics/perf/cascadelakex_metrics_perf.json index 07d45c30..06d29878 100644 --- a/CLX/metrics/perf/cascadelakex_metrics_perf.json +++ b/CLX/metrics/perf/cascadelakex_metrics_perf.json @@ -1,125 +1,125 @@ [ { - "BriefDescription": "CPU operating frequency (in GHz).", - "MetricExpr": "(( CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000)", + "BriefDescription": "CPU operating frequency (in GHz)", + "MetricExpr": "((CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000)", "MetricGroup": "", "MetricName": "cpu_operating_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Percentage of time spent in the active CPU power state C0.", + "BriefDescription": "Percentage of time spent in the active CPU power state C0", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "", "MetricName": "cpu_utilization", "ScaleUnit": "100%" }, { - "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles.", + "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles", "MetricExpr": "CPU_CLK_UNHALTED.THREAD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions", "MetricExpr": "MEM_INST_RETIRED.ALL_LOADS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "loads_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions", "MetricExpr": "MEM_INST_RETIRED.ALL_STORES / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "stores_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L1D.REPLACEMENT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L1_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions", "MetricExpr": "L2_RQSTS.ALL_CODE_RD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1_i_code_read_misses_with_prefetches_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L2_LINES_IN.ALL / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions", "MetricExpr": "L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_code_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", "MetricExpr": "cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x12d40433@ / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_data_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", "MetricExpr": "cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x12cc0233@ / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_code_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) in nano seconds.", - "MetricExpr": "( 1000000000 * ( cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS\\,config1\\=0x40433@ / cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40433@ ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) in nano seconds", + "MetricExpr": "( 1000000000 * (cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS\\,config1\\=0x40433@ / cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40433@) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_data_read_demand_plus_prefetch_miss_latency", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to local memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS\\,config1\\=0x40432@ / cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@ ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to local memory in nano seconds", + "MetricExpr": "( 1000000000 * (cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS\\,config1\\=0x40432@ / cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_data_read_demand_plus_prefetch_miss_latency_for_local_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to remote memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS\\,config1\\=0x40431@ / cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@ ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to remote memory in nano seconds", + "MetricExpr": "( 1000000000 * (cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS\\,config1\\=0x40431@ / cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_data_read_demand_plus_prefetch_miss_latency_for_remote_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_mpi", @@ -127,7 +127,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the ITLB (Instruction TLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED_2M_4M / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_large_page_mpi", @@ -135,7 +135,7 @@ "PublicDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the Instruction Translation Lookaside Buffer (ITLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions", "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_load_mpi", @@ -143,7 +143,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions", "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_2mb_large_page_load_mpi", @@ -151,7 +151,7 @@ "PublicDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the Data Translation Lookaside Buffer (DTLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions", "MetricExpr": "DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_store_mpi", @@ -159,1163 +159,1285 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@ / ( cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@ + cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@ )", + "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@ / (cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@ + cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_local_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@ / ( cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@ + cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@ )", + "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@ / (cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@ + cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_remote_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Uncore operating frequency in GHz.", - "MetricExpr": "( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) / 1000000000) / duration_time", + "BriefDescription": "Uncore operating frequency in GHz", + "MetricExpr": "(UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) / 1000000000) / duration_time", "MetricGroup": "", "MetricName": "uncore_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data transmit bandwidth (MB/sec).", - "MetricExpr": "( UNC_UPI_TxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", + "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data transmit bandwidth (MB/sec)", + "MetricExpr": "(UNC_UPI_TxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", "MetricGroup": "", "MetricName": "upi_data_transmit_bw", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory read bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory read bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory write bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory write bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory bandwidth (MB/sec)", + "MetricExpr": "((UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_total", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory read bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_PMM_RPQ_INSERTS * 64 / 1000000) / duration_time", + "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory read bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_PMM_RPQ_INSERTS * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "pmem_memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory write bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_PMM_WPQ_INSERTS * 64 / 1000000) / duration_time", + "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory write bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_PMM_WPQ_INSERTS * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "pmem_memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_PMM_RPQ_INSERTS + UNC_M_PMM_WPQ_INSERTS ) * 64 / 1000000) / duration_time", + "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory bandwidth (MB/sec)", + "MetricExpr": "((UNC_M_PMM_RPQ_INSERTS + UNC_M_PMM_WPQ_INSERTS) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "pmem_memory_bandwidth_total", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU.", - "MetricExpr": "(( UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART0 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART1 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART2 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART3 ) * 4 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU", + "MetricExpr": "((UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART0 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART1 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART2 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART3) * 4 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU.", - "MetricExpr": "(( UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART0 + UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART1 + UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART2 + UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART3 ) * 4 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU", + "MetricExpr": "((UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART0 + UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART1 + UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART2 + UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART3) * 4 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.DSB_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.DSB_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_decoded_icache", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MITE_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MITE_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_legacy_decode_pipeline", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MS_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MS_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_microcode_sequencer", "ScaleUnit": "100%" }, { - "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to local memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.READS_LOCAL * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to local memory", + "MetricExpr": "(UNC_CHA_REQUESTS.READS_LOCAL * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_local_memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to local memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.WRITES_LOCAL * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to local memory", + "MetricExpr": "(UNC_CHA_REQUESTS.WRITES_LOCAL * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_local_memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to remote memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.READS_REMOTE * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to remote memory", + "MetricExpr": "(UNC_CHA_REQUESTS.READS_REMOTE * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_remote_memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to remote memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.WRITES_REMOTE * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to remote memory", + "MetricExpr": "(UNC_CHA_REQUESTS.WRITES_REMOTE * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_remote_memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data receive bandwidth (MB/sec).", - "MetricExpr": "( UNC_UPI_RxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", + "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data receive bandwidth (MB/sec)", + "MetricExpr": "(UNC_UPI_RxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", "MetricGroup": "", "MetricName": "upi_data_receive_bw", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", + "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_mispredictions" + "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueBM", + "MetricName": "tma_bottleneck_mispredictions", + "MetricThreshold": "tma_bottleneck_mispredictions > 20", + "PublicDescription": "Total pipeline cost of Branch Misprediction related bottlenecks. Related metrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers." }, { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", + "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_big_code" + "MetricName": "tma_bottleneck_big_code", + "MetricThreshold": "tma_bottleneck_big_code > 20" }, { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", + "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) ) ) - tma_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_instruction_fetch_bw" + "MetricName": "tma_bottleneck_instruction_fetch_bw", + "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) )", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_cache_memory_bandwidth" + "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueBW", + "MetricName": "tma_bottleneck_cache_memory_bandwidth", + "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_cache_memory_latency" + "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueLat", + "MetricName": "tma_bottleneck_cache_memory_latency", + "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks. Related metrics: tma_l3_hit_latency, tma_mem_latency." }, { - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", + "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs)", "MetricExpr": "100 * ( tma_memory_bound * ( tma_l1_bound / max( tma_memory_bound , ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) ) * ( tma_dtlb_load / max( tma_l1_bound , ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_memory_data_tlbs" + "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueTLB", + "MetricName": "tma_bottleneck_memory_data_tlbs", + "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", + "PublicDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs). Related metrics: tma_dtlb_load, tma_dtlb_store." }, { - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", + "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors)", "MetricExpr": "100 * ( tma_memory_bound * ( ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) * tma_remote_cache / ( tma_local_mem + tma_remote_mem + tma_remote_cache ) + ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", - "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_memory_synchronization" + "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueSyncxn", + "MetricName": "tma_bottleneck_memory_synchronization", + "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", + "PublicDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", + "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation", "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Scaled_Slots", + "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", + "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", "PublicDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy." }, { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", + "BriefDescription": "Total pipeline cost of irregular execution (e.g", "MetricExpr": "100 * ( ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + tma_core_bound * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Scaled_Slots", + "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", - "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments)." + "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", + "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments). Related metrics: tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end.", + "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end", "MetricExpr": "100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", "MetricGroup": "BvOB;Cor;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", "MetricName": "tma_bottleneck_other_bottlenecks", + "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", "PublicDescription": "Total pipeline cost of remaining bottlenecks in the back-end. Examples include data-dependencies (Core Bound when Low ILP) and other unlisted memory-related stalls." }, { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", + "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA", "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Scaled_Slots", "MetricName": "tma_bottleneck_branching_overhead", + "MetricThreshold": "tma_bottleneck_branching_overhead > 5", "PublicDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound)." }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", + "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead", "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_useful_work" + "MetricName": "tma_bottleneck_useful_work", + "MetricThreshold": "tma_bottleneck_useful_work > 20" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "IDQ_UOPS_NOT_DELIVERED.CORE / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_frontend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_frontend_bound > 0.15", "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound. Sample with: FRONTEND_RETIRED.LATENCY_GE_4." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( 4 ) * IDQ_UOPS_NOT_DELIVERED.CYCLES_0_UOPS_DELIV.CORE / tma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", "MetricName": "tma_fetch_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_16, FRONTEND_RETIRED.LATENCY_GE_8." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "( ICACHE_16B.IFDATA_STALL + 2 * cpu@ICACHE_16B.IFDATA_STALL\\,cmask\\=0x1\\,edge\\=0x1@ ) / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_itlb_misses - tma_code_stlb_miss )", - "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group", + "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", - "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group", + "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + ( 9 ) * BACLEARS.ANY / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "( 1 - ( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "( 9 ) * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: BACLEARS.ANY." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 2 ) * IDQ.MS_SWITCHES / tma_info_thread_clks", - "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated", + "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES." + "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "DECODE.LCP / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." + "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", - "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS." + "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "tma_frontend_bound - tma_fetch_latency", - "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", + "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2." + "MetricThreshold": "tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.ALL_MITE_CYCLES_ANY_UOPS - IDQ.ALL_MITE_CYCLES_4_UOPS ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", "ScaleUnit": "100%", + "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder.", + "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder", "MetricExpr": "( cpu@INST_DECODED.DECODERS\\,cmask\\=0x1@ - cpu@INST_DECODED.DECODERS\\,cmask\\=0x2@ ) / tma_info_core_core_clks / 2", - "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated", + "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated;tma_issueD0", "MetricName": "tma_decoder0_alone", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder. Related metrics: tma_few_uops_instructions." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.DSB_CYCLES_ANY - IDQ.DSB_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", "ScaleUnit": "100%", + "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "( UOPS_ISSUED.ANY - ( UOPS_RETIRED.RETIRE_SLOTS ) + ( 4 ) * ( ( INT_MISC.RECOVERY_CYCLES_ANY / 2 ) if #SMT_on else INT_MISC.RECOVERY_CYCLES ) ) / tma_info_thread_slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_bad_speculation", "ScaleUnit": "100%", + "MetricThreshold": "tma_bad_speculation > 0.15", "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example." }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * tma_bad_speculation", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", + "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types)", "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Slots", "MetricName": "tma_other_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "tma_bad_speculation - tma_branch_mispredicts", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", + "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT." + "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches, tma_remote_cache." }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering", "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_machine_clears_group;Slots", "MetricName": "tma_other_nukes", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "1 - tma_frontend_bound - ( UOPS_ISSUED.ANY + ( 4 ) * ( ( INT_MISC.RECOVERY_CYCLES_ANY / 2 ) if #SMT_on else INT_MISC.RECOVERY_CYCLES ) ) / tma_info_thread_slots", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_backend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_backend_bound > 0.2", "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound." }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( ( CYCLE_ACTIVITY.STALLS_MEM_ANY + EXE_ACTIVITY.BOUND_ON_STORES ) / ( CYCLE_ACTIVITY.STALLS_TOTAL + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) + EXE_ACTIVITY.BOUND_ON_STORES ) ) * tma_backend_bound", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_memory_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two)." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", + "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT." + "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "min( ( 9 ) * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS." + "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", "MetricName": "tma_load_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", "MetricName": "tma_load_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", + "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache", "MetricExpr": "min( 2 * ( MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETIRED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS ) * 20 / 100 , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_l1_latency_dependency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( 12 * max( 0 , MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS.ALL_RFO ) + ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) * ( ( 11 ) * L2_RQSTS.RFO_HIT + ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) ) / tma_info_thread_clks", - "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks", + "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS." + "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset.", + "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset", "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_4k_aliasing", "ScaleUnit": "100%", + "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset. False match is possible; which incur a few cycles load re-issue. However; the short re-issue duration is often hidden by the out-of-order core and HW optimizations; hence a user may safely ignore a high value of this metric unless it manages to propagate up into parent nodes of the hierarchy (e.g. to L1_Bound)." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PEND_MISS.FB_FULL\\,cmask\\=0x1@ / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", + "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory)." + "MetricThreshold": "tma_fb_full > 0.3", + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( ( MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) ) ) / ( ( MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) ) ) + cpu@L1D_PEND_MISS.FB_FULL\\,cmask\\=0x1@ ) ) * ( ( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks )", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", + "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited)", "MetricExpr": "( 3.5 * tma_info_system_core_frequency ) * MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", "MetricName": "tma_l2_hit_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L2_MISS - CYCLE_ACTIVITY.STALLS_L3_MISS ) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( ( 47.5 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * ( OCR.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE / ( OCR.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE + OCR.DEMAND_DATA_RD.L3_HIT.HIT_OTHER_CORE_FWD ) ) ) + ( ( 47.5 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS." + "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( ( 47.5 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * ( 1 - ( OCR.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE / ( OCR.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE + OCR.DEMAND_DATA_RD.L3_HIT.HIT_OTHER_CORE_FWD ) ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT." + "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( ( 20.5 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_RETIRED.L3_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT." + "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_bottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "( ( OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 ) if #SMT_on else OFFCORE_REQUESTS_BUFFER.SQ_FULL ) / tma_info_core_core_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clks + ( ( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks ) - tma_l2_bound )", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that)." + "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that)." + "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory", "MetricExpr": "( ( 80 * tma_info_system_core_frequency ) - ( 20.5 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_local_mem", "ScaleUnit": "100%", + "MetricThreshold": "tma_local_mem > 0.1 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory. Caching will improve the latency and increase performance. Sample with: MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory", "MetricExpr": "( ( 147.5 * tma_info_system_core_frequency ) - ( 20.5 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_remote_mem", "ScaleUnit": "100%", + "MetricThreshold": "tma_remote_mem > 0.1 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues", "MetricExpr": "( ( ( 110 * tma_info_system_core_frequency ) - ( 20.5 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM + ( ( 110 * tma_info_system_core_frequency ) - ( 20.5 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", + "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_remote_cache", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD." + "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( L2_RQSTS.RFO_HIT * ( 11 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)." + "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( ( 110 * tma_info_system_core_frequency ) * ( OCR.DEMAND_RFO.L3_MISS.REMOTE_HITM + OCR.PF_L2_RFO.L3_MISS.REMOTE_HITM ) + ( 47.5 * tma_info_system_core_frequency ) * ( OCR.DEMAND_RFO.L3_HIT.HITM_OTHER_CORE + OCR.PF_L2_RFO.L3_HIT.HITM_OTHER_CORE ) ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, OCR.DEMAND_RFO.L3_HIT.HITM_OTHER_CORE." + "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, OCR.DEMAND_RFO.L3_HIT.HITM_OTHER_CORE. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization", + "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", "ScaleUnit": "100%", - "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." + "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES. Related metrics: tma_port_4." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( ( 9 ) * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_STORE_MISSES.WALK_ACTIVE ) / tma_info_core_core_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES." + "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_load." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", "MetricName": "tma_store_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", "MetricName": "tma_store_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "tma_backend_bound - tma_memory_bound", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_core_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations)." }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", "ScaleUnit": "100%", + "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIVIDER_ACTIVE." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations", "MetricExpr": "PARTIAL_RAT_STALLS.SCOREBOARD / tma_info_thread_clks", - "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Clocks;tma_issueSO", "MetricName": "tma_serializing_operation", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: PARTIAL_RAT_STALLS.SCOREBOARD." + "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: PARTIAL_RAT_STALLS.SCOREBOARD. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions", "MetricExpr": "40 * ROB_MISC_EVENTS.PAUSE_INST / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_slow_pause", "ScaleUnit": "100%", + "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions. Sample with: ROB_MISC_EVENTS.PAUSE_INST." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( tma_ports_utilized_0 * tma_info_thread_clks + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIVIDER_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - CYCLE_ACTIVITY.STALLS_MEM_ANY ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.EXE_BOUND_0_PORTS / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", + "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles)", "MetricExpr": "UOPS_ISSUED.VECTOR_WIDTH_MISMATCH / UOPS_ISSUED.ANY", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks;tma_issueMV", "MetricName": "tma_mixing_vectors", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic." + "MetricThreshold": "tma_mixing_vectors > 0.05", + "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 - UOPS_EXECUTED.CORE_CYCLES_GE_2 ) / 2 if #SMT_on else EXE_ACTIVITY.1_PORTS_UTIL ) / tma_info_core_core_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful." + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( UOPS_EXECUTED.CORE_CYCLES_GE_2 - UOPS_EXECUTED.CORE_CYCLES_GE_3 ) / 2 if #SMT_on else EXE_ACTIVITY.2_PORTS_UTIL ) / tma_info_core_core_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop." + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( UOPS_EXECUTED.CORE_CYCLES_GE_3 / 2 if #SMT_on else UOPS_EXECUTED.CORE_CYCLES_GE_3 ) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT.PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_0 / tma_info_core_core_clks", - "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0." + "MetricThreshold": "tma_port_0 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_1 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1." + "MetricThreshold": "tma_port_1 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_5 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_5", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5." + "MetricThreshold": "tma_port_5 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_6 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1." + "MetricThreshold": "tma_port_6 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "( UOPS_DISPATCHED_PORT.PORT_2 + UOPS_DISPATCHED_PORT.PORT_3 + UOPS_DISPATCHED_PORT.PORT_7 - UOPS_DISPATCHED_PORT.PORT_4 ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_op_utilization > 0.6" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_2 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_group;Core_Clocks", "MetricName": "tma_port_2", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_2 > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads). Sample with: UOPS_DISPATCHED_PORT.PORT_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_3 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_group;Core_Clocks", "MetricName": "tma_port_3", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_3 > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads). Sample with: UOPS_DISPATCHED_PORT.PORT_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_op_utilization > 0.6" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks;tma_issueSpSt", "MetricName": "tma_port_4", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data). Sample with: UOPS_DISPATCHED_PORT.PORT_4." + "MetricThreshold": "tma_port_4 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data). Sample with: UOPS_DISPATCHED_PORT.PORT_4. Related metrics: tma_split_stores." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_7 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks", "MetricName": "tma_port_7", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_7 > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address). Sample with: UOPS_DISPATCHED_PORT.PORT_7." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / tma_info_thread_slots", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_retiring", "ScaleUnit": "100%", + "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.1", "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.RETIRE_SLOTS." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_light_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", "ScaleUnit": "100%", + "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( cpu@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE\\,umask\\=0xFC@ ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_512b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses", "MetricExpr": "tma_light_operations * MEM_INST_RETIRED.ANY / INST_RETIRED.ANY", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_memory_operations", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_operations > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions", "MetricExpr": "tma_light_operations * UOPS_RETIRED.MACRO_FUSED / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_fused_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Other_Light_Ops in MTL!)}." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused", "MetricExpr": "tma_light_operations * ( BR_INST_RETIRED.ALL_BRANCHES - UOPS_RETIRED.MACRO_FUSED ) / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_non_fused_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused. Non-conditional branches like direct JMP or CALL would count here. Can be used to examine fusible conditional jumps that were not fused." }, { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", + "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes", "MetricExpr": "max( 0 , tma_light_operations - ( tma_fp_arith + tma_memory_operations + tma_fused_instructions + tma_non_fused_branches ) )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_other_light_ops", "ScaleUnit": "100%", + "MetricThreshold": "tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_light_operations_group;Slots", "MetricName": "tma_nop_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "( ( UOPS_RETIRED.RETIRE_SLOTS ) + UOPS_RETIRED.MACRO_FUSED - INST_RETIRED.ANY ) / tma_info_thread_slots", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_heavy_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+])." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops", "MetricExpr": "tma_heavy_operations - tma_microcode_sequencer", - "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueD0", "MetricName": "tma_few_uops_instructions", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions." + "MetricThreshold": "tma_few_uops_instructions > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions. Related metrics: tma_decoder0_alone." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "( ( UOPS_RETIRED.RETIRE_SLOTS ) / UOPS_ISSUED.ANY ) * IDQ.MS_UOPS / tma_info_thread_slots", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS." + "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( 34 ) * ( FP_ASSIST.ANY + OTHER_ASSISTS.ANY ) / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists", "MetricExpr": "34 * FP_ASSIST.ANY / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_fp_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_assists > 0.1", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", "ScaleUnit": "100%", + "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources." }, { - "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts.", + "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts", "MetricExpr": "100 * ( 1 - tma_core_bound / tma_ports_utilization if tma_core_bound < tma_ports_utilization else 1 ) if tma_info_system_smt_2t_utilization > 0.5 else 0", "MetricGroup": "Cor;SMT;Metric", "MetricName": "tma_info_botlnk_l0_core_bound_likely", "MetricThreshold": "tma_info_botlnk_core_bound_likely > 0.5" }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", "MetricName": "tma_info_thread_uoppi", "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", "MetricName": "tma_info_thread_uptb", "MetricThreshold": "tma_info_thread_uptb < 4 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "( 4 ) * tma_info_core_core_clks", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / tma_info_core_core_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( cpu@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE\\,umask\\=0xFC@ ) ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "uops Executed per Cycle.", + "BriefDescription": "uops Executed per Cycle", "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", "MetricGroup": "Power;Metric", "MetricName": "tma_info_core_epc" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "( CPU_CLK_UNHALTED.THREAD_ANY / 2 ) if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipload", "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipstore", "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipbranch", "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", "MetricName": "tma_info_inst_mix_ipcall", "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", "MetricThreshold": "tma_info_inst_mix_iptb < 4 * 2 + 1", - "PublicDescription": "Instructions per taken branch. Related metrics: tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage." + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipflop", "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( cpu@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE\\,umask\\=0xFC@ ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", @@ -1323,7 +1445,7 @@ "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", @@ -1331,7 +1453,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", @@ -1339,7 +1461,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", @@ -1347,7 +1469,7 @@ "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", @@ -1355,7 +1477,7 @@ "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx512", @@ -1363,33 +1485,33 @@ "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / ROB_MISC_EVENTS.PAUSE_INST", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ippause" }, { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / SW_PREFETCH_ACCESS.ANY", "MetricGroup": "Prefetches;Inst_Metric", "MetricName": "tma_info_inst_mix_ipswpf", "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / cpu@UOPS_RETIRED.RETIRE_SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Instructions per a microcode Assist invocation.", + "BriefDescription": "Instructions per a microcode Assist invocation", "MetricExpr": "INST_RETIRED.ANY / ( FP_ASSIST.ANY + OTHER_ASSISTS.ANY )", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", "MetricName": "tma_info_pipeline_ipassist", @@ -1397,96 +1519,98 @@ "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core.", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core", "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@ )", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", + "BriefDescription": "Average number of uops fetched from DSB per cycle", "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_dsb" }, { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", + "BriefDescription": "Average number of uops fetched from MITE per cycle", "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_mite" }, { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", + "BriefDescription": "Average number of Uops issued by front-end when it issued something", "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_frontend_fetch_upc" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS )", "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", "MetricName": "tma_info_frontend_dsb_coverage", "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 4 > 0.35", - "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb." + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", + "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / DSB2MITE_SWITCHES.COUNT", "MetricGroup": "DSBmiss;Metric", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Average Latency for L1 instruction cache misses.", + "BriefDescription": "Average Latency for L1 instruction cache misses", "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STALL\\,cmask\\=0x1\\,edge\\=0x1@ + 2", "MetricGroup": "Fed;FetchLat;IcMiss;Metric", "MetricName": "tma_info_frontend_icache_miss_latency" }, { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", "MetricGroup": "DSBmiss;Fed;Inst_Metric", "MetricName": "tma_info_frontend_ipdsb_miss_ret", "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache true code cacheline misses per kilo instruction", "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code" }, { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction", "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code_all" }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb ) )", "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_dsb_misses > 10", - "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_info_inst_mix_iptb, tma_info_frontend_dsb_coverage." + "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb ) ) )", - "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_bandwidth" + "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", + "MetricThreshold": "tma_info_botlnk_dsb_bandwidth > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", + "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", @@ -1494,284 +1618,284 @@ "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmispredict", "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / ( ( ( UOPS_RETIRED.RETIRE_SLOTS ) / UOPS_ISSUED.ANY ) * BR_MISP_EXEC.INDIRECT )", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_indirect", "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 4 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", "MetricName": "tma_info_bad_spec_branch_misprediction_cost", - "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)." + "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_mispredicts_resteers." }, { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", + "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", "MetricGroup": "BrMispredicts;Metric", "MetricName": "tma_info_bad_spec_spec_clears_ratio" }, { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", + "BriefDescription": "Fraction of branches that are non-taken conditionals", "MetricExpr": "BR_INST_RETIRED.NOT_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_nt" }, { - "BriefDescription": "Fraction of branches that are taken conditionals.", + "BriefDescription": "Fraction of branches that are taken conditionals", "MetricExpr": "( BR_INST_RETIRED.CONDITIONAL - BR_INST_RETIRED.NOT_TAKEN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_tk" }, { - "BriefDescription": "Fraction of branches that are CALL or RET.", + "BriefDescription": "Fraction of branches that are CALL or RET", "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_callret" }, { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", + "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - ( BR_INST_RETIRED.COND - BR_INST_RETIRED.NOT_TAKEN ) - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_jump" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_RETIRED.L1_MISS + MEM_LOAD_RETIRED.FB_HIT )", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki_load" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * OFFCORE_REQUESTS.DEMAND_RFO / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_all" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", + "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_fb_hpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Offcore;Metric", "MetricName": "tma_info_memory_l3_cache_access_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING + EPT.WALK_PENDING ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", "MetricName": "tma_info_memory_tlb_page_walks_utilization", "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Fed;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_code_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_load_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_access_bw", "MetricGroup": "Mem;MemoryBW;Offcore;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_access_bw_2t" }, { - "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory).", + "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory)", "MetricExpr": "1000 * L2_LINES_OUT.SILENT / tma_info_inst_mix_instructions", "MetricGroup": "L2Evicts;Mem;Server;Core_Metric", "MetricName": "tma_info_memory_core_l2_evictions_silent_pki" }, { - "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction.", + "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction", "MetricExpr": "1000 * L2_LINES_OUT.NON_SILENT / tma_info_inst_mix_instructions", "MetricGroup": "L2Evicts;Mem;Server;Core_Metric", "MetricName": "tma_info_memory_core_l2_evictions_nonsilent_pki" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", + "BriefDescription": "Un-cacheable retired load per kilo instruction", "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_uc_load_pki" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz].", + "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz]", "MetricExpr": "tma_info_system_socket_clks / 1e9 / tma_info_system_time", "MetricGroup": "SoC;System_Metric", "MetricName": "tma_info_system_uncore_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0.", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0", "MetricExpr": "CORE_POWER.LVL0_TURBO_LICENSE / 2 / tma_info_core_core_clks if #SMT_on else CORE_POWER.LVL0_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license0_utilization", "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0. This includes non-AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes." }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1.", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1", "MetricExpr": "CORE_POWER.LVL1_TURBO_LICENSE / 2 / tma_info_core_core_clks if #SMT_on else CORE_POWER.LVL1_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license1_utilization", @@ -1779,7 +1903,7 @@ "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1. This includes high current AVX 256-bit instructions as well as low current AVX 512-bit instructions." }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX).", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX)", "MetricExpr": "CORE_POWER.LVL2_TURBO_LICENSE / 2 / tma_info_core_core_clks if #SMT_on else CORE_POWER.LVL2_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license2_utilization", @@ -1787,94 +1911,95 @@ "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX). This includes high current AVX 512-bit instructions." }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / ( CPU_CLK_UNHALTED.REF_XCLK_ANY / 2 ) if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_utilization", "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "( 64 * ( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", "MetricName": "tma_info_system_dram_bw_use", - "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]." + "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Average latency of data read request to external memory (in nanoseconds).", + "BriefDescription": "Average latency of data read request to external memory (in nanoseconds)", "MetricExpr": "( 1000000000 ) * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( tma_info_system_socket_clks / tma_info_system_time )", "MetricGroup": "Mem;MemoryLat;SoC;NanoSeconds", "MetricName": "tma_info_system_mem_read_latency", "PublicDescription": "Average latency of data read request to external memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetches. ([RKL+]memory-controller only)." }, { - "BriefDescription": "Average number of parallel data read requests to external memory.", + "BriefDescription": "Average number of parallel data read requests to external memory", "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD\\,thresh\\=0x1@", "MetricGroup": "Mem;MemoryBW;SoC;System_Metric", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches." }, { - "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds].", + "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]", "MetricExpr": "( 1000000000 ) * ( UNC_M_RPQ_OCCUPANCY / UNC_M_RPQ_INSERTS ) / imc_0@event\\=0x0@", "MetricGroup": "MemOffcore;MemoryLat;SoC;Server;NanoSeconds", "MetricName": "tma_info_system_mem_dram_read_latency", "PublicDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]. Accounts for demand loads and L1/L2 data-read prefetches." }, { - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec].", + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]", "MetricExpr": "( UNC_IIO_DATA_REQ_OF_CPU.MEM_WRITE.PART0 + UNC_IIO_DATA_REQ_OF_CPU.MEM_WRITE.PART1 + UNC_IIO_DATA_REQ_OF_CPU.MEM_WRITE.PART2 + UNC_IIO_DATA_REQ_OF_CPU.MEM_WRITE.PART3 ) * 4 / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "IoBW;MemOffcore;SoC;Server;GB/sec", "MetricName": "tma_info_system_io_read_bw", "PublicDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]. Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU." }, { - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec].", + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]", "MetricExpr": "( UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART0 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART1 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART2 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART3 ) * 4 / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "IoBW;MemOffcore;SoC;Server;GB/sec", "MetricName": "tma_info_system_io_write_bw", "PublicDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]. Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU." }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "( power@energy\\-pkg@ * ( 61 ) + 15.6 * power@energy\\-ram@ ) / ( ( duration_time ) * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", "MetricName": "tma_info_system_mux", "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "cha_0@event\\=0x0@", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", "MetricName": "tma_info_system_ipfarbranch", diff --git a/EMR/metrics/emeraldrapids_metrics.json b/EMR/metrics/emeraldrapids_metrics.json index eea13d02..e66196ab 100644 --- a/EMR/metrics/emeraldrapids_metrics.json +++ b/EMR/metrics/emeraldrapids_metrics.json @@ -2,7 +2,7 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for 5th Generation Intel(R) Xeon(R) Processor Scalable Family0", - "DatePublished": "11/06/2024", + "DatePublished": "11/15/2024", "Version": "1.0", "Legend": "", "TmaVersion": "5.01", @@ -1112,7 +1112,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Mispredictions > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Mispredictions" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Mispredictions > 20", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP", @@ -1200,7 +1208,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Big_Code > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Big_Code" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Big_Code > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB", @@ -1349,7 +1365,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Instruction_Fetch_BW" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", @@ -1538,7 +1562,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore", @@ -1628,21 +1660,25 @@ "Alias": "a_s" }, { - "Name": "OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM", + "Name": "OCR.DEMAND_RFO.L3_MISS:ocr_msr_val=0x103b800002", "Alias": "a_t" }, { - "Name": "OCR.STREAMING_WR.ANY_RESPONSE", + "Name": "OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM", "Alias": "a_u" }, { - "Name": "DTLB_STORE_MISSES.STLB_HIT:c1", + "Name": "OCR.STREAMING_WR.ANY_RESPONSE", "Alias": "a_v" }, { - "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", + "Name": "DTLB_STORE_MISSES.STLB_HIT:c1", "Alias": "a_w" }, + { + "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", + "Alias": "a_x" + }, { "Name": "PERF_METRICS.FRONTEND_BOUND", "Alias": "b" @@ -1758,12 +1794,20 @@ "Alias": "threads" } ], - "Formula": "100 * ( ( ( a / ( b + c + d + e ) ) * ( ( ( f / ( g ) ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( min( g , l ) ) / ( g ) - ( ( min( g , m ) ) / ( g ) ) ) / ( ( ( min( g , m ) ) / ( g ) ) + ( ( min( g , l ) ) / ( g ) - ( ( min( g , m ) ) / ( g ) ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( ( j - f ) / ( g ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( q * ( 1 + ( r / s ) / 2 ) ) / ( g ) ) / ( ( ( ( ( 81 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( t * ( u / ( u + v ) ) ) + ( ( 79 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( w ) ) * ( 1 + ( r / s ) / 2 ) / ( g ) ) + ( ( ( 79 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( x + t * ( 1 - ( u / ( u + v ) ) ) ) * ( 1 + ( r / s ) / 2 ) / ( g ) ) + ( ( ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( q * ( 1 + ( r / s ) / 2 ) ) / ( g ) ) + ( ( y + z ) / ( g ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( i - j ) / ( g ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( max( ( h - i ) / ( g ) , 0 ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( min( 2 * ( a_a - r - s ) * dependentloadsweight / 100 , max( a_c - a_d , 0 ) ) / ( g ) ) / ( ( min( ( 7 ) * a_e + a_f , max( a_c - a_d , 0 ) ) / ( g ) ) + ( 13 * a_g / ( g ) ) + ( min( 2 * ( a_a - r - s ) * dependentloadsweight / 100 , max( a_c - a_d , 0 ) ) / ( g ) ) + ( ( 16 * max( 0 , a_h - a_i ) + ( a_h / a_j ) * ( ( 10 ) * a_k + ( min( g , a_l ) ) ) ) / ( g ) ) + ( ( a_m / a_n ) * a_o / ( g ) ) + ( a_p / ( g ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( max( ( h - i ) / ( g ) , 0 ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( 16 * max( 0 , a_h - a_i ) + ( a_h / a_j ) * ( ( 10 ) * a_k + ( min( g , a_l ) ) ) ) / ( g ) ) / ( ( min( ( 7 ) * a_e + a_f , max( a_c - a_d , 0 ) ) / ( g ) ) + ( 13 * a_g / ( g ) ) + ( min( 2 * ( a_a - r - s ) * dependentloadsweight / 100 , max( a_c - a_d , 0 ) ) / ( g ) ) + ( ( 16 * max( 0 , a_h - a_i ) + ( a_h / a_j ) * ( ( 10 ) * a_k + ( min( g , a_l ) ) ) ) / ( g ) ) + ( ( a_m / a_n ) * a_o / ( g ) ) + ( a_p / ( g ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( max( ( h - i ) / ( g ) , 0 ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( a_m / a_n ) * a_o / ( g ) ) / ( ( min( ( 7 ) * a_e + a_f , max( a_c - a_d , 0 ) ) / ( g ) ) + ( 13 * a_g / ( g ) ) + ( min( 2 * ( a_a - r - s ) * dependentloadsweight / 100 , max( a_c - a_d , 0 ) ) / ( g ) ) + ( ( 16 * max( 0 , a_h - a_i ) + ( a_h / a_j ) * ( ( 10 ) * a_k + ( min( g , a_l ) ) ) ) / ( g ) ) + ( ( a_m / a_n ) * a_o / ( g ) ) + ( a_p / ( g ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( k / ( g ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( a_q / ( a_r if smt_on else ( g ) ) ) / ( ( ( ( a_s * ( 10 ) * ( 1 - ( a_h / a_j ) ) ) + ( 1 - ( a_h / a_j ) ) * ( min( g , a_l ) ) ) / ( g ) ) + ( ( 81 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_t / ( g ) ) + ( a_q / ( a_r if smt_on else ( g ) ) ) + ( 9 * a_u / ( g ) ) + ( ( ( 7 ) * a_v + a_w ) / ( a_r if smt_on else ( g ) ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( k / ( g ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( ( a_s * ( 10 ) * ( 1 - ( a_h / a_j ) ) ) + ( 1 - ( a_h / a_j ) ) * ( min( g , a_l ) ) ) / ( g ) ) / ( ( ( ( a_s * ( 10 ) * ( 1 - ( a_h / a_j ) ) ) + ( 1 - ( a_h / a_j ) ) * ( min( g , a_l ) ) ) / ( g ) ) + ( ( 81 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_t / ( g ) ) + ( a_q / ( a_r if smt_on else ( g ) ) ) + ( 9 * a_u / ( g ) ) + ( ( ( 7 ) * a_v + a_w ) / ( a_r if smt_on else ( g ) ) ) ) ) ) )", + "Formula": "100 * ( ( ( a / ( b + c + d + e ) ) * ( ( ( f / ( g ) ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( min( g , l ) ) / ( g ) - ( ( min( g , m ) ) / ( g ) ) ) / ( ( ( min( g , m ) ) / ( g ) ) + ( ( min( g , l ) ) / ( g ) - ( ( min( g , m ) ) / ( g ) ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( ( j - f ) / ( g ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( q * ( 1 + ( r / s ) / 2 ) ) / ( g ) ) / ( ( ( ( ( 81 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( t * ( u / ( u + v ) ) ) + ( ( 79 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( w ) ) * ( 1 + ( r / s ) / 2 ) / ( g ) ) + ( ( ( 79 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( x + t * ( 1 - ( u / ( u + v ) ) ) ) * ( 1 + ( r / s ) / 2 ) / ( g ) ) + ( ( ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( q * ( 1 + ( r / s ) / 2 ) ) / ( g ) ) + ( ( y + z ) / ( g ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( i - j ) / ( g ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( max( ( h - i ) / ( g ) , 0 ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( min( 2 * ( a_a - r - s ) * dependentloadsweight / 100 , max( a_c - a_d , 0 ) ) / ( g ) ) / ( ( min( ( 7 ) * a_e + a_f , max( a_c - a_d , 0 ) ) / ( g ) ) + ( 13 * a_g / ( g ) ) + ( min( 2 * ( a_a - r - s ) * dependentloadsweight / 100 , max( a_c - a_d , 0 ) ) / ( g ) ) + ( ( 16 * max( 0 , a_h - a_i ) + ( a_h / a_j ) * ( ( 10 ) * a_k + ( min( g , a_l ) ) ) ) / ( g ) ) + ( ( a_m / a_n ) * a_o / ( g ) ) + ( a_p / ( g ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( max( ( h - i ) / ( g ) , 0 ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( 16 * max( 0 , a_h - a_i ) + ( a_h / a_j ) * ( ( 10 ) * a_k + ( min( g , a_l ) ) ) ) / ( g ) ) / ( ( min( ( 7 ) * a_e + a_f , max( a_c - a_d , 0 ) ) / ( g ) ) + ( 13 * a_g / ( g ) ) + ( min( 2 * ( a_a - r - s ) * dependentloadsweight / 100 , max( a_c - a_d , 0 ) ) / ( g ) ) + ( ( 16 * max( 0 , a_h - a_i ) + ( a_h / a_j ) * ( ( 10 ) * a_k + ( min( g , a_l ) ) ) ) / ( g ) ) + ( ( a_m / a_n ) * a_o / ( g ) ) + ( a_p / ( g ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( max( ( h - i ) / ( g ) , 0 ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( a_m / a_n ) * a_o / ( g ) ) / ( ( min( ( 7 ) * a_e + a_f , max( a_c - a_d , 0 ) ) / ( g ) ) + ( 13 * a_g / ( g ) ) + ( min( 2 * ( a_a - r - s ) * dependentloadsweight / 100 , max( a_c - a_d , 0 ) ) / ( g ) ) + ( ( 16 * max( 0 , a_h - a_i ) + ( a_h / a_j ) * ( ( 10 ) * a_k + ( min( g , a_l ) ) ) ) / ( g ) ) + ( ( a_m / a_n ) * a_o / ( g ) ) + ( a_p / ( g ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( k / ( g ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( a_q / ( a_r if smt_on else ( g ) ) ) / ( ( ( ( a_s * ( 10 ) * ( 1 - ( a_h / a_j ) ) ) + ( 1 - ( a_h / a_j ) ) * ( min( g , a_l ) ) ) / ( g ) ) + ( ( ( 170 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_t + ( 81 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_u ) / ( g ) ) + ( a_q / ( a_r if smt_on else ( g ) ) ) + ( 9 * a_v / ( g ) ) + ( ( ( 7 ) * a_w + a_x ) / ( a_r if smt_on else ( g ) ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( k / ( g ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( ( a_s * ( 10 ) * ( 1 - ( a_h / a_j ) ) ) + ( 1 - ( a_h / a_j ) ) * ( min( g , a_l ) ) ) / ( g ) ) / ( ( ( ( a_s * ( 10 ) * ( 1 - ( a_h / a_j ) ) ) + ( 1 - ( a_h / a_j ) ) * ( min( g , a_l ) ) ) / ( g ) ) + ( ( ( 170 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_t + ( 81 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_u ) / ( g ) ) + ( a_q / ( a_r if smt_on else ( g ) ) ) + ( 9 * a_v / ( g ) ) + ( ( ( 7 ) * a_w + a_x ) / ( a_r if smt_on else ( g ) ) ) ) ) ) )", "BaseFormula": " 100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Latency" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;Mem;MemoryLat;Offcore", @@ -1813,17 +1857,21 @@ "Alias": "a_h" }, { - "Name": "OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM", + "Name": "OCR.DEMAND_RFO.L3_MISS:ocr_msr_val=0x103b800002", "Alias": "a_k" }, { - "Name": "MEM_INST_RETIRED.SPLIT_STORES", + "Name": "OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM", "Alias": "a_l" }, { - "Name": "OCR.STREAMING_WR.ANY_RESPONSE", + "Name": "MEM_INST_RETIRED.SPLIT_STORES", "Alias": "a_m" }, + { + "Name": "OCR.STREAMING_WR.ANY_RESPONSE", + "Alias": "a_n" + }, { "Name": "PERF_METRICS.FRONTEND_BOUND", "Alias": "b" @@ -1943,12 +1991,20 @@ "Alias": "threads" } ], - "Formula": "100 * ( ( a / ( b + c + d + e ) ) * ( ( max( ( f - g ) / ( h ) , 0 ) ) / max( ( a / ( b + c + d + e ) ) , ( ( max( ( f - g ) / ( h ) , 0 ) ) + ( ( g - i ) / ( h ) ) + ( ( i - j ) / ( h ) ) + ( ( j / ( h ) ) ) + ( k / ( h ) ) ) ) ) * ( ( min( ( 7 ) * l + m , max( n - o , 0 ) ) / ( h ) ) / max( ( max( ( f - g ) / ( h ) , 0 ) ) , ( ( min( ( 7 ) * l + m , max( n - o , 0 ) ) / ( h ) ) + ( 13 * p / ( h ) ) + ( min( 2 * ( q - r - s ) * dependentloadsweight / 100 , max( n - o , 0 ) ) / ( h ) ) + ( ( 16 * max( 0 , u - v ) + ( u / w ) * ( ( 10 ) * x + ( min( h , y ) ) ) ) / ( h ) ) + ( ( z / a_a ) * a_b / ( h ) ) + ( a_c / ( h ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( k / ( h ) ) / ( ( max( ( f - g ) / ( h ) , 0 ) ) + ( ( g - i ) / ( h ) ) + ( ( i - j ) / ( h ) ) + ( ( j / ( h ) ) ) + ( k / ( h ) ) ) ) * ( ( ( ( 7 ) * a_d + a_e ) / ( a_f if smt_on else ( h ) ) ) / ( ( ( ( a_g * ( 10 ) * ( 1 - ( u / w ) ) ) + ( 1 - ( u / w ) ) * ( min( h , y ) ) ) / ( h ) ) + ( ( 81 * ( ( ( h ) / a_h ) * a_i / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_k / ( h ) ) + ( a_l / ( a_f if smt_on else ( h ) ) ) + ( 9 * a_m / ( h ) ) + ( ( ( 7 ) * a_d + a_e ) / ( a_f if smt_on else ( h ) ) ) ) ) ) )", + "Formula": "100 * ( ( a / ( b + c + d + e ) ) * ( ( max( ( f - g ) / ( h ) , 0 ) ) / max( ( a / ( b + c + d + e ) ) , ( ( max( ( f - g ) / ( h ) , 0 ) ) + ( ( g - i ) / ( h ) ) + ( ( i - j ) / ( h ) ) + ( ( j / ( h ) ) ) + ( k / ( h ) ) ) ) ) * ( ( min( ( 7 ) * l + m , max( n - o , 0 ) ) / ( h ) ) / max( ( max( ( f - g ) / ( h ) , 0 ) ) , ( ( min( ( 7 ) * l + m , max( n - o , 0 ) ) / ( h ) ) + ( 13 * p / ( h ) ) + ( min( 2 * ( q - r - s ) * dependentloadsweight / 100 , max( n - o , 0 ) ) / ( h ) ) + ( ( 16 * max( 0 , u - v ) + ( u / w ) * ( ( 10 ) * x + ( min( h , y ) ) ) ) / ( h ) ) + ( ( z / a_a ) * a_b / ( h ) ) + ( a_c / ( h ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( k / ( h ) ) / ( ( max( ( f - g ) / ( h ) , 0 ) ) + ( ( g - i ) / ( h ) ) + ( ( i - j ) / ( h ) ) + ( ( j / ( h ) ) ) + ( k / ( h ) ) ) ) * ( ( ( ( 7 ) * a_d + a_e ) / ( a_f if smt_on else ( h ) ) ) / ( ( ( ( a_g * ( 10 ) * ( 1 - ( u / w ) ) ) + ( 1 - ( u / w ) ) * ( min( h , y ) ) ) / ( h ) ) + ( ( ( 170 * ( ( ( h ) / a_h ) * a_i / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_k + ( 81 * ( ( ( h ) / a_h ) * a_i / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_l ) / ( h ) ) + ( a_m / ( a_f if smt_on else ( h ) ) ) + ( 9 * a_n / ( h ) ) + ( ( ( 7 ) * a_d + a_e ) / ( a_f if smt_on else ( h ) ) ) ) ) ) )", "BaseFormula": " 100 * ( tma_memory_bound * ( tma_l1_bound / max( tma_memory_bound , ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) ) * ( tma_dtlb_load / max( tma_l1_bound , ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Data_TLBs" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore", @@ -1982,65 +2038,69 @@ "Alias": "a_d" }, { - "Name": "OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM", + "Name": "OCR.DEMAND_RFO.L3_MISS:ocr_msr_val=0x103b800002", "Alias": "a_e" }, { - "Name": "MEM_STORE_RETIRED.L2_HIT", + "Name": "OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM", "Alias": "a_f" }, { - "Name": "MEM_INST_RETIRED.LOCK_LOADS", + "Name": "MEM_STORE_RETIRED.L2_HIT", "Alias": "a_g" }, { - "Name": "MEM_INST_RETIRED.ALL_STORES", + "Name": "MEM_INST_RETIRED.LOCK_LOADS", "Alias": "a_h" }, { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO", + "Name": "MEM_INST_RETIRED.ALL_STORES", "Alias": "a_i" }, { - "Name": "MEM_INST_RETIRED.SPLIT_STORES", + "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO", "Alias": "a_j" }, { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", + "Name": "MEM_INST_RETIRED.SPLIT_STORES", "Alias": "a_k" }, { - "Name": "OCR.STREAMING_WR.ANY_RESPONSE", + "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", "Alias": "a_l" }, { - "Name": "DTLB_STORE_MISSES.STLB_HIT:c1", + "Name": "OCR.STREAMING_WR.ANY_RESPONSE", "Alias": "a_m" }, { - "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", + "Name": "DTLB_STORE_MISSES.STLB_HIT:c1", "Alias": "a_n" }, { - "Name": "INT_MISC.UOP_DROPPING", + "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", "Alias": "a_o" }, { - "Name": "TOPDOWN.SLOTS:perf_metrics", + "Name": "INT_MISC.UOP_DROPPING", "Alias": "a_p" }, { - "Name": "PERF_METRICS.BRANCH_MISPREDICTS", + "Name": "TOPDOWN.SLOTS:perf_metrics", "Alias": "a_q" }, { - "Name": "MACHINE_CLEARS.MEMORY_ORDERING", + "Name": "PERF_METRICS.BRANCH_MISPREDICTS", "Alias": "a_r" }, { - "Name": "MACHINE_CLEARS.COUNT", + "Name": "MACHINE_CLEARS.MEMORY_ORDERING", "Alias": "a_s" }, + { + "Name": "MACHINE_CLEARS.COUNT", + "Alias": "a_t" + }, { "Name": "PERF_METRICS.FRONTEND_BOUND", "Alias": "b" @@ -2152,15 +2212,23 @@ "Alias": "threads" } ], - "Formula": "100 * ( ( a / ( b + c + d + e ) ) * ( ( ( ( f / ( g ) ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( min( g , l ) ) / ( g ) - ( ( min( g , m ) ) / ( g ) ) ) / ( ( ( min( g , m ) ) / ( g ) ) + ( ( min( g , l ) ) / ( g ) - ( ( min( g , m ) ) / ( g ) ) ) ) ) * ( ( ( ( 170 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * q + ( ( 170 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * r ) * ( 1 + ( s / t ) / 2 ) / ( g ) ) / ( ( ( ( 109 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * u * ( 1 + ( s / t ) / 2 ) / ( g ) ) + ( ( ( 190 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * v * ( 1 + ( s / t ) / 2 ) / ( g ) ) + ( ( ( ( 170 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * q + ( ( 170 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * r ) * ( 1 + ( s / t ) / 2 ) / ( g ) ) ) + ( ( ( j - f ) / ( g ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( ( ( 81 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( w * ( x / ( x + y ) ) ) + ( ( 79 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( z ) ) * ( 1 + ( s / t ) / 2 ) / ( g ) ) + ( ( ( 79 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_a + w * ( 1 - ( x / ( x + y ) ) ) ) * ( 1 + ( s / t ) / 2 ) / ( g ) ) ) / ( ( ( ( ( 81 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( w * ( x / ( x + y ) ) ) + ( ( 79 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( z ) ) * ( 1 + ( s / t ) / 2 ) / ( g ) ) + ( ( ( 79 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_a + w * ( 1 - ( x / ( x + y ) ) ) ) * ( 1 + ( s / t ) / 2 ) / ( g ) ) + ( ( ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_b * ( 1 + ( s / t ) / 2 ) ) / ( g ) ) + ( ( a_c + a_d ) / ( g ) ) ) + ( ( k / ( g ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( 81 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_e / ( g ) ) / ( ( ( ( ( a_f * ( 10 ) * ( 1 - ( a_g / a_h ) ) ) + ( 1 - ( a_g / a_h ) ) * ( min( g , a_i ) ) ) / ( g ) ) + ( ( 81 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_e / ( g ) ) + ( a_j / ( a_k if smt_on else ( g ) ) ) + ( 9 * a_l / ( g ) ) + ( ( ( 7 ) * a_m + a_n ) / ( a_k if smt_on else ( g ) ) ) ) - ( ( ( a_f * ( 10 ) * ( 1 - ( a_g / a_h ) ) ) + ( 1 - ( a_g / a_h ) ) * ( min( g , a_i ) ) ) / ( g ) ) ) ) + ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - a_o / ( a_p ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( a_q / ( b + c + d + e ) ) ) ) * ( 1 - ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - a_o / ( a_p ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( a_q / ( b + c + d + e ) ) ) ) * ( 1 - a_r / a_s ) , 0.0001 ) ) / ( ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - a_o / ( a_p ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( a_q / ( b + c + d + e ) ) ) ) * ( 1 - a_r / a_s ) , 0.0001 ) ) ) ) )", + "Formula": "100 * ( ( a / ( b + c + d + e ) ) * ( ( ( ( f / ( g ) ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( min( g , l ) ) / ( g ) - ( ( min( g , m ) ) / ( g ) ) ) / ( ( ( min( g , m ) ) / ( g ) ) + ( ( min( g , l ) ) / ( g ) - ( ( min( g , m ) ) / ( g ) ) ) ) ) * ( ( ( ( 170 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * q + ( ( 170 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * r ) * ( 1 + ( s / t ) / 2 ) / ( g ) ) / ( ( ( ( 109 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * u * ( 1 + ( s / t ) / 2 ) / ( g ) ) + ( ( ( 190 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * v * ( 1 + ( s / t ) / 2 ) / ( g ) ) + ( ( ( ( 170 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * q + ( ( 170 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * r ) * ( 1 + ( s / t ) / 2 ) / ( g ) ) ) + ( ( ( j - f ) / ( g ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( ( ( 81 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( w * ( x / ( x + y ) ) ) + ( ( 79 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( z ) ) * ( 1 + ( s / t ) / 2 ) / ( g ) ) + ( ( ( 79 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_a + w * ( 1 - ( x / ( x + y ) ) ) ) * ( 1 + ( s / t ) / 2 ) / ( g ) ) ) / ( ( ( ( ( 81 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( w * ( x / ( x + y ) ) ) + ( ( 79 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( z ) ) * ( 1 + ( s / t ) / 2 ) / ( g ) ) + ( ( ( 79 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_a + w * ( 1 - ( x / ( x + y ) ) ) ) * ( 1 + ( s / t ) / 2 ) / ( g ) ) + ( ( ( 37 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_b * ( 1 + ( s / t ) / 2 ) ) / ( g ) ) + ( ( a_c + a_d ) / ( g ) ) ) + ( ( k / ( g ) ) / ( ( max( ( h - i ) / ( g ) , 0 ) ) + ( ( i - j ) / ( g ) ) + ( ( j - f ) / ( g ) ) + ( ( f / ( g ) ) ) + ( k / ( g ) ) ) ) * ( ( ( 170 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_e + ( 81 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_f ) / ( g ) ) / ( ( ( ( ( a_g * ( 10 ) * ( 1 - ( a_h / a_i ) ) ) + ( 1 - ( a_h / a_i ) ) * ( min( g , a_j ) ) ) / ( g ) ) + ( ( ( 170 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_e + ( 81 * ( ( ( g ) / n ) * o / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_f ) / ( g ) ) + ( a_k / ( a_l if smt_on else ( g ) ) ) + ( 9 * a_m / ( g ) ) + ( ( ( 7 ) * a_n + a_o ) / ( a_l if smt_on else ( g ) ) ) ) - ( ( ( a_g * ( 10 ) * ( 1 - ( a_h / a_i ) ) ) + ( 1 - ( a_h / a_i ) ) * ( min( g , a_j ) ) ) / ( g ) ) ) ) + ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - a_p / ( a_q ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( a_r / ( b + c + d + e ) ) ) ) * ( 1 - ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - a_p / ( a_q ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( a_r / ( b + c + d + e ) ) ) ) * ( 1 - a_s / a_t ) , 0.0001 ) ) / ( ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - a_p / ( a_q ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( a_r / ( b + c + d + e ) ) ) ) * ( 1 - a_s / a_t ) , 0.0001 ) ) ) ) )", "BaseFormula": " 100 * ( tma_memory_bound * ( ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) * tma_remote_cache / ( tma_local_mem + tma_remote_mem + tma_remote_cache ) + ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Synchronization > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Synchronization" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Synchronization > 10", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvMS;Mem;Offcore", + "MetricGroup": "BvMS;LockCont;Mem;Offcore", "LocateWith": "" }, { @@ -2262,7 +2330,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Compute_Bound_Est" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20", + "ThresholdIssues": "$issueComp" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Cor", @@ -2467,7 +2543,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Irregular_Overhead > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Irregular_Overhead" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Irregular_Overhead > 10", + "ThresholdIssues": "$issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BvIO;Cor;Ret", @@ -2649,39 +2733,39 @@ "Alias": "b_q" }, { - "Name": "OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM", + "Name": "OCR.DEMAND_RFO.L3_MISS:ocr_msr_val=0x103b800002", "Alias": "b_r" }, { - "Name": "OCR.STREAMING_WR.ANY_RESPONSE", + "Name": "OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM", "Alias": "b_s" }, { - "Name": "DTLB_STORE_MISSES.STLB_HIT:c1", + "Name": "OCR.STREAMING_WR.ANY_RESPONSE", "Alias": "b_t" }, { - "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", + "Name": "DTLB_STORE_MISSES.STLB_HIT:c1", "Alias": "b_u" }, { - "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM", + "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", "Alias": "b_v" }, { - "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD", + "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM", "Alias": "b_w" }, { - "Name": "MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM", + "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD", "Alias": "b_x" }, { - "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM", + "Name": "MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM", "Alias": "b_y" }, { - "Name": "MACHINE_CLEARS.MEMORY_ORDERING", + "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM", "Alias": "b_z" }, { @@ -2689,69 +2773,73 @@ "Alias": "c" }, { - "Name": "ARITH.DIV_ACTIVE", + "Name": "MACHINE_CLEARS.MEMORY_ORDERING", "Alias": "c_a" }, { - "Name": "RESOURCE_STALLS.SCOREBOARD", + "Name": "ARITH.DIV_ACTIVE", "Alias": "c_b" }, { - "Name": "CPU_CLK_UNHALTED.C02", + "Name": "RESOURCE_STALLS.SCOREBOARD", "Alias": "c_c" }, { - "Name": "EXE.AMX_BUSY", + "Name": "CPU_CLK_UNHALTED.C02", "Alias": "c_d" }, { - "Name": "EXE_ACTIVITY.EXE_BOUND_0_PORTS", + "Name": "EXE.AMX_BUSY", "Alias": "c_e" }, { - "Name": "RS.EMPTY_RESOURCE", + "Name": "EXE_ACTIVITY.EXE_BOUND_0_PORTS", "Alias": "c_f" }, { - "Name": "CYCLE_ACTIVITY.STALLS_TOTAL", + "Name": "RS.EMPTY_RESOURCE", "Alias": "c_g" }, { - "Name": "EXE_ACTIVITY.1_PORTS_UTIL", + "Name": "CYCLE_ACTIVITY.STALLS_TOTAL", "Alias": "c_h" }, { - "Name": "EXE_ACTIVITY.2_3_PORTS_UTIL", + "Name": "EXE_ACTIVITY.1_PORTS_UTIL", "Alias": "c_i" }, { - "Name": "UOPS_EXECUTED.CYCLES_GE_3", + "Name": "EXE_ACTIVITY.2_3_PORTS_UTIL", "Alias": "c_j" }, { - "Name": "EXE_ACTIVITY.2_PORTS_UTIL", + "Name": "UOPS_EXECUTED.CYCLES_GE_3", "Alias": "c_k" }, { - "Name": "PERF_METRICS.HEAVY_OPERATIONS", + "Name": "EXE_ACTIVITY.2_PORTS_UTIL", "Alias": "c_l" }, { - "Name": "ASSISTS.ANY", + "Name": "PERF_METRICS.HEAVY_OPERATIONS", "Alias": "c_m" }, { - "Name": "BR_INST_RETIRED.ALL_BRANCHES", + "Name": "ASSISTS.ANY", "Alias": "c_n" }, { - "Name": "BR_INST_RETIRED.NEAR_CALL", + "Name": "BR_INST_RETIRED.ALL_BRANCHES", "Alias": "c_o" }, { - "Name": "INST_RETIRED.NOP", + "Name": "BR_INST_RETIRED.NEAR_CALL", "Alias": "c_p" }, + { + "Name": "INST_RETIRED.NOP", + "Alias": "c_q" + }, { "Name": "PERF_METRICS.RETIRING", "Alias": "d" @@ -2867,12 +2955,20 @@ "Alias": "threads" } ], - "Formula": "100 - ( ( 100 * ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( h / ( i ) ) + ( j / ( i ) ) + ( k / ( i ) ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) ) + ( 100 * ( ( b / ( b + c + d + e ) - f / ( g ) ) - ( 1 - ( 10 * ( r / ( g ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) ) * ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) - ( ( 1 - w / x ) * ( ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) * ( ( ( 1 - ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) ) * l / ( i ) ) + ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) / ( ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) + ( ( 1 - ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) ) * l / ( i ) ) + ( k / ( i ) ) ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) + ( max( 0 , ( b / ( b + c + d + e ) - f / ( g ) ) - ( ( a / ( b + c + d + e ) - f / ( g ) ) ) ) ) * ( max( y , x / ( n / o ) ) / ( z if smt_on else ( i ) ) / 2 ) / ( ( ( a_a - a_b ) / ( z if smt_on else ( i ) ) / 2 ) + ( ( a_c - a_d ) / ( z if smt_on else ( i ) ) / 2 ) + ( max( y , x / ( n / o ) ) / ( z if smt_on else ( i ) ) / 2 ) ) ) ) ) - ( 100 * ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( h / ( i ) ) + ( j / ( i ) ) + ( k / ( i ) ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) ) ) + ( 100 * ( 1 - ( 10 * ( r / ( g ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) ) * ( ( s / ( b + c + d + e ) ) + ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) ) ) + ( 100 * ( ( ( a_e / ( b + c + d + e ) ) * ( ( ( a_f / ( i ) ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( min( i , a_k ) ) / ( i ) ) / ( ( ( min( i , a_k ) ) / ( i ) ) + ( ( min( i , a_l ) ) / ( i ) - ( ( min( i , a_k ) ) / ( i ) ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( ( a_i - a_f ) / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( a_m + a_n ) / ( i ) ) / ( ( ( ( ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_r * ( a_s / ( a_s + a_t ) ) ) + ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_u ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_x + a_r * ( 1 - ( a_s / ( a_s + a_t ) ) ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_y * ( 1 + ( a_v / a_w ) / 2 ) ) / ( i ) ) + ( ( a_m + a_n ) / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( a_z / ( i ) ) / ( ( min( ( 7 ) * b_a + b_b , max( b_c - b_d , 0 ) ) / ( i ) ) + ( 13 * b_e / ( i ) ) + ( min( 2 * ( b_f - a_v - a_w ) * dependentloadsweight / 100 , max( b_c - b_d , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_h - b_i ) + ( b_h / b_j ) * ( ( 10 ) * b_k + ( min( i , b_l ) ) ) ) / ( i ) ) + ( ( b_m / b_n ) * b_o / ( i ) ) + ( a_z / ( i ) ) ) ) ) ) ) + ( 100 * ( ( ( a_e / ( b + c + d + e ) ) * ( ( ( a_f / ( i ) ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( min( i , a_l ) ) / ( i ) - ( ( min( i , a_k ) ) / ( i ) ) ) / ( ( ( min( i , a_k ) ) / ( i ) ) + ( ( min( i , a_l ) ) / ( i ) - ( ( min( i , a_k ) ) / ( i ) ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( ( a_i - a_f ) / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_y * ( 1 + ( a_v / a_w ) / 2 ) ) / ( i ) ) / ( ( ( ( ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_r * ( a_s / ( a_s + a_t ) ) ) + ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_u ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_x + a_r * ( 1 - ( a_s / ( a_s + a_t ) ) ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_y * ( 1 + ( a_v / a_w ) / 2 ) ) / ( i ) ) + ( ( a_m + a_n ) / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( a_h - a_i ) / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( min( 2 * ( b_f - a_v - a_w ) * dependentloadsweight / 100 , max( b_c - b_d , 0 ) ) / ( i ) ) / ( ( min( ( 7 ) * b_a + b_b , max( b_c - b_d , 0 ) ) / ( i ) ) + ( 13 * b_e / ( i ) ) + ( min( 2 * ( b_f - a_v - a_w ) * dependentloadsweight / 100 , max( b_c - b_d , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_h - b_i ) + ( b_h / b_j ) * ( ( 10 ) * b_k + ( min( i , b_l ) ) ) ) / ( i ) ) + ( ( b_m / b_n ) * b_o / ( i ) ) + ( a_z / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( 16 * max( 0 , b_h - b_i ) + ( b_h / b_j ) * ( ( 10 ) * b_k + ( min( i , b_l ) ) ) ) / ( i ) ) / ( ( min( ( 7 ) * b_a + b_b , max( b_c - b_d , 0 ) ) / ( i ) ) + ( 13 * b_e / ( i ) ) + ( min( 2 * ( b_f - a_v - a_w ) * dependentloadsweight / 100 , max( b_c - b_d , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_h - b_i ) + ( b_h / b_j ) * ( ( 10 ) * b_k + ( min( i , b_l ) ) ) ) / ( i ) ) + ( ( b_m / b_n ) * b_o / ( i ) ) + ( a_z / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( b_m / b_n ) * b_o / ( i ) ) / ( ( min( ( 7 ) * b_a + b_b , max( b_c - b_d , 0 ) ) / ( i ) ) + ( 13 * b_e / ( i ) ) + ( min( 2 * ( b_f - a_v - a_w ) * dependentloadsweight / 100 , max( b_c - b_d , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_h - b_i ) + ( b_h / b_j ) * ( ( 10 ) * b_k + ( min( i , b_l ) ) ) ) / ( i ) ) + ( ( b_m / b_n ) * b_o / ( i ) ) + ( a_z / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( a_j / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( b_p / ( z if smt_on else ( i ) ) ) / ( ( ( ( b_q * ( 10 ) * ( 1 - ( b_h / b_j ) ) ) + ( 1 - ( b_h / b_j ) ) * ( min( i , b_l ) ) ) / ( i ) ) + ( ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_r / ( i ) ) + ( b_p / ( z if smt_on else ( i ) ) ) + ( 9 * b_s / ( i ) ) + ( ( ( 7 ) * b_t + b_u ) / ( z if smt_on else ( i ) ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( a_j / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( ( b_q * ( 10 ) * ( 1 - ( b_h / b_j ) ) ) + ( 1 - ( b_h / b_j ) ) * ( min( i , b_l ) ) ) / ( i ) ) / ( ( ( ( b_q * ( 10 ) * ( 1 - ( b_h / b_j ) ) ) + ( 1 - ( b_h / b_j ) ) * ( min( i , b_l ) ) ) / ( i ) ) + ( ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_r / ( i ) ) + ( b_p / ( z if smt_on else ( i ) ) ) + ( 9 * b_s / ( i ) ) + ( ( ( 7 ) * b_t + b_u ) / ( z if smt_on else ( i ) ) ) ) ) ) ) ) + ( 100 * ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) / max( ( a_e / ( b + c + d + e ) ) , ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) ) * ( ( min( ( 7 ) * b_a + b_b , max( b_c - b_d , 0 ) ) / ( i ) ) / max( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) , ( ( min( ( 7 ) * b_a + b_b , max( b_c - b_d , 0 ) ) / ( i ) ) + ( 13 * b_e / ( i ) ) + ( min( 2 * ( b_f - a_v - a_w ) * dependentloadsweight / 100 , max( b_c - b_d , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_h - b_i ) + ( b_h / b_j ) * ( ( 10 ) * b_k + ( min( i , b_l ) ) ) ) / ( i ) ) + ( ( b_m / b_n ) * b_o / ( i ) ) + ( a_z / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( a_j / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( ( 7 ) * b_t + b_u ) / ( z if smt_on else ( i ) ) ) / ( ( ( ( b_q * ( 10 ) * ( 1 - ( b_h / b_j ) ) ) + ( 1 - ( b_h / b_j ) ) * ( min( i , b_l ) ) ) / ( i ) ) + ( ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_r / ( i ) ) + ( b_p / ( z if smt_on else ( i ) ) ) + ( 9 * b_s / ( i ) ) + ( ( ( 7 ) * b_t + b_u ) / ( z if smt_on else ( i ) ) ) ) ) ) ) ) + ( 100 * ( ( a_e / ( b + c + d + e ) ) * ( ( ( ( a_f / ( i ) ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( min( i , a_l ) ) / ( i ) - ( ( min( i , a_k ) ) / ( i ) ) ) / ( ( ( min( i , a_k ) ) / ( i ) ) + ( ( min( i , a_l ) ) / ( i ) - ( ( min( i , a_k ) ) / ( i ) ) ) ) ) * ( ( ( ( 170 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_v + ( ( 170 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_w ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) / ( ( ( ( 109 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_x * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 190 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_y * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( ( 170 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_v + ( ( 170 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_w ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) ) + ( ( ( a_i - a_f ) / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( ( ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_r * ( a_s / ( a_s + a_t ) ) ) + ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_u ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_x + a_r * ( 1 - ( a_s / ( a_s + a_t ) ) ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) ) / ( ( ( ( ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_r * ( a_s / ( a_s + a_t ) ) ) + ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_u ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_x + a_r * ( 1 - ( a_s / ( a_s + a_t ) ) ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_y * ( 1 + ( a_v / a_w ) / 2 ) ) / ( i ) ) + ( ( a_m + a_n ) / ( i ) ) ) + ( ( a_j / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_r / ( i ) ) / ( ( ( ( ( b_q * ( 10 ) * ( 1 - ( b_h / b_j ) ) ) + ( 1 - ( b_h / b_j ) ) * ( min( i , b_l ) ) ) / ( i ) ) + ( ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_r / ( i ) ) + ( b_p / ( z if smt_on else ( i ) ) ) + ( 9 * b_s / ( i ) ) + ( ( ( 7 ) * b_t + b_u ) / ( z if smt_on else ( i ) ) ) ) - ( ( ( b_q * ( 10 ) * ( 1 - ( b_h / b_j ) ) ) + ( 1 - ( b_h / b_j ) ) * ( min( i , b_l ) ) ) / ( i ) ) ) ) + ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - b_z / v ) , 0.0001 ) ) / ( ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - b_z / v ) , 0.0001 ) ) ) ) ) ) + ( 100 * ( ( ( max( 0 , ( e / ( b + c + d + e ) ) - ( a_e / ( b + c + d + e ) ) ) ) * ( c_a / ( i ) ) / ( ( c_a / ( i ) ) + ( c_b / ( i ) + ( c_c / ( i ) ) ) + ( c_d / ( z if smt_on else ( i ) ) ) + ( ( ( ( c_e + max( c_f - c_b , 0 ) ) / ( i ) * ( c_g - a_g ) / ( i ) ) * ( i ) + ( c_h + ( d / ( b + c + d + e ) ) * c_i ) ) / ( i ) if ( c_a < ( c_g - a_g ) ) else ( c_h + ( d / ( b + c + d + e ) ) * c_i ) / ( i ) ) ) ) + ( ( max( 0 , ( e / ( b + c + d + e ) ) - ( a_e / ( b + c + d + e ) ) ) ) * ( c_d / ( z if smt_on else ( i ) ) ) / ( ( c_a / ( i ) ) + ( c_b / ( i ) + ( c_c / ( i ) ) ) + ( c_d / ( z if smt_on else ( i ) ) ) + ( ( ( ( c_e + max( c_f - c_b , 0 ) ) / ( i ) * ( c_g - a_g ) / ( i ) ) * ( i ) + ( c_h + ( d / ( b + c + d + e ) ) * c_i ) ) / ( i ) if ( c_a < ( c_g - a_g ) ) else ( c_h + ( d / ( b + c + d + e ) ) * c_i ) / ( i ) ) ) ) + ( ( max( 0 , ( e / ( b + c + d + e ) ) - ( a_e / ( b + c + d + e ) ) ) ) * ( ( ( ( ( c_e + max( c_f - c_b , 0 ) ) / ( i ) * ( c_g - a_g ) / ( i ) ) * ( i ) + ( c_h + ( d / ( b + c + d + e ) ) * c_i ) ) / ( i ) if ( c_a < ( c_g - a_g ) ) else ( c_h + ( d / ( b + c + d + e ) ) * c_i ) / ( i ) ) / ( ( c_a / ( i ) ) + ( c_b / ( i ) + ( c_c / ( i ) ) ) + ( c_d / ( z if smt_on else ( i ) ) ) + ( ( ( ( c_e + max( c_f - c_b , 0 ) ) / ( i ) * ( c_g - a_g ) / ( i ) ) * ( i ) + ( c_h + ( d / ( b + c + d + e ) ) * c_i ) ) / ( i ) if ( c_a < ( c_g - a_g ) ) else ( c_h + ( d / ( b + c + d + e ) ) * c_i ) / ( i ) ) ) ) * ( ( c_j / ( i ) ) / ( ( ( c_e + max( c_f - c_b , 0 ) ) / ( i ) * ( c_g - a_g ) / ( i ) ) + ( c_h / ( i ) ) + ( c_k / ( i ) ) + ( c_j / ( i ) ) ) ) ) ) ) + ( 100 * ( ( ( 1 - w / x ) * ( ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) * ( ( ( 1 - ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) ) * l / ( i ) ) + ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) / ( ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) + ( ( 1 - ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) ) * l / ( i ) ) + ( k / ( i ) ) ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) + ( max( 0 , ( b / ( b + c + d + e ) - f / ( g ) ) - ( ( a / ( b + c + d + e ) - f / ( g ) ) ) ) ) * ( max( y , x / ( n / o ) ) / ( z if smt_on else ( i ) ) / 2 ) / ( ( ( a_a - a_b ) / ( z if smt_on else ( i ) ) / 2 ) + ( ( a_c - a_d ) / ( z if smt_on else ( i ) ) / 2 ) + ( max( y , x / ( n / o ) ) / ( z if smt_on else ( i ) ) / 2 ) ) ) ) + ( 10 * ( r / ( g ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) * ( s / ( b + c + d + e ) ) + ( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - b_z / v ) , 0.0001 ) ) / ( ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - b_z / v ) , 0.0001 ) ) ) ) + ( ( max( 0 , ( e / ( b + c + d + e ) ) - ( a_e / ( b + c + d + e ) ) ) ) * ( ( c_b / ( i ) + ( c_c / ( i ) ) ) + c_f / ( i ) * ( ( c_e + max( c_f - c_b , 0 ) ) / ( i ) * ( c_g - a_g ) / ( i ) ) ) / ( ( c_a / ( i ) ) + ( c_b / ( i ) + ( c_c / ( i ) ) ) + ( c_d / ( z if smt_on else ( i ) ) ) + ( ( ( ( c_e + max( c_f - c_b , 0 ) ) / ( i ) * ( c_g - a_g ) / ( i ) ) * ( i ) + ( c_h + ( d / ( b + c + d + e ) ) * c_i ) ) / ( i ) if ( c_a < ( c_g - a_g ) ) else ( c_h + ( d / ( b + c + d + e ) ) * c_i ) / ( i ) ) ) ) + ( ( ( ( r / ( g ) ) / ( ( max( 0 , ( c_l / ( b + c + d + e ) ) - ( r / ( g ) ) ) ) + ( r / ( g ) ) ) ) * ( ( ( ( 99 * 3 + 63 + 30 ) / 5 ) * c_m / ( g ) ) / ( r / ( g ) ) ) ) * ( c_l / ( b + c + d + e ) ) ) ) ) + ( 100 * ( ( c_n + 2 * c_o + c_p ) / ( g ) ) ) + ( 100 * ( ( d / ( b + c + d + e ) ) - ( ( c_n + 2 * c_o + c_p ) / ( g ) ) - ( ( ( ( r / ( g ) ) / ( ( max( 0 , ( c_l / ( b + c + d + e ) ) - ( r / ( g ) ) ) ) + ( r / ( g ) ) ) ) * ( ( ( ( 99 * 3 + 63 + 30 ) / 5 ) * c_m / ( g ) ) / ( r / ( g ) ) ) ) * ( c_l / ( b + c + d + e ) ) ) ) ) )", + "Formula": "100 - ( ( 100 * ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( h / ( i ) ) + ( j / ( i ) ) + ( k / ( i ) ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) ) + ( 100 * ( ( b / ( b + c + d + e ) - f / ( g ) ) - ( 1 - ( 10 * ( r / ( g ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) ) * ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) - ( ( 1 - w / x ) * ( ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) * ( ( ( 1 - ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) ) * l / ( i ) ) + ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) / ( ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) + ( ( 1 - ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) ) * l / ( i ) ) + ( k / ( i ) ) ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) + ( max( 0 , ( b / ( b + c + d + e ) - f / ( g ) ) - ( ( a / ( b + c + d + e ) - f / ( g ) ) ) ) ) * ( max( y , x / ( n / o ) ) / ( z if smt_on else ( i ) ) / 2 ) / ( ( ( a_a - a_b ) / ( z if smt_on else ( i ) ) / 2 ) + ( ( a_c - a_d ) / ( z if smt_on else ( i ) ) / 2 ) + ( max( y , x / ( n / o ) ) / ( z if smt_on else ( i ) ) / 2 ) ) ) ) ) - ( 100 * ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( h / ( i ) ) + ( j / ( i ) ) + ( k / ( i ) ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) ) ) + ( 100 * ( 1 - ( 10 * ( r / ( g ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) ) * ( ( s / ( b + c + d + e ) ) + ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) ) ) + ( 100 * ( ( ( a_e / ( b + c + d + e ) ) * ( ( ( a_f / ( i ) ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( min( i , a_k ) ) / ( i ) ) / ( ( ( min( i , a_k ) ) / ( i ) ) + ( ( min( i , a_l ) ) / ( i ) - ( ( min( i , a_k ) ) / ( i ) ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( ( a_i - a_f ) / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( a_m + a_n ) / ( i ) ) / ( ( ( ( ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_r * ( a_s / ( a_s + a_t ) ) ) + ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_u ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_x + a_r * ( 1 - ( a_s / ( a_s + a_t ) ) ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_y * ( 1 + ( a_v / a_w ) / 2 ) ) / ( i ) ) + ( ( a_m + a_n ) / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( a_z / ( i ) ) / ( ( min( ( 7 ) * b_a + b_b , max( b_c - b_d , 0 ) ) / ( i ) ) + ( 13 * b_e / ( i ) ) + ( min( 2 * ( b_f - a_v - a_w ) * dependentloadsweight / 100 , max( b_c - b_d , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_h - b_i ) + ( b_h / b_j ) * ( ( 10 ) * b_k + ( min( i , b_l ) ) ) ) / ( i ) ) + ( ( b_m / b_n ) * b_o / ( i ) ) + ( a_z / ( i ) ) ) ) ) ) ) + ( 100 * ( ( ( a_e / ( b + c + d + e ) ) * ( ( ( a_f / ( i ) ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( min( i , a_l ) ) / ( i ) - ( ( min( i , a_k ) ) / ( i ) ) ) / ( ( ( min( i , a_k ) ) / ( i ) ) + ( ( min( i , a_l ) ) / ( i ) - ( ( min( i , a_k ) ) / ( i ) ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( ( a_i - a_f ) / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_y * ( 1 + ( a_v / a_w ) / 2 ) ) / ( i ) ) / ( ( ( ( ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_r * ( a_s / ( a_s + a_t ) ) ) + ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_u ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_x + a_r * ( 1 - ( a_s / ( a_s + a_t ) ) ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_y * ( 1 + ( a_v / a_w ) / 2 ) ) / ( i ) ) + ( ( a_m + a_n ) / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( a_h - a_i ) / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( min( 2 * ( b_f - a_v - a_w ) * dependentloadsweight / 100 , max( b_c - b_d , 0 ) ) / ( i ) ) / ( ( min( ( 7 ) * b_a + b_b , max( b_c - b_d , 0 ) ) / ( i ) ) + ( 13 * b_e / ( i ) ) + ( min( 2 * ( b_f - a_v - a_w ) * dependentloadsweight / 100 , max( b_c - b_d , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_h - b_i ) + ( b_h / b_j ) * ( ( 10 ) * b_k + ( min( i , b_l ) ) ) ) / ( i ) ) + ( ( b_m / b_n ) * b_o / ( i ) ) + ( a_z / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( 16 * max( 0 , b_h - b_i ) + ( b_h / b_j ) * ( ( 10 ) * b_k + ( min( i , b_l ) ) ) ) / ( i ) ) / ( ( min( ( 7 ) * b_a + b_b , max( b_c - b_d , 0 ) ) / ( i ) ) + ( 13 * b_e / ( i ) ) + ( min( 2 * ( b_f - a_v - a_w ) * dependentloadsweight / 100 , max( b_c - b_d , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_h - b_i ) + ( b_h / b_j ) * ( ( 10 ) * b_k + ( min( i , b_l ) ) ) ) / ( i ) ) + ( ( b_m / b_n ) * b_o / ( i ) ) + ( a_z / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( b_m / b_n ) * b_o / ( i ) ) / ( ( min( ( 7 ) * b_a + b_b , max( b_c - b_d , 0 ) ) / ( i ) ) + ( 13 * b_e / ( i ) ) + ( min( 2 * ( b_f - a_v - a_w ) * dependentloadsweight / 100 , max( b_c - b_d , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_h - b_i ) + ( b_h / b_j ) * ( ( 10 ) * b_k + ( min( i , b_l ) ) ) ) / ( i ) ) + ( ( b_m / b_n ) * b_o / ( i ) ) + ( a_z / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( a_j / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( b_p / ( z if smt_on else ( i ) ) ) / ( ( ( ( b_q * ( 10 ) * ( 1 - ( b_h / b_j ) ) ) + ( 1 - ( b_h / b_j ) ) * ( min( i , b_l ) ) ) / ( i ) ) + ( ( ( 170 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_r + ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_s ) / ( i ) ) + ( b_p / ( z if smt_on else ( i ) ) ) + ( 9 * b_t / ( i ) ) + ( ( ( 7 ) * b_u + b_v ) / ( z if smt_on else ( i ) ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( a_j / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( ( b_q * ( 10 ) * ( 1 - ( b_h / b_j ) ) ) + ( 1 - ( b_h / b_j ) ) * ( min( i , b_l ) ) ) / ( i ) ) / ( ( ( ( b_q * ( 10 ) * ( 1 - ( b_h / b_j ) ) ) + ( 1 - ( b_h / b_j ) ) * ( min( i , b_l ) ) ) / ( i ) ) + ( ( ( 170 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_r + ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_s ) / ( i ) ) + ( b_p / ( z if smt_on else ( i ) ) ) + ( 9 * b_t / ( i ) ) + ( ( ( 7 ) * b_u + b_v ) / ( z if smt_on else ( i ) ) ) ) ) ) ) ) + ( 100 * ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) / max( ( a_e / ( b + c + d + e ) ) , ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) ) * ( ( min( ( 7 ) * b_a + b_b , max( b_c - b_d , 0 ) ) / ( i ) ) / max( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) , ( ( min( ( 7 ) * b_a + b_b , max( b_c - b_d , 0 ) ) / ( i ) ) + ( 13 * b_e / ( i ) ) + ( min( 2 * ( b_f - a_v - a_w ) * dependentloadsweight / 100 , max( b_c - b_d , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_h - b_i ) + ( b_h / b_j ) * ( ( 10 ) * b_k + ( min( i , b_l ) ) ) ) / ( i ) ) + ( ( b_m / b_n ) * b_o / ( i ) ) + ( a_z / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( a_j / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( ( 7 ) * b_u + b_v ) / ( z if smt_on else ( i ) ) ) / ( ( ( ( b_q * ( 10 ) * ( 1 - ( b_h / b_j ) ) ) + ( 1 - ( b_h / b_j ) ) * ( min( i , b_l ) ) ) / ( i ) ) + ( ( ( 170 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_r + ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_s ) / ( i ) ) + ( b_p / ( z if smt_on else ( i ) ) ) + ( 9 * b_t / ( i ) ) + ( ( ( 7 ) * b_u + b_v ) / ( z if smt_on else ( i ) ) ) ) ) ) ) ) + ( 100 * ( ( a_e / ( b + c + d + e ) ) * ( ( ( ( a_f / ( i ) ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( min( i , a_l ) ) / ( i ) - ( ( min( i , a_k ) ) / ( i ) ) ) / ( ( ( min( i , a_k ) ) / ( i ) ) + ( ( min( i , a_l ) ) / ( i ) - ( ( min( i , a_k ) ) / ( i ) ) ) ) ) * ( ( ( ( 170 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_w + ( ( 170 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_x ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) / ( ( ( ( 109 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_y * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 190 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_z * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( ( 170 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_w + ( ( 170 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_x ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) ) + ( ( ( a_i - a_f ) / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( ( ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_r * ( a_s / ( a_s + a_t ) ) ) + ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_u ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_x + a_r * ( 1 - ( a_s / ( a_s + a_t ) ) ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) ) / ( ( ( ( ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_r * ( a_s / ( a_s + a_t ) ) ) + ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_u ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 79 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_x + a_r * ( 1 - ( a_s / ( a_s + a_t ) ) ) ) * ( 1 + ( a_v / a_w ) / 2 ) / ( i ) ) + ( ( ( 37 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_y * ( 1 + ( a_v / a_w ) / 2 ) ) / ( i ) ) + ( ( a_m + a_n ) / ( i ) ) ) + ( ( a_j / ( i ) ) / ( ( max( ( a_g - a_h ) / ( i ) , 0 ) ) + ( ( a_h - a_i ) / ( i ) ) + ( ( a_i - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) ) + ( a_j / ( i ) ) ) ) * ( ( ( 170 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_r + ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_s ) / ( i ) ) / ( ( ( ( ( b_q * ( 10 ) * ( 1 - ( b_h / b_j ) ) ) + ( 1 - ( b_h / b_j ) ) * ( min( i , b_l ) ) ) / ( i ) ) + ( ( ( 170 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_r + ( 81 * ( ( ( i ) / a_o ) * a_p / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_s ) / ( i ) ) + ( b_p / ( z if smt_on else ( i ) ) ) + ( 9 * b_t / ( i ) ) + ( ( ( 7 ) * b_u + b_v ) / ( z if smt_on else ( i ) ) ) ) - ( ( ( b_q * ( 10 ) * ( 1 - ( b_h / b_j ) ) ) + ( 1 - ( b_h / b_j ) ) * ( min( i , b_l ) ) ) / ( i ) ) ) ) + ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - c_a / v ) , 0.0001 ) ) / ( ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - c_a / v ) , 0.0001 ) ) ) ) ) ) + ( 100 * ( ( ( max( 0 , ( e / ( b + c + d + e ) ) - ( a_e / ( b + c + d + e ) ) ) ) * ( c_b / ( i ) ) / ( ( c_b / ( i ) ) + ( c_c / ( i ) + ( c_d / ( i ) ) ) + ( c_e / ( z if smt_on else ( i ) ) ) + ( ( ( ( c_f + max( c_g - c_c , 0 ) ) / ( i ) * ( c_h - a_g ) / ( i ) ) * ( i ) + ( c_i + ( d / ( b + c + d + e ) ) * c_j ) ) / ( i ) if ( c_b < ( c_h - a_g ) ) else ( c_i + ( d / ( b + c + d + e ) ) * c_j ) / ( i ) ) ) ) + ( ( max( 0 , ( e / ( b + c + d + e ) ) - ( a_e / ( b + c + d + e ) ) ) ) * ( c_e / ( z if smt_on else ( i ) ) ) / ( ( c_b / ( i ) ) + ( c_c / ( i ) + ( c_d / ( i ) ) ) + ( c_e / ( z if smt_on else ( i ) ) ) + ( ( ( ( c_f + max( c_g - c_c , 0 ) ) / ( i ) * ( c_h - a_g ) / ( i ) ) * ( i ) + ( c_i + ( d / ( b + c + d + e ) ) * c_j ) ) / ( i ) if ( c_b < ( c_h - a_g ) ) else ( c_i + ( d / ( b + c + d + e ) ) * c_j ) / ( i ) ) ) ) + ( ( max( 0 , ( e / ( b + c + d + e ) ) - ( a_e / ( b + c + d + e ) ) ) ) * ( ( ( ( ( c_f + max( c_g - c_c , 0 ) ) / ( i ) * ( c_h - a_g ) / ( i ) ) * ( i ) + ( c_i + ( d / ( b + c + d + e ) ) * c_j ) ) / ( i ) if ( c_b < ( c_h - a_g ) ) else ( c_i + ( d / ( b + c + d + e ) ) * c_j ) / ( i ) ) / ( ( c_b / ( i ) ) + ( c_c / ( i ) + ( c_d / ( i ) ) ) + ( c_e / ( z if smt_on else ( i ) ) ) + ( ( ( ( c_f + max( c_g - c_c , 0 ) ) / ( i ) * ( c_h - a_g ) / ( i ) ) * ( i ) + ( c_i + ( d / ( b + c + d + e ) ) * c_j ) ) / ( i ) if ( c_b < ( c_h - a_g ) ) else ( c_i + ( d / ( b + c + d + e ) ) * c_j ) / ( i ) ) ) ) * ( ( c_k / ( i ) ) / ( ( ( c_f + max( c_g - c_c , 0 ) ) / ( i ) * ( c_h - a_g ) / ( i ) ) + ( c_i / ( i ) ) + ( c_l / ( i ) ) + ( c_k / ( i ) ) ) ) ) ) ) + ( 100 * ( ( ( 1 - w / x ) * ( ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) * ( ( ( 1 - ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) ) * l / ( i ) ) + ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) / ( ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) + ( ( 1 - ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) ) * l / ( i ) ) + ( k / ( i ) ) ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) + ( max( 0 , ( b / ( b + c + d + e ) - f / ( g ) ) - ( ( a / ( b + c + d + e ) - f / ( g ) ) ) ) ) * ( max( y , x / ( n / o ) ) / ( z if smt_on else ( i ) ) / 2 ) / ( ( ( a_a - a_b ) / ( z if smt_on else ( i ) ) / 2 ) + ( ( a_c - a_d ) / ( z if smt_on else ( i ) ) / 2 ) + ( max( y , x / ( n / o ) ) / ( z if smt_on else ( i ) ) / 2 ) ) ) ) + ( 10 * ( r / ( g ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) * ( s / ( b + c + d + e ) ) + ( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - c_a / v ) , 0.0001 ) ) / ( ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - c_a / v ) , 0.0001 ) ) ) ) + ( ( max( 0 , ( e / ( b + c + d + e ) ) - ( a_e / ( b + c + d + e ) ) ) ) * ( ( c_c / ( i ) + ( c_d / ( i ) ) ) + c_g / ( i ) * ( ( c_f + max( c_g - c_c , 0 ) ) / ( i ) * ( c_h - a_g ) / ( i ) ) ) / ( ( c_b / ( i ) ) + ( c_c / ( i ) + ( c_d / ( i ) ) ) + ( c_e / ( z if smt_on else ( i ) ) ) + ( ( ( ( c_f + max( c_g - c_c , 0 ) ) / ( i ) * ( c_h - a_g ) / ( i ) ) * ( i ) + ( c_i + ( d / ( b + c + d + e ) ) * c_j ) ) / ( i ) if ( c_b < ( c_h - a_g ) ) else ( c_i + ( d / ( b + c + d + e ) ) * c_j ) / ( i ) ) ) ) + ( ( ( ( r / ( g ) ) / ( ( max( 0 , ( c_m / ( b + c + d + e ) ) - ( r / ( g ) ) ) ) + ( r / ( g ) ) ) ) * ( ( ( ( 99 * 3 + 63 + 30 ) / 5 ) * c_n / ( g ) ) / ( r / ( g ) ) ) ) * ( c_m / ( b + c + d + e ) ) ) ) ) + ( 100 * ( ( c_o + 2 * c_p + c_q ) / ( g ) ) ) + ( 100 * ( ( d / ( b + c + d + e ) ) - ( ( c_o + 2 * c_p + c_q ) / ( g ) ) - ( ( ( ( r / ( g ) ) / ( ( max( 0 , ( c_m / ( b + c + d + e ) ) - ( r / ( g ) ) ) ) + ( r / ( g ) ) ) ) * ( ( ( ( 99 * 3 + 63 + 30 ) / 5 ) * c_n / ( g ) ) / ( r / ( g ) ) ) ) * ( c_m / ( b + c + d + e ) ) ) ) ) )", "BaseFormula": " 100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Other_Bottlenecks" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;Cor;Offcore", @@ -2908,7 +3004,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Branching_Overhead > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Branching_Overhead" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Bottleneck_Branching_Overhead > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Ret", @@ -2972,7 +3076,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Useful_Work > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Useful_Work" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Useful_Work > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;Ret", @@ -3016,7 +3128,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;BvIO;TmaL1;PGO", @@ -3065,7 +3185,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Frontend;TmaL2", @@ -3094,7 +3226,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss", @@ -3127,7 +3275,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -3156,7 +3324,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -3185,7 +3373,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB", @@ -3218,7 +3422,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3247,8 +3471,28 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", "LocateWith": "" @@ -3284,7 +3528,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3321,7 +3589,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3354,7 +3646,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -3411,7 +3719,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Mispredicts_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP", @@ -3468,7 +3796,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Clears_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;MachineClears", @@ -3497,7 +3845,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Unknown_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat", @@ -3534,7 +3902,23 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC, $issueMS, $issueMV, $issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MicroSeq", @@ -3563,7 +3947,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LCP(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -3592,7 +3992,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchLat", @@ -3641,7 +4057,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;Frontend;TmaL2", @@ -3687,7 +4111,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -3733,7 +4169,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Decoder0_Alone(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -3779,7 +4231,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;FetchBW", @@ -3833,7 +4297,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 5 & b > 20", + "BaseFormula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -3877,7 +4353,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", @@ -3918,7 +4402,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2", @@ -3971,7 +4467,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;BrMispredicts", @@ -4020,7 +4532,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueMC, $issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2", @@ -4077,7 +4601,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Nukes(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;Machine_Clears", @@ -4113,7 +4653,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;TmaL1", @@ -4154,7 +4702,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20", + "BaseFormula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2", @@ -4187,7 +4747,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1, $issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -4228,7 +4804,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -4269,14 +4865,38 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Load_STLB_Miss", + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" + }, + "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", + "MetricGroup": "MemoryTLB", + "LocateWith": "" + }, + { + "MetricName": "Load_STLB_Miss", "LegacyName": "metric_TMA_........Load_STLB_Miss(%)", "ParentCategory": "DTLB_Load", "Level": 5, @@ -4298,7 +4918,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4339,7 +4983,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4380,7 +5052,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4421,7 +5121,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4450,7 +5178,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Fwd_Blk(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4500,7 +5248,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L1_Latency_Dependency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -4545,7 +5313,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Lock_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Offcore", @@ -4582,7 +5370,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......Split_Loads(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Loads(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......Split_Loads(%) > 30", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4611,7 +5407,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......FB_Full(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FB_Full(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......FB_Full(%) > 30", + "ThresholdIssues": "$issueBW, $issueSL, $issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW", @@ -4644,7 +5448,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem", @@ -4694,7 +5514,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L2_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryLat", @@ -4727,7 +5567,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -4789,7 +5645,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Contested_Accesses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -4851,7 +5727,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Data_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;Offcore;Snoop", @@ -4901,7 +5797,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L3_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -4934,7 +5850,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......SQ_Full(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 30 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -4963,7 +5899,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -4972,6 +5924,7 @@ { "MetricName": "MEM_Bandwidth", "LegacyName": "metric_TMA_......MEM_Bandwidth(%)", + "ParentCategory": "DRAM_Bound", "Level": 4, "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that).", "UnitOfMeasure": "percent", @@ -4991,7 +5944,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Bandwidth(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -5020,7 +5993,31 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_........MBA_Stalls(%) > 10 & metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........MBA_Stalls(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......MEM_Bandwidth(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........MBA_Stalls(%) > 10 & metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBW;Offcore;Server", @@ -5029,6 +6026,7 @@ { "MetricName": "MEM_Latency", "LegacyName": "metric_TMA_......MEM_Latency(%)", + "ParentCategory": "DRAM_Bound", "Level": 4, "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that).", "UnitOfMeasure": "percent", @@ -5052,7 +6050,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat;Offcore", @@ -5102,7 +6120,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Local_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Local_MEM(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Local_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Server", @@ -5152,7 +6194,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Remote_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Remote_MEM(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Remote_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Server;Snoop", @@ -5206,7 +6272,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Remote_Cache(%) > 5 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Remote_Cache(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Remote_Cache(%) > 5 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Offcore;Server;Snoop", @@ -5235,7 +6325,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -5276,7 +6382,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO, $issueSL, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore", @@ -5322,7 +6448,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......False_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -5364,7 +6510,27 @@ "Category": "TMA", "CountDomain": "Core_Utilization", "Threshold": { - "Formula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5393,7 +6559,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Streaming_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBW;Offcore", @@ -5439,7 +6625,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -5485,7 +6691,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5527,7 +6757,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5581,7 +6835,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5635,7 +6917,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5689,7 +6999,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5730,7 +7068,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2;Compute", @@ -5759,7 +7109,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB", @@ -5788,7 +7154,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5821,7 +7207,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......INT_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5854,7 +7260,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;PortsUtil", @@ -5883,7 +7305,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Slow_Pause(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5912,7 +7354,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......C01_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......C01_Wait(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......C01_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", @@ -5941,7 +7403,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......C02_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......C02_Wait(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......C02_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", @@ -5970,7 +7452,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Memory_Fence(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Memory_Fence(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Memory_Fence(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6012,7 +7514,23 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_....AMX_Busy(%) > 50 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....AMX_Busy(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 50 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....AMX_Busy(%) > 50 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Compute;HPC;Server", @@ -6085,7 +7603,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -6130,7 +7664,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_0(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -6159,7 +7713,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_........Mixing_Vectors(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Mixing_Vectors(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Mixing_Vectors(%) > 5", + "ThresholdIssues": "$issueMV" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6188,7 +7750,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_1(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -6217,7 +7799,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_2(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -6246,7 +7848,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_3m(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 40 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;PortsUtil", @@ -6300,7 +7922,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........ALU_Op_Utilization(%) > 40" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........ALU_Op_Utilization(%)" + } + ], + "Formula": "a > 40", + "BaseFormula": "metric_TMA_........ALU_Op_Utilization(%) > 40", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6342,7 +7972,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_0(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_0(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_0(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -6384,7 +8022,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_1(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_1(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_1(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6426,7 +8072,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_6(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_6(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_6(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6468,7 +8122,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Load_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Load_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6514,7 +8176,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Store_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Store_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6550,7 +8220,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Retiring(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 70 | b > 10", + "BaseFormula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;TmaL1", @@ -6591,7 +8273,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -6656,7 +8346,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 20 & b > 60", + "BaseFormula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -6701,7 +8403,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......X87_Use(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -6750,7 +8468,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Scalar(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -6799,7 +8533,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -6852,7 +8602,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -6905,7 +8675,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -6958,7 +8748,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_512b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -7019,7 +8829,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -7068,7 +8890,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......Int_Vector_128b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Int_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 60", + "BaseFormula": "metric_TMA_......Int_Vector_128b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;IntVector;Pipeline", @@ -7121,7 +8959,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......Int_Vector_256b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Int_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 60", + "BaseFormula": "metric_TMA_......Int_Vector_256b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;IntVector;Pipeline", @@ -7170,7 +9024,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Memory_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -7219,7 +9085,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Fused_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -7272,7 +9150,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Non_Fused_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -7373,7 +9263,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 30 & b > 60", + "BaseFormula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -7422,7 +9324,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Nop_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Pipeline", @@ -7471,7 +9389,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Shuffles_256b(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Shuffles_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Shuffles_256b(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC;Pipeline", @@ -7512,7 +9446,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -7561,7 +9503,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Few_Uops_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -7590,7 +9544,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueMC, $issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -7619,7 +9585,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Assists(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO", @@ -7648,7 +9630,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........Page_Faults(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Page_Faults(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Page_Faults(%) > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -7677,7 +9667,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........FP_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........FP_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -7706,7 +9704,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........AVX_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........AVX_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........AVX_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -7739,7 +9745,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......CISC(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -9217,6 +11239,11 @@ "BaseFormula": " uops_executed.thread / ( ( uops_executed.core_cycles_ge_1 / 2 ) if smt_on else uops_executed.thread:c1 )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "LocateWith": "" @@ -9432,6 +11459,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -11273,7 +13305,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Useless_HWPF" + } + ], + "Formula": "a > 0.15", + "BaseFormula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", diff --git a/EMR/metrics/perf/emeraldrapids_metrics_perf.json b/EMR/metrics/perf/emeraldrapids_metrics_perf.json index 056cfd3b..0c15d57c 100644 --- a/EMR/metrics/perf/emeraldrapids_metrics_perf.json +++ b/EMR/metrics/perf/emeraldrapids_metrics_perf.json @@ -1,132 +1,132 @@ [ { - "BriefDescription": "CPU operating frequency (in GHz).", - "MetricExpr": "( CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000", + "BriefDescription": "CPU operating frequency (in GHz)", + "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000", "MetricGroup": "", "MetricName": "cpu_operating_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Percentage of time spent in the active CPU power state C0.", + "BriefDescription": "Percentage of time spent in the active CPU power state C0", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "", "MetricName": "cpu_utilization", "ScaleUnit": "100%" }, { - "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles.", + "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles", "MetricExpr": "CPU_CLK_UNHALTED.THREAD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions", "MetricExpr": "MEM_INST_RETIRED.ALL_LOADS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "loads_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions", "MetricExpr": "MEM_INST_RETIRED.ALL_STORES / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "stores_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L1D.REPLACEMENT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L1_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions", "MetricExpr": "L2_RQSTS.ALL_CODE_RD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1_i_code_read_misses_with_prefetches_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L2_LINES_IN.ALL / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions", "MetricExpr": "L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_code_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_LLCPREFDATA + UNC_CHA_TOR_INSERTS.IA_MISS_DRD + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF ) / INST_RETIRED.ANY", + "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IA_MISS_LLCPREFDATA + UNC_CHA_TOR_INSERTS.IA_MISS_DRD + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF) / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_data_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", "MetricExpr": "UNC_CHA_TOR_INSERTS.IA_MISS_CRD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_code_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_latency", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to local memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to local memory in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_latency_for_local_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to remote memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to remote memory in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_latency_for_remote_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to DRAM in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to DRAM in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_to_dram_latency", "ScaleUnit": "1ns" }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_2nd_level_mpi", @@ -134,7 +134,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the ITLB (Instruction TLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED_2M_4M / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_2nd_level_large_page_mpi", @@ -142,7 +142,7 @@ "PublicDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the Instruction Translation Lookaside Buffer (ITLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions", "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_2nd_level_load_mpi", @@ -150,7 +150,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions", "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_2nd_level_2mb_large_page_load_mpi", @@ -158,7 +158,7 @@ "PublicDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the Data Translation Lookaside Buffer (DTLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions", "MetricExpr": "DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_2nd_level_store_mpi", @@ -166,1295 +166,1430 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL ) / ( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE )", + "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL) / (UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_local_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE ) / ( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE )", + "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE) / (UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_remote_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Uncore operating frequency in GHz.", - "MetricExpr": "( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) / 1000000000) / duration_time", + "BriefDescription": "Uncore operating frequency in GHz", + "MetricExpr": "(UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) / 1000000000) / duration_time", "MetricGroup": "", "MetricName": "uncore_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data transmit bandwidth (MB/sec).", - "MetricExpr": "( UNC_UPI_TxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", + "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data transmit bandwidth (MB/sec)", + "MetricExpr": "(UNC_UPI_TxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", "MetricGroup": "", "MetricName": "upi_data_transmit_bw", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory read bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory read bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory write bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory write bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory bandwidth (MB/sec)", + "MetricExpr": "((UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_total", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Memory write bandwidth (MB/sec) caused by directory updates; includes DDR and Intel(R) Optane(TM) Persistent Memory(PMEM).", - "MetricExpr": "(( UNC_CHA_DIR_UPDATE.HA + UNC_CHA_DIR_UPDATE.TOR + UNC_M2M_DIRECTORY_UPDATE.ANY ) * 64 / 1000000) / duration_time", + "BriefDescription": "Memory write bandwidth (MB/sec) caused by directory updates; includes DDR and Intel(R) Optane(TM) Persistent Memory(PMEM)", + "MetricExpr": "((UNC_CHA_DIR_UPDATE.HA + UNC_CHA_DIR_UPDATE.TOR + UNC_M2M_DIRECTORY_UPDATE.ANY) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_extra_write_bw_due_to_directory_updates", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_ITOM + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR ) * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_ITOM + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Percentage of inbound reads initiated by end device controllers that miss the L3 cache.", + "BriefDescription": "Percentage of inbound reads initiated by end device controllers that miss the L3 cache", "MetricExpr": "UNC_CHA_TOR_INSERTS.IO_MISS_PCIRDCUR / UNC_CHA_TOR_INSERTS.IO_PCIRDCUR", "MetricGroup": "", "MetricName": "io_percent_of_inbound_reads_that_miss_l3", "ScaleUnit": "100%" }, { - "BriefDescription": "Percentage of inbound partial cacheline writes initiated by end device controllers that miss the L3 cache.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_MISS_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_MISS_RFO ) / ( UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_RFO ) )", + "BriefDescription": "Percentage of inbound partial cacheline writes initiated by end device controllers that miss the L3 cache", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_MISS_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_MISS_RFO) / (UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_RFO) )", "MetricGroup": "", "MetricName": "io_percent_of_inbound_partial_writes_that_miss_l3", "ScaleUnit": "100%" }, { - "BriefDescription": "Percentage of inbound full cacheline writes initiated by end device controllers that miss the L3 cache.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_MISS_ITOM / UNC_CHA_TOR_INSERTS.IO_ITOM )", + "BriefDescription": "Percentage of inbound full cacheline writes initiated by end device controllers that miss the L3 cache", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IO_MISS_ITOM / UNC_CHA_TOR_INSERTS.IO_ITOM)", "MetricGroup": "", "MetricName": "io_percent_of_inbound_full_writes_that_miss_l3", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.DSB_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.DSB_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_decoded_icache", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MITE_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MITE_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_legacy_decode_pipeline", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MS_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MS_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_microcode_sequencer", "ScaleUnit": "100%" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the local CPU socket.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_LOCAL * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the local CPU socket", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_LOCAL * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read_local", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from a remote CPU socket.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_REMOTE * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from a remote CPU socket", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_REMOTE * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read_remote", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the local CPU socket.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_ITOM_LOCAL + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_LOCAL ) * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the local CPU socket", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_ITOM_LOCAL + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_LOCAL) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write_local", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to a remote CPU socket.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_ITOM_REMOTE + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_REMOTE ) * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to a remote CPU socket", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_ITOM_REMOTE + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_REMOTE) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write_remote", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", + "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", + "MetricThreshold": "tma_bottleneck_mispredictions > 20", + "PublicDescription": "Total pipeline cost of Branch Misprediction related bottlenecks. Related metrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", + "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_big_code", + "MetricThreshold": "tma_bottleneck_big_code > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", + "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_ms ) ) ) ) - tma_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_instruction_fetch_bw", + "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) )", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", + "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", + "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks. Related metrics: tma_l3_hit_latency, tma_mem_latency.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", + "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs)", "MetricExpr": "100 * ( tma_memory_bound * ( tma_l1_bound / max( tma_memory_bound , ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) ) * ( tma_dtlb_load / max( tma_l1_bound , ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", + "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", + "PublicDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs). Related metrics: tma_dtlb_load, tma_dtlb_store.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", + "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors)", "MetricExpr": "100 * ( tma_memory_bound * ( ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) * tma_remote_cache / ( tma_local_mem + tma_remote_mem + tma_remote_cache ) + ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", - "MetricGroup": "BvMS;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", + "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", + "PublicDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears, tma_remote_cache.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", + "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation", "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( tma_core_bound * tma_amx_busy / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", + "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", "PublicDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", + "BriefDescription": "Total pipeline cost of irregular execution (e.g", "MetricExpr": "100 * ( ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_ms ) ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + RS.EMPTY_RESOURCE / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", - "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments).", + "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", + "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments). Related metrics: tma_microcode_sequencer, tma_ms_switches.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end.", + "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end", "MetricExpr": "100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", "MetricGroup": "BvOB;Cor;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_other_bottlenecks", + "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", "PublicDescription": "Total pipeline cost of remaining bottlenecks in the back-end. Examples include data-dependencies (Core Bound when Low ILP) and other unlisted memory-related stalls.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", + "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA", "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_branching_overhead", + "MetricThreshold": "tma_bottleneck_branching_overhead > 5", "PublicDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound).", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", + "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead", "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_useful_work", + "MetricThreshold": "tma_bottleneck_useful_work > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "topdown\\-fe\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_frontend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_frontend_bound > 0.15", "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound. Sample with: FRONTEND_RETIRED.LATENCY_GE_4.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( topdown\\-fetch\\-lat / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots )", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", "MetricName": "tma_fetch_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_16, FRONTEND_RETIRED.LATENCY_GE_8.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache", "MetricExpr": "max( 0 , tma_icache_misses - tma_code_l2_miss )", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_CODE_RD / tma_info_thread_clks", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_itlb_misses - tma_code_stlb_miss )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + tma_unknown_branches", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "( tma_branch_mispredicts / tma_bad_speculation ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "( 1 - ( tma_branch_mispredicts / tma_bad_speculation ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 3 ) * cpu@UOPS_RETIRED.MS\\,cmask\\=0x1\\,edge\\=0x1@ / ( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) / tma_info_thread_clks", - "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated", + "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: FRONTEND_RETIRED.MS_FLOWS." + "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: FRONTEND_RETIRED.MS_FLOWS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "DECODE.LCP / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." + "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", - "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS." + "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "max( 0 , tma_frontend_bound - tma_fetch_latency )", - "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", + "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2.", + "MetricThreshold": "tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.MITE_CYCLES_ANY - IDQ.MITE_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", "ScaleUnit": "100%", + "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder.", + "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder", "MetricExpr": "( cpu@INST_DECODED.DECODERS\\,cmask\\=0x1@ - cpu@INST_DECODED.DECODERS\\,cmask\\=0x2@ ) / tma_info_core_core_clks / 2", - "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated", + "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated;tma_issueD0", "MetricName": "tma_decoder0_alone", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder. Related metrics: tma_few_uops_instructions." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.DSB_CYCLES_ANY - IDQ.DSB_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", "ScaleUnit": "100%", + "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details", "MetricExpr": "max( IDQ.MS_CYCLES_ANY , cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ / ( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) ) / tma_info_core_core_clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_ms", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "max( 1 - ( tma_frontend_bound + tma_backend_bound + tma_retiring ) , 0 )", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_bad_speculation", "ScaleUnit": "100%", + "MetricThreshold": "tma_bad_speculation > 0.15", "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( topdown\\-br\\-mispredict / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS.", + "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", + "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types)", "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Slots", "MetricName": "tma_other_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "max( 0 , tma_bad_speculation - tma_branch_mispredicts )", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots", + "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT.", + "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches, tma_remote_cache.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering", "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_machine_clears_group;Slots", "MetricName": "tma_other_nukes", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "( topdown\\-be\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_backend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_backend_bound > 0.2", "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( topdown\\-mem\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", "MetricName": "tma_memory_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two).", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( EXE_ACTIVITY.BOUND_ON_LOADS - MEMORY_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", + "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT." + "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "min( ( 7 ) * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - MEMORY_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS." + "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", "MetricName": "tma_load_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", "MetricName": "tma_load_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", + "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache", "MetricExpr": "min( 2 * ( MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETIRED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS ) * 20 / 100 , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - MEMORY_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_l1_latency_dependency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( 16 * max( 0 , MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS.ALL_RFO ) + ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) * ( ( 10 ) * L2_RQSTS.RFO_HIT + ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) ) / tma_info_thread_clks", - "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks", + "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS." + "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "L1D_PEND_MISS.FB_FULL / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", + "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory)." + "MetricThreshold": "tma_fb_full > 0.3", + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L1D_MISS - MEMORY_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", + "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited)", "MetricExpr": "( 4.4 * tma_info_system_core_frequency ) * MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", "MetricName": "tma_l2_hit_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L2_MISS - MEMORY_ACTIVITY.STALLS_L3_MISS ) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( ( 81 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) + ( ( 79 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS." + "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( ( 79 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD + MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 1 - ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD." + "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( ( 37 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_RETIRED.L3_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT." + "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_bottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "( XQ.FULL_CYCLES + L1D_PEND_MISS.L2_STALLS ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clks )", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that)." + "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to memory bandwidth Allocation feature (RDT's memory bandwidth throttling).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to memory bandwidth Allocation feature (RDT's memory bandwidth throttling)", "MetricExpr": "INT_MISC.MBA_STALLS / tma_info_thread_clks", "MetricGroup": "MemoryBW;Offcore;Server;TopdownL5;tma_L5_group;tma_mem_bandwidth_group;Clocks", "MetricName": "tma_mba_stalls", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_mba_stalls > 0.1 & tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;Clocks", + "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that)." + "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory", "MetricExpr": "( ( 109 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_local_mem", "ScaleUnit": "100%", + "MetricThreshold": "tma_local_mem > 0.1 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory. Caching will improve the latency and increase performance. Sample with: MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory", "MetricExpr": "( ( 190 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_remote_mem", "ScaleUnit": "100%", + "MetricThreshold": "tma_remote_mem > 0.1 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues", "MetricExpr": "( ( ( 170 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM + ( ( 170 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", + "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_remote_cache", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD." + "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( MEM_STORE_RETIRED.L2_HIT * ( 10 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)." + "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( ( 170 * tma_info_system_core_frequency ) * cpu@OCR.DEMAND_RFO.L3_MISS\\,offcore_rsp\\=0x103b800002@ + ( 81 * tma_info_system_core_frequency ) * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM." + "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization", + "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores.", + "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores", "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread_clks", - "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSmSt", "MetricName": "tma_streaming_stores", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE." + "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma_fb_full." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( ( 7 ) * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_STORE_MISSES.WALK_ACTIVE ) / tma_info_core_core_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES." + "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_load." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", "MetricName": "tma_store_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", "MetricName": "tma_store_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "max( 0 , tma_backend_bound - tma_memory_bound )", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", "MetricName": "tma_core_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations).", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.DIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", "ScaleUnit": "100%", + "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIVIDER_ACTIVE." }, { - "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active", "MetricExpr": "ARITH.FPDIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_fp_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_int_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations", "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks + tma_c02_wait", - "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Clocks;tma_issueSO", "MetricName": "tma_serializing_operation", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD." + "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions", "MetricExpr": "CPU_CLK_UNHALTED.PAUSE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_slow_pause", "ScaleUnit": "100%", + "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.PAUSE_INST." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings).", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings)", "MetricExpr": "CPU_CLK_UNHALTED.C01 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_c01_wait", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_c01_wait > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings).", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings)", "MetricExpr": "CPU_CLK_UNHALTED.C02 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_c02_wait", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_c02_wait > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions", "MetricExpr": "13 * MISC2_RETIRED.LFENCE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_memory_fence", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_fence > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates fraction of cycles where the Advanced Matrix eXtensions (AMX) execution engine was busy with tile (arithmetic) operations.", + "BriefDescription": "This metric estimates fraction of cycles where the Advanced Matrix eXtensions (AMX) execution engine was busy with tile (arithmetic) operations", "MetricExpr": "EXE.AMX_BUSY / tma_info_core_core_clks", "MetricGroup": "BvCB;Compute;HPC;Server;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Core_Clocks", "MetricName": "tma_amx_busy", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_amx_busy > 0.5 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( tma_ports_utilized_0 * tma_info_thread_clks + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIV_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - EXE_ACTIVITY.BOUND_ON_LOADS ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( EXE_ACTIVITY.EXE_BOUND_0_PORTS + max( RS.EMPTY_RESOURCE - RESOURCE_STALLS.SCOREBOARD , 0 ) ) / tma_info_thread_clks * ( CYCLE_ACTIVITY.STALLS_TOTAL - EXE_ACTIVITY.BOUND_ON_LOADS ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", + "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles)", "MetricExpr": "160 * ASSISTS.SSE_AVX_MIX / tma_info_thread_clks", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks;tma_issueMV", "MetricName": "tma_mixing_vectors", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic." + "MetricThreshold": "tma_mixing_vectors > 0.05", + "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Sample with: UOPS_EXECUTED.CYCLES_GE_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + UOPS_DISPATCHED.PORT_5_11 + UOPS_DISPATCHED.PORT_6 ) / ( 5 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED.PORT_0 / tma_info_core_core_clks", - "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0." + "MetricThreshold": "tma_port_0 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_1 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1." + "MetricThreshold": "tma_port_1 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_6 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1." + "MetricThreshold": "tma_port_6 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "UOPS_DISPATCHED.PORT_2_3_10 / ( 3 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_load_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations. Sample with: UOPS_DISPATCHED.PORT_2_3_10." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_4_9 + UOPS_DISPATCHED.PORT_7_8 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations. Sample with: UOPS_DISPATCHED.PORT_7_8." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( topdown\\-retiring / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_retiring", "ScaleUnit": "100%", + "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.1", "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "max( 0 , tma_retiring - tma_heavy_operations )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", "MetricName": "tma_light_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", "ScaleUnit": "100%", + "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR + FP_ARITH_INST_RETIRED2.VECTOR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.128B_PACKED_HALF ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.256B_PACKED_HALF ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.512B_PACKED_HALF ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_512b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired)", "MetricExpr": "tma_int_vector_128b + tma_int_vector_256b", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_int_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_int_operations > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired). Vector/Matrix Int operations and shuffles are counted. Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain." }, { - "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", + "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired", "MetricExpr": "( INT_VEC_RETIRED.ADD_128 + INT_VEC_RETIRED.VNNI_128 ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops", + "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops;tma_issue2P", "MetricName": "tma_int_vector_128b", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_vector_128b > 0.1 & tma_int_operations > 0.1 & tma_light_operations > 0.6", + "PublicDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", + "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired", "MetricExpr": "( INT_VEC_RETIRED.ADD_256 + INT_VEC_RETIRED.MUL_256 + INT_VEC_RETIRED.VNNI_256 ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops", + "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops;tma_issue2P", "MetricName": "tma_int_vector_256b", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_vector_256b > 0.1 & tma_int_operations > 0.1 & tma_light_operations > 0.6", + "PublicDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses", "MetricExpr": "tma_light_operations * MEM_UOP_RETIRED.ANY / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_memory_operations", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_operations > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.MACRO_FUSED / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_fused_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Other_Light_Ops in MTL!)}." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused", "MetricExpr": "tma_light_operations * ( BR_INST_RETIRED.ALL_BRANCHES - INST_RETIRED.MACRO_FUSED ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_non_fused_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused. Non-conditional branches like direct JMP or CALL would count here. Can be used to examine fusible conditional jumps that were not fused." }, { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", + "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes", "MetricExpr": "max( 0 , tma_light_operations - ( tma_fp_arith + tma_int_operations + tma_memory_operations + tma_fused_instructions + tma_non_fused_branches ) )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_other_light_ops", "ScaleUnit": "100%", + "MetricThreshold": "tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_light_operations_group;Slots", "MetricName": "tma_nop_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer)", "MetricExpr": "tma_light_operations * INT_VEC_RETIRED.SHUFFLES / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Slots", "MetricName": "tma_shuffles_256b", "ScaleUnit": "100%", + "MetricThreshold": "tma_shuffles_256b > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer). Shuffles may incur slow cross \"vector lane\" data transfers." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "( topdown\\-heavy\\-ops / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", "MetricName": "tma_heavy_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+]). Sample with: UOPS_RETIRED.HEAVY.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops", "MetricExpr": "max( 0 , tma_heavy_operations - tma_microcode_sequencer )", - "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueD0", "MetricName": "tma_few_uops_instructions", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions." + "MetricThreshold": "tma_few_uops_instructions > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions. Related metrics: tma_decoder0_alone." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "UOPS_RETIRED.MS / tma_info_thread_slots", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: UOPS_RETIRED.MS." + "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: UOPS_RETIRED.MS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( ( 99 *3 + 63 + 30 ) / 5 ) * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults", "MetricExpr": "99 * ASSISTS.PAGE_FAULT / tma_info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_page_faults", "ScaleUnit": "100%", + "MetricThreshold": "tma_page_faults > 0.05", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults. A Page Fault may apply on first application access to a memory page. Note operating system handling of page faults accounts for the majority of its cost." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists", "MetricExpr": "30 * ASSISTS.FP / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_fp_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_assists > 0.1", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists", "MetricExpr": "63 * ASSISTS.SSE_AVX_MIX / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_avx_assists", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_avx_assists > 0.1" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", "ScaleUnit": "100%", + "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources. Sample with: FRONTEND_RETIRED.MS_FLOWS." }, { - "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts.", + "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts", "MetricExpr": "100 * ( 1 - tma_core_bound / tma_ports_utilization if tma_core_bound < tma_ports_utilization else 1 ) if tma_info_system_smt_2t_utilization > 0.5 else 0", "MetricGroup": "Cor;SMT;Metric", "MetricName": "tma_info_botlnk_l0_core_bound_likely", "MetricThreshold": "tma_info_botlnk_core_bound_likely > 0.5" }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", "MetricName": "tma_info_thread_uoppi", "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", "MetricName": "tma_info_thread_uptb", "MetricThreshold": "tma_info_thread_uptb < 6 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor.", + "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor", "MetricExpr": "tma_info_thread_slots / ( TOPDOWN.SLOTS / 2 ) if #SMT_on else 1", "MetricGroup": "SMT;TmaL1;TopdownL1;tma_L1_group;Metric", "MetricName": "tma_info_thread_slots_utilization" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR_HALF ) + 2 * ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED2.COMPLEX_SCALAR_HALF ) + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * ( FP_ARITH_INST_RETIRED2.128B_PACKED_HALF + FP_ARITH_INST_RETIRED.8_FLOPS ) + 16 * ( FP_ARITH_INST_RETIRED2.256B_PACKED_HALF + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) + 32 * FP_ARITH_INST_RETIRED2.512B_PACKED_HALF ) ) / tma_info_core_core_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( FP_ARITH_DISPATCHED.PORT_0 + FP_ARITH_DISPATCHED.PORT_1 + FP_ARITH_DISPATCHED.PORT_5 ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "uops Executed per Cycle.", + "BriefDescription": "uops Executed per Cycle", "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", "MetricGroup": "Power;Metric", "MetricName": "tma_info_core_epc" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipload", "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipstore", "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipbranch", "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", "MetricName": "tma_info_inst_mix_ipcall", "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", "MetricThreshold": "tma_info_inst_mix_iptb < 6 * 2 + 1", - "PublicDescription": "Instructions per taken branch. Related metrics: tma_info_frontend_dsb_coverage, tma_info_botlnk_l2_dsb_misses, tma_info_botlnk_l2_dsb_bandwidth." + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR_HALF ) + 2 * ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED2.COMPLEX_SCALAR_HALF ) + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * ( FP_ARITH_INST_RETIRED2.128B_PACKED_HALF + FP_ARITH_INST_RETIRED.8_FLOPS ) + 16 * ( FP_ARITH_INST_RETIRED2.256B_PACKED_HALF + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) + 32 * FP_ARITH_INST_RETIRED2.512B_PACKED_HALF ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipflop", "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR + FP_ARITH_INST_RETIRED2.VECTOR ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", @@ -1462,7 +1597,7 @@ "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Half-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Half-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED2.SCALAR", "MetricGroup": "Flops;FpScalar;InsType;Server;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_hp", @@ -1470,7 +1605,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Half-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", @@ -1478,7 +1613,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", @@ -1486,7 +1621,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.128B_PACKED_HALF )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", @@ -1494,7 +1629,7 @@ "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.256B_PACKED_HALF )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", @@ -1502,7 +1637,7 @@ "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.512B_PACKED_HALF )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx512", @@ -1510,40 +1645,40 @@ "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / CPU_CLK_UNHALTED.PAUSE_INST", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ippause" }, { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / SW_PREFETCH_ACCESS.ANY", "MetricGroup": "Prefetches;Inst_Metric", "MetricName": "tma_info_inst_mix_ipswpf", "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions.", + "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions", "MetricExpr": "INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "MicroSeq;Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_strings_cycles", "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1" }, { - "BriefDescription": "Instructions per a microcode Assist invocation.", + "BriefDescription": "Instructions per a microcode Assist invocation", "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", "MetricName": "tma_info_pipeline_ipassist", @@ -1551,105 +1686,105 @@ "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core.", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core", "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@ )", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", + "BriefDescription": "Average number of uops fetched from DSB per cycle", "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_dsb" }, { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", + "BriefDescription": "Average number of uops fetched from MITE per cycle", "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_mite" }, { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", + "BriefDescription": "Average number of Uops issued by front-end when it issued something", "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_frontend_fetch_upc" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( UOPS_ISSUED.ANY )", "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", "MetricName": "tma_info_frontend_dsb_coverage", "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 6 > 0.35", - "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_info_inst_mix_iptb, tma_info_botlnk_l2_dsb_misses, tma_info_botlnk_l2_dsb_bandwidth." + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection.", + "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection", "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / cpu@INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_unknown_branch_cost", "PublicDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection. See Unknown_Branches node." }, { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", + "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "DSBmiss;Metric", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Average Latency for L1 instruction cache misses.", + "BriefDescription": "Average Latency for L1 instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / cpu@ICACHE_DATA.STALLS\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "Fed;FetchLat;IcMiss;Metric", "MetricName": "tma_info_frontend_icache_miss_latency" }, { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", "MetricGroup": "DSBmiss;Fed;Inst_Metric", "MetricName": "tma_info_frontend_ipdsb_miss_ret", "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache true code cacheline misses per kilo instruction", "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code" }, { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction", "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code_all" }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb + tma_ms ) )", "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_dsb_misses > 10", - "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_info_inst_mix_iptb, tma_info_frontend_dsb_coverage, tma_info_botlnk_l2_dsb_bandwidth." + "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb + tma_ms ) ) )", "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_dsb_bandwidth > 10", - "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_info_inst_mix_iptb, tma_info_frontend_dsb_coverage, tma_info_botlnk_l2_dsb_misses." + "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", + "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", @@ -1657,462 +1792,463 @@ "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmispredict", "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_ret", "MetricThreshold": "tma_info_bad_spec_ipmisp_ret < 500" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_indirect", "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 6 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", "MetricName": "tma_info_bad_spec_branch_misprediction_cost", - "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)." + "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_mispredicts_resteers." }, { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", + "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", "MetricGroup": "BrMispredicts;Metric", "MetricName": "tma_info_bad_spec_spec_clears_ratio" }, { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", + "BriefDescription": "Fraction of branches that are non-taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_nt" }, { - "BriefDescription": "Fraction of branches that are taken conditionals.", + "BriefDescription": "Fraction of branches that are taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_tk" }, { - "BriefDescription": "Fraction of branches that are CALL or RET.", + "BriefDescription": "Fraction of branches that are CALL or RET", "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_callret" }, { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", + "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_jump" }, { - "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group).", + "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group)", "MetricExpr": "1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_other_branches" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / MEM_LOAD_COMPLETED.L1_MISS_ANY", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki_load" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * L2_RQSTS.RFO_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_all" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", + "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_fb_hpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Offcore;Metric", "MetricName": "tma_info_memory_l3_cache_access_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", "MetricName": "tma_info_memory_tlb_page_walks_utilization", "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Fed;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_code_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_load_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_access_bw", "MetricGroup": "Mem;MemoryBW;Offcore;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_access_bw_2t" }, { - "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory).", + "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory)", "MetricExpr": "1000 * L2_LINES_OUT.SILENT / tma_info_inst_mix_instructions", "MetricGroup": "L2Evicts;Mem;Server;Core_Metric", "MetricName": "tma_info_memory_core_l2_evictions_silent_pki" }, { - "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction.", + "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction", "MetricExpr": "1000 * L2_LINES_OUT.NON_SILENT / tma_info_inst_mix_instructions", "MetricGroup": "L2Evicts;Mem;Server;Core_Metric", "MetricName": "tma_info_memory_core_l2_evictions_nonsilent_pki" }, { - "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses.", + "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses", "MetricExpr": "L2_LINES_OUT.USELESS_HWPF / ( L2_LINES_OUT.SILENT + L2_LINES_OUT.NON_SILENT )", "MetricGroup": "Prefetches;Metric", - "MetricName": "tma_info_memory_prefetches_useless_hwpf" + "MetricName": "tma_info_memory_prefetches_useless_hwpf", + "MetricThreshold": "tma_info_memory_useless_hwpf > 0.15" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Latency for L3 cache miss demand Loads.", + "BriefDescription": "Average Latency for L3 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.L3_MISS_DEMAND_DATA_RD / OFFCORE_REQUESTS.L3_MISS_DEMAND_DATA_RD", "MetricGroup": "Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l3_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=0x1@", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Off-core accesses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches).", + "BriefDescription": "Off-core accesses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches)", "MetricExpr": "1000 * OCR.READS_TO_CORE.ANY_RESPONSE / tma_info_inst_mix_instructions", "MetricGroup": "CacheHits;Offcore;Metric", "MetricName": "tma_info_memory_mix_offcore_read_any_pki" }, { - "BriefDescription": "L3 cache misses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches).", + "BriefDescription": "L3 cache misses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches)", "MetricExpr": "1000 * OCR.READS_TO_CORE.L3_MISS / tma_info_inst_mix_instructions", "MetricGroup": "Offcore;Metric", "MetricName": "tma_info_memory_mix_offcore_read_l3m_pki" }, { - "BriefDescription": "Off-core accesses per kilo instruction for modified write requests.", + "BriefDescription": "Off-core accesses per kilo instruction for modified write requests", "MetricExpr": "1000 * OCR.MODIFIED_WRITE.ANY_RESPONSE / tma_info_inst_mix_instructions", "MetricGroup": "Offcore;Metric", "MetricName": "tma_info_memory_mix_offcore_mwrite_any_pki" }, { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", + "BriefDescription": "Un-cacheable retired load per kilo instruction", "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_uc_load_pki" }, { - "BriefDescription": "\"Bus lock\" per kilo instruction.", + "BriefDescription": "\"Bus lock\" per kilo instruction", "MetricExpr": "1000 * SQ_MISC.BUS_LOCK / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_bus_lock_pki" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz].", + "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz]", "MetricExpr": "tma_info_system_socket_clks / 1e9 / tma_info_system_time", "MetricGroup": "SoC;System_Metric", "MetricName": "tma_info_system_uncore_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR_HALF ) + 2 * ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED2.COMPLEX_SCALAR_HALF ) + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * ( FP_ARITH_INST_RETIRED2.128B_PACKED_HALF + FP_ARITH_INST_RETIRED.8_FLOPS ) + 16 * ( FP_ARITH_INST_RETIRED2.256B_PACKED_HALF + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) + 32 * FP_ARITH_INST_RETIRED2.512B_PACKED_HALF ) ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_utilization", "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states.", + "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states", "MetricExpr": "CPU_CLK_UNHALTED.C0_WAIT / tma_info_thread_clks", "MetricGroup": "C0Wait;Metric", "MetricName": "tma_info_system_c0_wait", "MetricThreshold": "tma_info_system_c0_wait > 0.05" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "( 64 * ( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", "MetricName": "tma_info_system_dram_bw_use", - "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]." + "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Average latency of data read request to external memory (in nanoseconds).", + "BriefDescription": "Average latency of data read request to external memory (in nanoseconds)", "MetricExpr": "( 1000000000 ) * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( tma_info_system_socket_clks / tma_info_system_time )", "MetricGroup": "Mem;MemoryLat;SoC;NanoSeconds", "MetricName": "tma_info_system_mem_read_latency", "PublicDescription": "Average latency of data read request to external memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetches. ([RKL+]memory-controller only)." }, { - "BriefDescription": "Average number of parallel data read requests to external memory.", + "BriefDescription": "Average number of parallel data read requests to external memory", "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD\\,thresh\\=0x1@", "MetricGroup": "Mem;MemoryBW;SoC;System_Metric", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches." }, { - "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds].", + "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]", "MetricExpr": "( 1000000000 ) * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR ) / cha_0@event\\=0x0@", "MetricGroup": "MemOffcore;MemoryLat;SoC;Server;NanoSeconds", "MetricName": "tma_info_system_mem_dram_read_latency", "PublicDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]. Accounts for demand loads and L1/L2 data-read prefetches." }, { - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec].", + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]", "MetricExpr": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "IoBW;MemOffcore;SoC;Server;GB/sec", "MetricName": "tma_info_system_io_read_bw", "PublicDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]. Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU." }, { - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec].", + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]", "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_ITOM + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR ) * 64 / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "IoBW;MemOffcore;SoC;Server;GB/sec", "MetricName": "tma_info_system_io_write_bw", "PublicDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]. Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU." }, { - "BriefDescription": "Cross-socket Ultra Path Interconnect (UPI) data transmit bandwidth for data only [MB / sec].", - "MetricExpr": "UNC_UPI_TxL_FLITS.ALL_DATA * 64 / 9 / 1000000", + "BriefDescription": "Cross-socket Ultra Path Interconnect (UPI) data transmit bandwidth for data only [MB / sec]", + "MetricExpr": "UNC_UPI_TXL_FLITS.ALL_DATA * 64 / 9 / 1000000", "MetricGroup": "SoC;Server;MB/sec", "MetricName": "tma_info_system_upi_data_transmit_bw" }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "( power@energy\\-pkg@ * ( 61 ) + 15.6 * power@energy\\-ram@ ) / ( ( duration_time ) * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", "MetricName": "tma_info_system_time", "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", "MetricName": "tma_info_system_mux", "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "cha_0@event\\=0x0@", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", "MetricName": "tma_info_system_ipfarbranch", "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" }, { - "BriefDescription": "Average Off-core access BW for Reads-to-Core (R2C).", + "BriefDescription": "Average Off-core access BW for Reads-to-Core (R2C)", "MetricExpr": "64 * OCR.READS_TO_CORE.ANY_RESPONSE / 1e9 / tma_info_system_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC;GB/sec", "MetricName": "tma_info_memory_soc_r2c_offcore_bw", "PublicDescription": "Average Off-core access BW for Reads-to-Core (R2C). R2C account for demand or prefetch load/RFO/code access that fill data into the Core caches." }, { - "BriefDescription": "Average L3-cache miss BW for Reads-to-Core (R2C).", + "BriefDescription": "Average L3-cache miss BW for Reads-to-Core (R2C)", "MetricExpr": "64 * OCR.READS_TO_CORE.L3_MISS / 1e9 / tma_info_system_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC;GB/sec", "MetricName": "tma_info_memory_soc_r2c_l3m_bw", "PublicDescription": "Average L3-cache miss BW for Reads-to-Core (R2C). This covering going to DRAM or other memory off-chip memory tears. See R2C_Offcore_BW." }, { - "BriefDescription": "Average DRAM BW for Reads-to-Core (R2C) covering for memory attached to local- and remote-socket.", + "BriefDescription": "Average DRAM BW for Reads-to-Core (R2C) covering for memory attached to local- and remote-socket", "MetricExpr": "64 * OCR.READS_TO_CORE.DRAM / 1e9 / tma_info_system_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC;GB/sec", "MetricName": "tma_info_memory_soc_r2c_dram_bw", diff --git a/GNR/metrics/graniterapids_metrics.json b/GNR/metrics/graniterapids_metrics.json index 5ea72048..93d5cefc 100644 --- a/GNR/metrics/graniterapids_metrics.json +++ b/GNR/metrics/graniterapids_metrics.json @@ -2,7 +2,7 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for Intel(R) Xeon(R) 6 Processor with P-cores0", - "DatePublished": "11/08/2024", + "DatePublished": "11/15/2024", "Version": "1.0", "Legend": "", "TmaVersion": "5.01", @@ -11629,6 +11629,11 @@ "BaseFormula": " uops_executed.thread / ( ( uops_executed.core_cycles_ge_1 / 2 ) if smt_on else uops_executed.thread:c1 )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "LocateWith": "" diff --git a/GNR/metrics/perf/graniterapids_metrics_perf.json b/GNR/metrics/perf/graniterapids_metrics_perf.json index 5b9eb475..5b020ab1 100644 --- a/GNR/metrics/perf/graniterapids_metrics_perf.json +++ b/GNR/metrics/perf/graniterapids_metrics_perf.json @@ -1,181 +1,181 @@ [ { - "BriefDescription": "CPU operating frequency (in GHz).", - "MetricExpr": "( CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000", + "BriefDescription": "CPU operating frequency (in GHz)", + "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000", "MetricGroup": "", "MetricName": "cpu_operating_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Uncore operating frequency in GHz.", - "MetricExpr": "( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) / 1000000000) / duration_time", + "BriefDescription": "Uncore operating frequency in GHz", + "MetricExpr": "(UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) / 1000000000) / duration_time", "MetricGroup": "", "MetricName": "uncore_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Percentage of time spent in the active CPU power state C0.", + "BriefDescription": "Percentage of time spent in the active CPU power state C0", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "", "MetricName": "cpu_utilization", "ScaleUnit": "100%" }, { - "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles.", + "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles", "MetricExpr": "CPU_CLK_UNHALTED.THREAD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L1D.REPLACEMENT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L2_LINES_IN.ALL / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL ) / ( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE )", + "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL) / (UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_local_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE ) / ( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE )", + "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE) / (UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_remote_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "DDR memory read bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_CAS_COUNT_SCH0.RD + UNC_M_CAS_COUNT_SCH1.RD ) * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory read bandwidth (MB/sec)", + "MetricExpr": "((UNC_M_CAS_COUNT_SCH0.RD + UNC_M_CAS_COUNT_SCH1.RD) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory write bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_CAS_COUNT_SCH0.WR + UNC_M_CAS_COUNT_SCH1.WR ) * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory write bandwidth (MB/sec)", + "MetricExpr": "((UNC_M_CAS_COUNT_SCH0.WR + UNC_M_CAS_COUNT_SCH1.WR) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_CAS_COUNT_SCH0.RD + UNC_M_CAS_COUNT_SCH1.RD + UNC_M_CAS_COUNT_SCH0.WR + UNC_M_CAS_COUNT_SCH1.WR ) * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory bandwidth (MB/sec)", + "MetricExpr": "((UNC_M_CAS_COUNT_SCH0.RD + UNC_M_CAS_COUNT_SCH1.RD + UNC_M_CAS_COUNT_SCH0.WR + UNC_M_CAS_COUNT_SCH1.WR) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_total", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data transmit bandwidth (MB/sec).", - "MetricExpr": "( UNC_UPI_TxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", + "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data transmit bandwidth (MB/sec)", + "MetricExpr": "(UNC_UPI_TxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", "MetricGroup": "", "MetricName": "upi_data_transmit_bw", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions", "MetricExpr": "MEM_INST_RETIRED.ALL_LOADS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "loads_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions", "MetricExpr": "MEM_INST_RETIRED.ALL_STORES / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "stores_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L1_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions", "MetricExpr": "L2_RQSTS.ALL_CODE_RD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1_i_code_read_misses_with_prefetches_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions", "MetricExpr": "L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_code_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_LLCPREFDATA + UNC_CHA_TOR_INSERTS.IA_MISS_DRD + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF ) / INST_RETIRED.ANY", + "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IA_MISS_LLCPREFDATA + UNC_CHA_TOR_INSERTS.IA_MISS_DRD + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF) / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_data_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", "MetricExpr": "UNC_CHA_TOR_INSERTS.IA_MISS_CRD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_code_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_latency", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to local memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to local memory in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_latency_for_local_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to remote memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to remote memory in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_latency_for_remote_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to DRAM in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to DRAM in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_to_dram_latency", "ScaleUnit": "1ns" }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_2nd_level_mpi", @@ -183,7 +183,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the ITLB (Instruction TLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions", "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_2nd_level_load_mpi", @@ -191,7 +191,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions", "MetricExpr": "DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_2nd_level_store_mpi", @@ -199,131 +199,131 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_ITOM + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR ) * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_ITOM + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.DSB_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.DSB_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_decoded_icache", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MITE_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MITE_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_legacy_decode_pipeline", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MS_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MS_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_microcode_sequencer", "ScaleUnit": "100%" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the local CPU socket.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_LOCAL * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the local CPU socket", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_LOCAL * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read_local", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from a remote CPU socket.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_REMOTE * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from a remote CPU socket", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_REMOTE * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read_remote", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the local CPU socket.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_ITOM_LOCAL + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_LOCAL ) * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the local CPU socket", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_ITOM_LOCAL + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_LOCAL) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write_local", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to a remote CPU socket.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_ITOM_REMOTE + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_REMOTE ) * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to a remote CPU socket", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_ITOM_REMOTE + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_REMOTE) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write_remote", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to local memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.READS_LOCAL * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to local memory", + "MetricExpr": "(UNC_CHA_REQUESTS.READS_LOCAL * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_local_memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to local memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.WRITES_LOCAL * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to local memory", + "MetricExpr": "(UNC_CHA_REQUESTS.WRITES_LOCAL * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_local_memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to remote memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.READS_REMOTE * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to remote memory", + "MetricExpr": "(UNC_CHA_REQUESTS.READS_REMOTE * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_remote_memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to remote memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.WRITES_REMOTE * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to remote memory", + "MetricExpr": "(UNC_CHA_REQUESTS.WRITES_REMOTE * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_remote_memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data receive bandwidth (MB/sec).", - "MetricExpr": "( UNC_UPI_RxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", + "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data receive bandwidth (MB/sec)", + "MetricExpr": "(UNC_UPI_RxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", "MetricGroup": "", "MetricName": "upi_data_receive_bw", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth observed by the integrated I/O traffic controller (IIO) of IO reads that are initiated by end device controllers that are requesting memory from the CPU.", - "MetricExpr": "( UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.ALL_PARTS * 4 / 1000000) / duration_time", + "BriefDescription": "Bandwidth observed by the integrated I/O traffic controller (IIO) of IO reads that are initiated by end device controllers that are requesting memory from the CPU", + "MetricExpr": "(UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.ALL_PARTS * 4 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "iio_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth observed by the integrated I/O traffic controller (IIO) of IO writes that are initiated by end device controllers that are writing memory to the CPU.", - "MetricExpr": "( UNC_IIO_DATA_REQ_OF_CPU.MEM_WRITE.ALL_PARTS * 4 / 1000000) / duration_time", + "BriefDescription": "Bandwidth observed by the integrated I/O traffic controller (IIO) of IO writes that are initiated by end device controllers that are writing memory to the CPU", + "MetricExpr": "(UNC_IIO_DATA_REQ_OF_CPU.MEM_WRITE.ALL_PARTS * 4 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "iio_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Percent of time that cores are in cstate C0 as observed by the power control unit (PCU).", - "MetricExpr": "100 * (( UNC_P_POWER_STATE_OCCUPANCY_CORES_C0 / (UNC_P_CLOCKTICKS[0] / #num_packages ) ) / ( #num_cores / #num_packages * #num_packages ) )", + "BriefDescription": "Percent of time that cores are in cstate C0 as observed by the power control unit (PCU)", + "MetricExpr": "100 * ((UNC_P_POWER_STATE_OCCUPANCY_CORES_C0 / (UNC_P_CLOCKTICKS[0] / #num_packages ) ) / ( #num_cores / #num_packages * #num_packages ) )", "MetricGroup": "", "MetricName": "cpu_cstate_c0" }, { - "BriefDescription": "Percent of time that cores are in cstate C6 as observed by the power control unit (PCU).", - "MetricExpr": "100 * (( UNC_P_POWER_STATE_OCCUPANCY_CORES_C6 / (UNC_P_CLOCKTICKS[0] / #num_packages ) ) / ( #num_cores / #num_packages * #num_packages ) )", + "BriefDescription": "Percent of time that cores are in cstate C6 as observed by the power control unit (PCU)", + "MetricExpr": "100 * ((UNC_P_POWER_STATE_OCCUPANCY_CORES_C6 / (UNC_P_CLOCKTICKS[0] / #num_packages ) ) / ( #num_cores / #num_packages * #num_packages ) )", "MetricGroup": "", "MetricName": "cpu_cstate_c6" }, { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", + "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", @@ -333,7 +333,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", + "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_big_code", @@ -342,7 +342,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", + "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_ms ) ) ) ) - tma_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_instruction_fetch_bw", @@ -351,7 +351,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) )", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", @@ -361,7 +361,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", @@ -371,7 +371,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", + "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs)", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_load / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", @@ -381,7 +381,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", + "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors)", "MetricExpr": "100 * ( tma_memory_bound * ( ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) * tma_remote_cache / ( tma_local_mem + tma_remote_mem + tma_remote_cache ) + ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", @@ -391,7 +391,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", + "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation", "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( tma_core_bound * tma_amx_busy / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", @@ -401,7 +401,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", + "BriefDescription": "Total pipeline cost of irregular execution (e.g", "MetricExpr": "100 * ( ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_ms ) ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + RS.EMPTY_RESOURCE / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", @@ -411,7 +411,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end.", + "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end", "MetricExpr": "100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", "MetricGroup": "BvOB;Cor;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_other_bottlenecks", @@ -421,7 +421,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", + "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA", "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_branching_overhead", @@ -431,7 +431,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", + "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead", "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_useful_work", @@ -440,7 +440,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "topdown\\-fe\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_frontend_bound", @@ -451,7 +451,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( topdown\\-fetch\\-lat / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots )", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", "MetricName": "tma_fetch_latency", @@ -462,7 +462,7 @@ "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", @@ -471,7 +471,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache", "MetricExpr": "max( 0 , ( FRONTEND_RETIRED.L1I_MISS * FRONTEND_RETIRED.L1I_MISS:R ) / tma_info_thread_clks - tma_code_l2_miss )", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_hit", @@ -479,7 +479,7 @@ "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache", "MetricExpr": "( FRONTEND_RETIRED.L2_MISS * FRONTEND_RETIRED.L2_MISS:R ) / tma_info_thread_clks", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_miss", @@ -487,7 +487,7 @@ "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", @@ -496,7 +496,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , ( FRONTEND_RETIRED.ITLB_MISS * FRONTEND_RETIRED.ITLB_MISS:R ) / tma_info_thread_clks - tma_code_stlb_miss )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_hit", @@ -504,7 +504,7 @@ "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", "MetricExpr": "( FRONTEND_RETIRED.STLB_MISS * FRONTEND_RETIRED.STLB_MISS:R ) / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_miss", @@ -512,7 +512,7 @@ "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_4k", @@ -520,7 +520,7 @@ "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_2m", @@ -528,7 +528,7 @@ "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", @@ -537,7 +537,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "( tma_branch_mispredicts / tma_bad_speculation ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", @@ -546,7 +546,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "( 1 - ( tma_branch_mispredicts / tma_bad_speculation ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", @@ -555,7 +555,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", @@ -564,7 +564,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 3 ) * cpu@UOPS_RETIRED.MS\\,cmask\\=0x1\\,edge\\=0x1@ / ( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", @@ -573,7 +573,7 @@ "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "DECODE.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", @@ -582,7 +582,7 @@ "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", @@ -591,7 +591,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "max( 0 , tma_frontend_bound - tma_fetch_latency )", "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", @@ -602,7 +602,7 @@ "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.MITE_CYCLES_ANY - IDQ.MITE_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", @@ -611,7 +611,7 @@ "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder.", + "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder", "MetricExpr": "( cpu@INST_DECODED.DECODERS\\,cmask\\=0x1@ - cpu@INST_DECODED.DECODERS\\,cmask\\=0x2@ ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated;tma_issueD0", "MetricName": "tma_decoder0_alone", @@ -620,7 +620,7 @@ "PublicDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder. Related metrics: tma_few_uops_instructions." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.DSB_CYCLES_ANY - IDQ.DSB_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", @@ -629,7 +629,7 @@ "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details", "MetricExpr": "max( IDQ.MS_CYCLES_ANY , cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ / ( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) ) / tma_info_core_core_clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_ms", @@ -637,7 +637,7 @@ "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "max( 1 - ( tma_frontend_bound + tma_backend_bound + tma_retiring ) , 0 )", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_bad_speculation", @@ -648,7 +648,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( topdown\\-br\\-mispredict / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", @@ -659,7 +659,7 @@ "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by non-taken conditional branches.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by non-taken conditional branches", "MetricExpr": "( BR_MISP_RETIRED.COND_NTAKEN_COST * BR_MISP_RETIRED.COND_NTAKEN_COST:R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_cond_nt_mispredicts", @@ -667,7 +667,7 @@ "MetricThreshold": "tma_cond_nt_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to misprediction by taken conditional branches.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to misprediction by taken conditional branches", "MetricExpr": "( BR_MISP_RETIRED.COND_TAKEN_COST * BR_MISP_RETIRED.COND_TAKEN_COST:R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_cond_tk_mispredicts", @@ -675,7 +675,7 @@ "MetricThreshold": "tma_cond_tk_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect CALL instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect CALL instructions", "MetricExpr": "( BR_MISP_RETIRED.INDIRECT_CALL_COST * BR_MISP_RETIRED.INDIRECT_CALL_COST:R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_ind_call_mispredicts", @@ -683,7 +683,7 @@ "MetricThreshold": "tma_ind_call_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect JMP instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect JMP instructions", "MetricExpr": "max( ( ( BR_MISP_RETIRED.INDIRECT_COST * BR_MISP_RETIRED.INDIRECT_COST:R ) - ( BR_MISP_RETIRED.INDIRECT_CALL_COST * BR_MISP_RETIRED.INDIRECT_CALL_COST:R ) ) / tma_info_thread_clks , 0 )", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_ind_jump_mispredicts", @@ -691,7 +691,7 @@ "MetricThreshold": "tma_ind_jump_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by (indirect) RET instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by (indirect) RET instructions", "MetricExpr": "( BR_MISP_RETIRED.RET_COST * BR_MISP_RETIRED.RET_COST:R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_ret_mispredicts", @@ -699,7 +699,7 @@ "MetricThreshold": "tma_ret_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", + "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types)", "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Slots", "MetricName": "tma_other_mispredicts", @@ -707,7 +707,7 @@ "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "max( 0 , tma_bad_speculation - tma_branch_mispredicts )", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", @@ -718,7 +718,7 @@ "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering", "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_machine_clears_group;Slots", "MetricName": "tma_other_nukes", @@ -726,7 +726,7 @@ "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "( topdown\\-be\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_backend_bound", @@ -737,7 +737,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( topdown\\-mem\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", "MetricName": "tma_memory_bound", @@ -748,7 +748,7 @@ "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( EXE_ACTIVITY.BOUND_ON_LOADS - MEMORY_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", @@ -757,7 +757,7 @@ "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "( min( ( MEM_INST_RETIRED.STLB_HIT_LOADS * MEM_INST_RETIRED.STLB_HIT_LOADS:R ) , MEM_INST_RETIRED.STLB_HIT_LOADS * ( 7 ) ) if ( MEM_INST_RETIRED.STLB_HIT_LOADS:R >= 0 ) else ( MEM_INST_RETIRED.STLB_HIT_LOADS * ( 7 ) ) ) / tma_info_thread_clks + tma_load_stlb_miss", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", @@ -766,7 +766,7 @@ "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_dtlb_load - tma_load_stlb_miss )", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", "MetricName": "tma_load_stlb_hit", @@ -774,7 +774,7 @@ "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", "MetricName": "tma_load_stlb_miss", @@ -782,7 +782,7 @@ "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_4k", @@ -790,7 +790,7 @@ "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_2m", @@ -798,7 +798,7 @@ "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_1g", @@ -806,7 +806,7 @@ "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", @@ -815,7 +815,7 @@ "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", + "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache", "MetricExpr": "min( 2 * ( MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETIRED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS ) * 20 / 100 , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - MEMORY_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_l1_latency_dependency", @@ -824,7 +824,7 @@ "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( MEM_INST_RETIRED.LOCK_LOADS * MEM_INST_RETIRED.LOCK_LOADS:R ) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", @@ -833,7 +833,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "( min( ( MEM_INST_RETIRED.SPLIT_LOADS * MEM_INST_RETIRED.SPLIT_LOADS:R ) , MEM_INST_RETIRED.SPLIT_LOADS * tma_info_memory_load_miss_real_latency ) if ( MEM_INST_RETIRED.SPLIT_LOADS:R >= 0 ) else ( MEM_INST_RETIRED.SPLIT_LOADS * tma_info_memory_load_miss_real_latency ) ) / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", @@ -842,7 +842,7 @@ "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "L1D_PEND_MISS.FB_FULL / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", @@ -851,7 +851,7 @@ "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L1D_MISS - MEMORY_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", @@ -860,7 +860,7 @@ "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", + "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited)", "MetricExpr": "( min( ( MEM_LOAD_RETIRED.L2_HIT * MEM_LOAD_RETIRED.L2_HIT:R ) , MEM_LOAD_RETIRED.L2_HIT * ( 4.4 * tma_info_system_core_frequency ) ) if ( MEM_LOAD_RETIRED.L2_HIT:R >= 0 ) else ( MEM_LOAD_RETIRED.L2_HIT * ( 4.4 * tma_info_system_core_frequency ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", "MetricName": "tma_l2_hit_latency", @@ -869,7 +869,7 @@ "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L2_MISS - MEMORY_ACTIVITY.STALLS_L3_MISS ) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", @@ -878,7 +878,7 @@ "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS:R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * ( 79 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) if ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS:R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * ( 79 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) ) + ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD:R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 81 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) if ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD:R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 81 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) ) * ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", @@ -887,7 +887,7 @@ "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD:R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * ( 79 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) if ( MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD:R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * ( 79 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) ) + ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD:R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 79 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) if ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD:R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 79 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) ) * ( 1 - ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", @@ -896,7 +896,7 @@ "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( min( ( MEM_LOAD_RETIRED.L3_HIT * MEM_LOAD_RETIRED.L3_HIT:R ) , MEM_LOAD_RETIRED.L3_HIT * ( 37 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) if ( MEM_LOAD_RETIRED.L3_HIT:R >= 0 ) else ( MEM_LOAD_RETIRED.L3_HIT * ( 37 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", @@ -905,7 +905,7 @@ "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_bottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "( XQ.FULL_CYCLES + L1D_PEND_MISS.L2_STALLS ) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", @@ -914,7 +914,7 @@ "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clks )", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", @@ -923,7 +923,7 @@ "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", @@ -932,7 +932,7 @@ "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to memory bandwidth Allocation feature (RDT's memory bandwidth throttling).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to memory bandwidth Allocation feature (RDT's memory bandwidth throttling)", "MetricExpr": "INT_MISC.MBA_STALLS / tma_info_thread_clks", "MetricGroup": "MemoryBW;Offcore;Server;TopdownL5;tma_L5_group;tma_mem_bandwidth_group;Clocks", "MetricName": "tma_mba_stalls", @@ -940,7 +940,7 @@ "MetricThreshold": "tma_mba_stalls > 0.1 & tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", @@ -949,7 +949,7 @@ "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory", "MetricExpr": "( MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM * MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM:R ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_local_mem", @@ -958,7 +958,7 @@ "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory. Caching will improve the latency and increase performance. Sample with: MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory", "MetricExpr": "( MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM * MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM:R ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_remote_mem", @@ -967,7 +967,7 @@ "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues", "MetricExpr": "( ( MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM * MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM:R ) + ( MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD * MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD:R ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_remote_cache", @@ -976,7 +976,7 @@ "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", @@ -985,7 +985,7 @@ "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( MEM_STORE_RETIRED.L2_HIT * ( 10 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", @@ -994,7 +994,7 @@ "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( ( 170 * tma_info_system_core_frequency ) * cpu@OCR.DEMAND_RFO.L3_MISS\\,offcore_rsp\\=0x103b800002@ + ( 81 * tma_info_system_core_frequency ) * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM ) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", @@ -1003,7 +1003,7 @@ "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "( min( ( MEM_INST_RETIRED.SPLIT_STORES * MEM_INST_RETIRED.SPLIT_STORES:R ) , MEM_INST_RETIRED.SPLIT_STORES * 1 ) if ( MEM_INST_RETIRED.SPLIT_STORES:R >= 0 ) else ( MEM_INST_RETIRED.SPLIT_STORES * 1 ) ) / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", @@ -1012,7 +1012,7 @@ "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores.", + "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores", "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread_clks", "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSmSt", "MetricName": "tma_streaming_stores", @@ -1021,7 +1021,7 @@ "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma_fb_full." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( min( ( MEM_INST_RETIRED.STLB_HIT_STORES * MEM_INST_RETIRED.STLB_HIT_STORES:R ) , MEM_INST_RETIRED.STLB_HIT_STORES * ( 7 ) ) if ( MEM_INST_RETIRED.STLB_HIT_STORES:R >= 0 ) else ( MEM_INST_RETIRED.STLB_HIT_STORES * ( 7 ) ) ) / tma_info_thread_clks + tma_store_stlb_miss", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", @@ -1030,7 +1030,7 @@ "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_load." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_dtlb_store - tma_store_stlb_miss )", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", "MetricName": "tma_store_stlb_hit", @@ -1038,7 +1038,7 @@ "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", "MetricName": "tma_store_stlb_miss", @@ -1046,7 +1046,7 @@ "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_4k", @@ -1054,7 +1054,7 @@ "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_2m", @@ -1062,7 +1062,7 @@ "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_1g", @@ -1070,7 +1070,7 @@ "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "max( 0 , tma_backend_bound - tma_memory_bound )", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", "MetricName": "tma_core_bound", @@ -1081,7 +1081,7 @@ "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.DIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", @@ -1090,7 +1090,7 @@ "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIV_ACTIVE." }, { - "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active", "MetricExpr": "ARITH.FPDIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_fp_divider", @@ -1098,7 +1098,7 @@ "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_int_divider", @@ -1106,7 +1106,7 @@ "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations", "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks + tma_c02_wait", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks;tma_issueSO", "MetricName": "tma_serializing_operation", @@ -1115,7 +1115,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions", "MetricExpr": "CPU_CLK_UNHALTED.PAUSE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_slow_pause", @@ -1124,7 +1124,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.PAUSE_INST." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings).", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings)", "MetricExpr": "CPU_CLK_UNHALTED.C01 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_c01_wait", @@ -1132,7 +1132,7 @@ "MetricThreshold": "tma_c01_wait > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings).", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings)", "MetricExpr": "CPU_CLK_UNHALTED.C02 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_c02_wait", @@ -1140,7 +1140,7 @@ "MetricThreshold": "tma_c02_wait > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions", "MetricExpr": "13 * MISC2_RETIRED.LFENCE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_memory_fence", @@ -1148,7 +1148,7 @@ "MetricThreshold": "tma_memory_fence > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates fraction of cycles where the Advanced Matrix eXtensions (AMX) execution engine was busy with tile (arithmetic) operations.", + "BriefDescription": "This metric estimates fraction of cycles where the Advanced Matrix eXtensions (AMX) execution engine was busy with tile (arithmetic) operations", "MetricExpr": "EXE.AMX_BUSY / tma_info_core_core_clks", "MetricGroup": "BvCB;Compute;HPC;Server;TopdownL3;tma_L3_group;tma_core_bound_group;Core_Clocks", "MetricName": "tma_amx_busy", @@ -1156,7 +1156,7 @@ "MetricThreshold": "tma_amx_busy > 0.5 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( tma_ports_utilized_0 * tma_info_thread_clks + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIV_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - EXE_ACTIVITY.BOUND_ON_LOADS ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", @@ -1165,7 +1165,7 @@ "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "max( EXE_ACTIVITY.EXE_BOUND_0_PORTS - RESOURCE_STALLS.SCOREBOARD , 0 ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", @@ -1174,7 +1174,7 @@ "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", + "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles)", "MetricExpr": "160 * ASSISTS.SSE_AVX_MIX / tma_info_thread_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks;tma_issueMV", "MetricName": "tma_mixing_vectors", @@ -1183,7 +1183,7 @@ "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", @@ -1192,7 +1192,7 @@ "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", @@ -1201,7 +1201,7 @@ "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", @@ -1210,7 +1210,7 @@ "PublicDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Sample with: UOPS_EXECUTED.CYCLES_GE_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + UOPS_DISPATCHED.PORT_5_11 + UOPS_DISPATCHED.PORT_6 ) / ( 5 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", @@ -1218,7 +1218,7 @@ "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED.PORT_0 / tma_info_core_core_clks", "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", @@ -1227,7 +1227,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_1 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", @@ -1236,7 +1236,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_6 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", @@ -1245,7 +1245,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "UOPS_DISPATCHED.PORT_2_3_10 / ( 3 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", @@ -1254,7 +1254,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations. Sample with: UOPS_DISPATCHED.PORT_2_3_10." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_4_9 + UOPS_DISPATCHED.PORT_7_8 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", @@ -1263,7 +1263,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations. Sample with: UOPS_DISPATCHED.PORT_7_8." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( topdown\\-retiring / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_retiring", @@ -1274,7 +1274,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "max( 0 , tma_retiring - tma_heavy_operations )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", "MetricName": "tma_light_operations", @@ -1285,7 +1285,7 @@ "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", @@ -1294,7 +1294,7 @@ "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", @@ -1303,7 +1303,7 @@ "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", @@ -1312,7 +1312,7 @@ "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR + FP_ARITH_INST_RETIRED2.VECTOR ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", @@ -1321,7 +1321,7 @@ "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.128B_PACKED_HALF ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", @@ -1330,7 +1330,7 @@ "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.256B_PACKED_HALF ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", @@ -1339,7 +1339,7 @@ "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.512B_PACKED_HALF ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_512b", @@ -1348,7 +1348,7 @@ "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired)", "MetricExpr": "tma_int_vector_128b + tma_int_vector_256b", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_int_operations", @@ -1357,7 +1357,7 @@ "PublicDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired). Vector/Matrix Int operations and shuffles are counted. Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain." }, { - "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", + "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired", "MetricExpr": "( INT_VEC_RETIRED.ADD_128 + INT_VEC_RETIRED.VNNI_128 ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops;tma_issue2P", "MetricName": "tma_int_vector_128b", @@ -1366,7 +1366,7 @@ "PublicDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", + "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired", "MetricExpr": "( INT_VEC_RETIRED.ADD_256 + INT_VEC_RETIRED.MUL_256 + INT_VEC_RETIRED.VNNI_256 ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops;tma_issue2P", "MetricName": "tma_int_vector_256b", @@ -1375,7 +1375,7 @@ "PublicDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses", "MetricExpr": "tma_light_operations * MEM_UOP_RETIRED.ANY / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_memory_operations", @@ -1383,7 +1383,7 @@ "MetricThreshold": "tma_memory_operations > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.MACRO_FUSED / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_fused_instructions", @@ -1392,7 +1392,7 @@ "PublicDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Other_Light_Ops in MTL!)}." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused", "MetricExpr": "tma_light_operations * ( BR_INST_RETIRED.ALL_BRANCHES - INST_RETIRED.MACRO_FUSED ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_non_fused_branches", @@ -1401,7 +1401,7 @@ "PublicDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused. Non-conditional branches like direct JMP or CALL would count here. Can be used to examine fusible conditional jumps that were not fused." }, { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", + "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes", "MetricExpr": "max( 0 , tma_light_operations - ( tma_fp_arith + tma_int_operations + tma_memory_operations + tma_fused_instructions + tma_non_fused_branches ) )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_other_light_ops", @@ -1410,7 +1410,7 @@ "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_light_ops_group;Slots", "MetricName": "tma_nop_instructions", @@ -1419,7 +1419,7 @@ "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer)", "MetricExpr": "tma_light_operations * INT_VEC_RETIRED.SHUFFLES / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_other_light_ops_group;Slots", "MetricName": "tma_shuffles_256b", @@ -1428,7 +1428,7 @@ "PublicDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer). Shuffles may incur slow cross \"vector lane\" data transfers." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "( topdown\\-heavy\\-ops / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", "MetricName": "tma_heavy_operations", @@ -1439,7 +1439,7 @@ "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops", "MetricExpr": "max( 0 , tma_heavy_operations - tma_microcode_sequencer )", "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueD0", "MetricName": "tma_few_uops_instructions", @@ -1448,7 +1448,7 @@ "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions. Related metrics: tma_decoder0_alone." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "UOPS_RETIRED.MS / tma_info_thread_slots", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", @@ -1457,7 +1457,7 @@ "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: UOPS_RETIRED.MS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( ( 99 *3 + 63 + 30 ) / 5 ) * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", @@ -1466,7 +1466,7 @@ "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults", "MetricExpr": "99 * ASSISTS.PAGE_FAULT / tma_info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_page_faults", @@ -1475,7 +1475,7 @@ "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults. A Page Fault may apply on first application access to a memory page. Note operating system handling of page faults accounts for the majority of its cost." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists", "MetricExpr": "30 * ASSISTS.FP / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_fp_assists", @@ -1484,7 +1484,7 @@ "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists", "MetricExpr": "63 * ASSISTS.SSE_AVX_MIX / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_avx_assists", @@ -1492,7 +1492,7 @@ "MetricThreshold": "tma_avx_assists > 0.1" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", @@ -1501,131 +1501,131 @@ "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources." }, { - "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts.", + "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts", "MetricExpr": "100 * ( 1 - tma_core_bound / tma_ports_utilization if tma_core_bound < tma_ports_utilization else 1 ) if tma_info_system_smt_2t_utilization > 0.5 else 0", "MetricGroup": "Cor;SMT;Metric", "MetricName": "tma_info_botlnk_l0_core_bound_likely", "MetricThreshold": "tma_info_botlnk_core_bound_likely > 0.5" }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", "MetricName": "tma_info_thread_uoppi", "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", "MetricName": "tma_info_thread_uptb", "MetricThreshold": "tma_info_thread_uptb < 6 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor.", + "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor", "MetricExpr": "tma_info_thread_slots / ( TOPDOWN.SLOTS / 2 ) if #SMT_on else 1", "MetricGroup": "SMT;TmaL1;TopdownL1;tma_L1_group;Metric", "MetricName": "tma_info_thread_slots_utilization" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( ( 1 * cpu@FP_ARITH_INST_RETIRED.SCALAR_SINGLE\\,umask\\=0x03@ + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * cpu@FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE\\,umask\\=0x18@ + 8 * cpu@FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE\\,umask\\=0x60@ + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) ) / tma_info_core_core_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( FP_ARITH_DISPATCHED.PORT_0 + FP_ARITH_DISPATCHED.PORT_1 + FP_ARITH_DISPATCHED.PORT_5 ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "uops Executed per Cycle.", + "BriefDescription": "uops Executed per Cycle", "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", "MetricGroup": "Power;Metric", "MetricName": "tma_info_core_epc" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipload", "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipstore", "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipbranch", "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", "MetricName": "tma_info_inst_mix_ipcall", "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", @@ -1633,20 +1633,20 @@ "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( 1 * cpu@FP_ARITH_INST_RETIRED.SCALAR_SINGLE\\,umask\\=0x03@ + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * cpu@FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE\\,umask\\=0x18@ + 8 * cpu@FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE\\,umask\\=0x60@ + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipflop", "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR + FP_ARITH_INST_RETIRED2.VECTOR ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", @@ -1654,7 +1654,7 @@ "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Half-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Half-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED2.SCALAR", "MetricGroup": "Flops;FpScalar;InsType;Server;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_hp", @@ -1662,7 +1662,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Half-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", @@ -1670,7 +1670,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", @@ -1678,7 +1678,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.128B_PACKED_HALF )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", @@ -1686,7 +1686,7 @@ "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.256B_PACKED_HALF )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", @@ -1694,7 +1694,7 @@ "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.512B_PACKED_HALF )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx512", @@ -1702,40 +1702,40 @@ "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / CPU_CLK_UNHALTED.PAUSE_INST", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ippause" }, { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / SW_PREFETCH_ACCESS.ANY", "MetricGroup": "Prefetches;Inst_Metric", "MetricName": "tma_info_inst_mix_ipswpf", "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions.", + "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions", "MetricExpr": "INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "MicroSeq;Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_strings_cycles", "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1" }, { - "BriefDescription": "Instructions per a microcode Assist invocation.", + "BriefDescription": "Instructions per a microcode Assist invocation", "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", "MetricName": "tma_info_pipeline_ipassist", @@ -1743,31 +1743,31 @@ "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core.", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core", "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@ )", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", + "BriefDescription": "Average number of uops fetched from DSB per cycle", "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_dsb" }, { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", + "BriefDescription": "Average number of uops fetched from MITE per cycle", "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_mite" }, { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", + "BriefDescription": "Average number of Uops issued by front-end when it issued something", "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_frontend_fetch_upc" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( UOPS_ISSUED.ANY )", "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", "MetricName": "tma_info_frontend_dsb_coverage", @@ -1775,77 +1775,77 @@ "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection.", + "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection", "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / cpu@INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_unknown_branch_cost", "PublicDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection. See Unknown_Branches node." }, { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", + "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "DSBmiss;Metric", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired DSB misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired DSB misses", "MetricExpr": "( FRONTEND_RETIRED.ANY_DSB_MISS * FRONTEND_RETIRED.ANY_DSB_MISS:R ) / tma_info_thread_clks", "MetricGroup": "DSBmiss;Fed;FetchLat;Clocks_Retired", "MetricName": "tma_info_frontend_dsb_switches_ret", "MetricThreshold": "tma_info_frontend_dsb_switches_ret > 0.05" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired operations that invoke the Microcode Sequencer.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired operations that invoke the Microcode Sequencer", "MetricExpr": "( FRONTEND_RETIRED.MS_FLOWS * FRONTEND_RETIRED.MS_FLOWS:R ) / tma_info_thread_clks", "MetricGroup": "Fed;FetchLat;MicroSeq;Clocks_Retired", "MetricName": "tma_info_frontend_ms_latency_ret", "MetricThreshold": "tma_info_frontend_ms_latency_ret > 0.05" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired branches who got branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired branches who got branch address clears", "MetricExpr": "( FRONTEND_RETIRED.UNKNOWN_BRANCH * FRONTEND_RETIRED.UNKNOWN_BRANCH:R ) / tma_info_thread_clks", "MetricGroup": "Fed;FetchLat;Clocks_Retired", "MetricName": "tma_info_frontend_unknown_branches_ret" }, { - "BriefDescription": "Average Latency for L1 instruction cache misses.", + "BriefDescription": "Average Latency for L1 instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / ICACHE_DATA.STALL_PERIODS", "MetricGroup": "Fed;FetchLat;IcMiss;Metric", "MetricName": "tma_info_frontend_icache_miss_latency" }, { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", "MetricGroup": "DSBmiss;Fed;Inst_Metric", "MetricName": "tma_info_frontend_ipdsb_miss_ret", "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache true code cacheline misses per kilo instruction", "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code" }, { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction", "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code_all" }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb + tma_ms ) )", "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", @@ -1853,7 +1853,7 @@ "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb + tma_ms ) ) )", "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", @@ -1861,7 +1861,7 @@ "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", + "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", @@ -1869,470 +1869,470 @@ "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmispredict", "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_ret", "MetricThreshold": "tma_info_bad_spec_ipmisp_ret < 500" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_indirect", "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 6 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", "MetricName": "tma_info_bad_spec_branch_misprediction_cost", "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_mispredicts_resteers." }, { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", + "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", "MetricGroup": "BrMispredicts;Metric", "MetricName": "tma_info_bad_spec_spec_clears_ratio" }, { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", + "BriefDescription": "Fraction of branches that are non-taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_nt" }, { - "BriefDescription": "Fraction of branches that are taken conditionals.", + "BriefDescription": "Fraction of branches that are taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_tk" }, { - "BriefDescription": "Fraction of branches that are CALL or RET.", + "BriefDescription": "Fraction of branches that are CALL or RET", "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_callret" }, { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", + "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_jump" }, { - "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group).", + "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group)", "MetricExpr": "1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_other_branches" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / MEM_LOAD_COMPLETED.L1_MISS_ANY", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki_load" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * L2_RQSTS.RFO_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_all" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", + "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_fb_hpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Offcore;Metric", "MetricName": "tma_info_memory_l3_cache_access_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", "MetricName": "tma_info_memory_tlb_page_walks_utilization", "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Fed;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_code_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_load_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand loads.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand loads", "MetricExpr": "( MEM_INST_RETIRED.STLB_MISS_LOADS * MEM_INST_RETIRED.STLB_MISS_LOADS:R ) / tma_info_thread_clks", "MetricGroup": "Mem;MemoryTLB;Clocks_Retired", "MetricName": "tma_info_memory_tlb_load_stlb_miss_ret", "MetricThreshold": "tma_info_memory_load_stlb_miss_ret > 0.05" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand stores.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand stores", "MetricExpr": "( MEM_INST_RETIRED.STLB_MISS_STORES * MEM_INST_RETIRED.STLB_MISS_STORES:R ) / tma_info_thread_clks", "MetricGroup": "Mem;MemoryTLB;Clocks_Retired", "MetricName": "tma_info_memory_tlb_store_stlb_miss_ret", "MetricThreshold": "tma_info_memory_store_stlb_miss_ret > 0.05" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_access_bw", "MetricGroup": "Mem;MemoryBW;Offcore;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_access_bw_2t" }, { - "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory).", + "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory)", "MetricExpr": "1000 * L2_LINES_OUT.SILENT / tma_info_inst_mix_instructions", "MetricGroup": "L2Evicts;Mem;Server;Core_Metric", "MetricName": "tma_info_memory_core_l2_evictions_silent_pki" }, { - "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction.", + "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction", "MetricExpr": "1000 * L2_LINES_OUT.NON_SILENT / tma_info_inst_mix_instructions", "MetricGroup": "L2Evicts;Mem;Server;Core_Metric", "MetricName": "tma_info_memory_core_l2_evictions_nonsilent_pki" }, { - "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses.", + "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses", "MetricExpr": "L2_LINES_OUT.USELESS_HWPF / ( L2_LINES_OUT.SILENT + L2_LINES_OUT.NON_SILENT )", "MetricGroup": "Prefetches;Metric", "MetricName": "tma_info_memory_prefetches_useless_hwpf", "MetricThreshold": "tma_info_memory_useless_hwpf > 0.15" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Latency for L3 cache miss demand Loads.", + "BriefDescription": "Average Latency for L3 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.L3_MISS_DEMAND_DATA_RD / OFFCORE_REQUESTS.L3_MISS_DEMAND_DATA_RD", "MetricGroup": "Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l3_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=0x1@", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Off-core accesses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches).", + "BriefDescription": "Off-core accesses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches)", "MetricExpr": "1000 * OCR.READS_TO_CORE.ANY_RESPONSE / tma_info_inst_mix_instructions", "MetricGroup": "CacheHits;Offcore;Metric", "MetricName": "tma_info_memory_mix_offcore_read_any_pki" }, { - "BriefDescription": "L3 cache misses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches).", + "BriefDescription": "L3 cache misses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches)", "MetricExpr": "1000 * OCR.READS_TO_CORE.L3_MISS / tma_info_inst_mix_instructions", "MetricGroup": "Offcore;Metric", "MetricName": "tma_info_memory_mix_offcore_read_l3m_pki" }, { - "BriefDescription": "Off-core accesses per kilo instruction for modified write requests.", + "BriefDescription": "Off-core accesses per kilo instruction for modified write requests", "MetricExpr": "1000 * OCR.MODIFIED_WRITE.ANY_RESPONSE / tma_info_inst_mix_instructions", "MetricGroup": "Offcore;Metric", "MetricName": "tma_info_memory_mix_offcore_mwrite_any_pki" }, { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", + "BriefDescription": "Un-cacheable retired load per kilo instruction", "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_uc_load_pki" }, { - "BriefDescription": "\"Bus lock\" per kilo instruction.", + "BriefDescription": "\"Bus lock\" per kilo instruction", "MetricExpr": "1000 * SQ_MISC.BUS_LOCK / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_bus_lock_pki" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz].", + "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz]", "MetricExpr": "tma_info_system_socket_clks / 1e9 / tma_info_system_time", "MetricGroup": "SoC;System_Metric", "MetricName": "tma_info_system_uncore_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( ( 1 * cpu@FP_ARITH_INST_RETIRED.SCALAR_SINGLE\\,umask\\=0x03@ + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * cpu@FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE\\,umask\\=0x18@ + 8 * cpu@FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE\\,umask\\=0x60@ + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_utilization", "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states.", + "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states", "MetricExpr": "CPU_CLK_UNHALTED.C0_WAIT / tma_info_thread_clks", "MetricGroup": "C0Wait;Metric", "MetricName": "tma_info_system_c0_wait", "MetricThreshold": "tma_info_system_c0_wait > 0.05" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "( 64 * ( UNC_M_CAS_COUNT_SCH0.RD + UNC_M_CAS_COUNT_SCH1.RD + UNC_M_CAS_COUNT_SCH0.WR + UNC_M_CAS_COUNT_SCH1.WR ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", "MetricName": "tma_info_system_dram_bw_use", "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Average number of parallel data read requests to external memory.", + "BriefDescription": "Average number of parallel data read requests to external memory", "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD\\,thresh\\=0x1@", "MetricGroup": "Mem;MemoryBW;SoC;System_Metric", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches." }, { - "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds].", + "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]", "MetricExpr": "( 1000000000 ) * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR ) / cha_0@event\\=0x0@", "MetricGroup": "MemOffcore;MemoryLat;SoC;Server;NanoSeconds", "MetricName": "tma_info_system_mem_dram_read_latency", "PublicDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]. Accounts for demand loads and L1/L2 data-read prefetches." }, { - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec].", + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]", "MetricExpr": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "IoBW;MemOffcore;SoC;Server;GB/sec", "MetricName": "tma_info_system_io_read_bw", "PublicDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]. Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU." }, { - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec].", + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]", "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_ITOM + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR ) * 64 / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "IoBW;MemOffcore;SoC;Server;GB/sec", "MetricName": "tma_info_system_io_write_bw", "PublicDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]. Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU." }, { - "BriefDescription": "Cross-socket Ultra Path Interconnect (UPI) data transmit bandwidth for data only [MB / sec].", + "BriefDescription": "Cross-socket Ultra Path Interconnect (UPI) data transmit bandwidth for data only [MB / sec]", "MetricExpr": "UNC_UPI_TxL_FLITS.ALL_DATA * 64 / 9 / 1000000", "MetricGroup": "SoC;Server;MB/sec", "MetricName": "tma_info_system_upi_data_transmit_bw" }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "( power@energy\\-pkg@ * ( 61 ) + 15.6 * power@energy\\-ram@ ) / ( ( duration_time ) * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", "MetricName": "tma_info_system_time", "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", "MetricName": "tma_info_system_mux", "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "cha_0@event\\=0x0@", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", "MetricName": "tma_info_system_ipfarbranch", "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" }, { - "BriefDescription": "Average Off-core access BW for Reads-to-Core (R2C).", + "BriefDescription": "Average Off-core access BW for Reads-to-Core (R2C)", "MetricExpr": "64 * OCR.READS_TO_CORE.ANY_RESPONSE / 1e9 / tma_info_system_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC;GB/sec", "MetricName": "tma_info_memory_soc_r2c_offcore_bw", "PublicDescription": "Average Off-core access BW for Reads-to-Core (R2C). R2C account for demand or prefetch load/RFO/code access that fill data into the Core caches." }, { - "BriefDescription": "Average L3-cache miss BW for Reads-to-Core (R2C).", + "BriefDescription": "Average L3-cache miss BW for Reads-to-Core (R2C)", "MetricExpr": "64 * OCR.READS_TO_CORE.L3_MISS / 1e9 / tma_info_system_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC;GB/sec", "MetricName": "tma_info_memory_soc_r2c_l3m_bw", "PublicDescription": "Average L3-cache miss BW for Reads-to-Core (R2C). This covering going to DRAM or other memory off-chip memory tears. See R2C_Offcore_BW." }, { - "BriefDescription": "Average DRAM BW for Reads-to-Core (R2C) covering for memory attached to local- and remote-socket.", + "BriefDescription": "Average DRAM BW for Reads-to-Core (R2C) covering for memory attached to local- and remote-socket", "MetricExpr": "64 * OCR.READS_TO_CORE.DRAM / 1e9 / tma_info_system_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC;GB/sec", "MetricName": "tma_info_memory_soc_r2c_dram_bw", diff --git a/GRR/metrics/grandridge_metrics.json b/GRR/metrics/grandridge_metrics.json index 4d5bfe42..a33b0afa 100644 --- a/GRR/metrics/grandridge_metrics.json +++ b/GRR/metrics/grandridge_metrics.json @@ -6,7 +6,7 @@ "Version": "0", "Legend": "", "TmaVersion": "3.6", - "TmaFlavor": "Private" + "TmaFlavor": "Public" }, "Metrics": [ { diff --git a/HSW/metrics/perf/haswell_metrics_perf.json b/HSW/metrics/perf/haswell_metrics_perf.json index 672ff2ea..b1fc8d01 100644 --- a/HSW/metrics/perf/haswell_metrics_perf.json +++ b/HSW/metrics/perf/haswell_metrics_perf.json @@ -1,794 +1,870 @@ [ { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "IDQ_UOPS_NOT_DELIVERED.CORE / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_frontend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_frontend_bound > 15", "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( 4 ) * ( min( CPU_CLK_UNHALTED.THREAD , IDQ_UOPS_NOT_DELIVERED.CYCLES_0_UOPS_DELIV.CORE ) ) / tma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", "MetricName": "tma_fetch_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_latency > 10 & tma_frontend_bound > 15", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: RS_EVENTS.EMPTY_END." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE.IFDATA_STALL / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_icache_misses > 5 & tma_fetch_latency > 10 & tma_frontend_bound > 15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "( ( 14 * ITLB_MISSES.STLB_HIT + ITLB_MISSES.WALK_DURATION ) ) / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_itlb_misses > 5 & tma_fetch_latency > 10 & tma_frontend_bound > 15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: ITLB_MISSES.WALK_COMPLETED." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "( 12 ) * ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT + BACLEARS.ANY ) / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_branch_resteers", "ScaleUnit": "100%", + "MetricThreshold": "tma_branch_resteers > 5 & tma_fetch_latency > 10 & tma_frontend_bound > 15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 2 ) * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", + "MetricThreshold": "tma_ms_switches > 5 & tma_fetch_latency > 10 & tma_frontend_bound > 15", "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "ILD_STALL.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_lcp", "ScaleUnit": "100%", + "MetricThreshold": "tma_lcp > 5 & tma_fetch_latency > 10 & tma_frontend_bound > 15", "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", + "MetricThreshold": "tma_dsb_switches > 5 & tma_fetch_latency > 10 & tma_frontend_bound > 15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "tma_frontend_bound - tma_fetch_latency", "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_bandwidth > 20", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.ALL_MITE_CYCLES_ANY_UOPS - IDQ.ALL_MITE_CYCLES_4_UOPS ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", "ScaleUnit": "100%", + "MetricThreshold": "tma_mite > 10 & tma_fetch_bandwidth > 20", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.ALL_DSB_CYCLES_ANY_UOPS - IDQ.ALL_DSB_CYCLES_4_UOPS ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", "ScaleUnit": "100%", + "MetricThreshold": "tma_dsb > 15 & tma_fetch_bandwidth > 20", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "( UOPS_ISSUED.ANY - ( UOPS_RETIRED.RETIRE_SLOTS ) + ( 4 ) * ( ( INT_MISC.RECOVERY_CYCLES_ANY / 2 ) if #SMT_on else INT_MISC.RECOVERY_CYCLES ) ) / tma_info_thread_slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_bad_speculation", "ScaleUnit": "100%", + "MetricThreshold": "tma_bad_speculation > 15", "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example." }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * tma_bad_speculation", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", + "MetricThreshold": "tma_branch_mispredicts > 10 & tma_bad_speculation > 15", "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "tma_bad_speculation - tma_branch_mispredicts", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", + "MetricThreshold": "tma_machine_clears > 10 & tma_bad_speculation > 15", "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT." }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "1 - ( tma_frontend_bound + tma_bad_speculation + tma_retiring )", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_backend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_backend_bound > 20", "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound." }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( ( ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.STALLS_LDM_PENDING ) ) + RESOURCE_STALLS.SB ) / ( ( ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.CYCLES_NO_EXECUTE ) ) + ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ - ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ if ( tma_info_thread_ipc > 1.8 ) else cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ ) ) / 2 - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) + RESOURCE_STALLS.SB ) if #SMT_on else ( ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.CYCLES_NO_EXECUTE ) ) + cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ - ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ if ( tma_info_thread_ipc > 1.8 ) else cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ ) - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) + RESOURCE_STALLS.SB ) ) ) * tma_backend_bound", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_memory_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two)." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.STALLS_LDM_PENDING ) ) - CYCLE_ACTIVITY.STALLS_L1D_PENDING ) / tma_info_thread_clks , 0 )", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_bound > 10 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "( ( 8 ) * DTLB_LOAD_MISSES.STLB_HIT + DTLB_LOAD_MISSES.WALK_DURATION ) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", + "MetricThreshold": "tma_dtlb_load > 10 & tma_l1_bound > 10 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_UOPS_RETIRED.STLB_MISS_LOADS." }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_fwd_blk > 10 & tma_l1_bound > 10 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_STORES ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks", "MetricName": "tma_lock_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_lock_latency > 20 & tma_l1_bound > 10 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOADS." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_loads > 30", "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_UOPS_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset.", + "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset", "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_4k_aliasing", "ScaleUnit": "100%", + "MetricThreshold": "tma_4k_aliasing > 20 & tma_l1_bound > 10 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset. False match is possible; which incur a few cycles load re-issue. However; the short re-issue duration is often hidden by the out-of-order core and HW optimizations; hence a user may safely ignore a high value of this metric unless it manages to propagate up into parent nodes of the hierarchy (e.g. to L1_Bound)." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PEND_MISS.REQUEST_FB_FULL\\,cmask\\=0x1@ / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_fb_full", "ScaleUnit": "100%", + "MetricThreshold": "tma_fb_full > 30", "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory)." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L1D_PENDING - CYCLE_ACTIVITY.STALLS_L2_PENDING ) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_bound > 5 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( MEM_LOAD_UOPS_RETIRED.L3_HIT / ( MEM_LOAD_UOPS_RETIRED.L3_HIT + ( 7 ) * MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) * CYCLE_ACTIVITY.STALLS_L2_PENDING / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l3_bound > 5 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( 60 ) * ( MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) ) + ( 43 ) * ( MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) ) ) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", + "MetricThreshold": "tma_contested_accesses > 5 & tma_l3_bound > 5 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( 43 ) * ( MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) ) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", + "MetricThreshold": "tma_data_sharing > 5 & tma_l3_bound > 5 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( 29 ) * ( MEM_LOAD_UOPS_RETIRED.L3_HIT * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l3_hit_latency > 10 & tma_l3_bound > 5 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "( ( OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 ) if #SMT_on else OFFCORE_REQUESTS_BUFFER.SQ_FULL ) / tma_info_core_core_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks", "MetricName": "tma_sq_full", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_sq_full > 30 & tma_l3_bound > 5 & tma_memory_bound > 20 & tma_backend_bound > 20" }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( 1 - ( MEM_LOAD_UOPS_RETIRED.L3_HIT / ( MEM_LOAD_UOPS_RETIRED.L3_HIT + ( 7 ) * MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) ) * CYCLE_ACTIVITY.STALLS_L2_PENDING / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_dram_bound > 10 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x6@ ) ) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", + "MetricThreshold": "tma_mem_bandwidth > 20 & tma_dram_bound > 10 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that)." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_mem_latency > 10 & tma_dram_bound > 10 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that)." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "RESOURCE_STALLS.SB / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_bound > 20 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_UOPS_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( L2_RQSTS.RFO_HIT * ( 9 ) * ( 1 - ( MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", "MetricName": "tma_store_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_latency > 10 & tma_store_bound > 20 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( 60 ) * OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.HITM_OTHER_CORE / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", + "MetricThreshold": "tma_false_sharing > 5 & tma_store_bound > 20 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.HITM_OTHER_CORE." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "2 * MEM_UOPS_RETIRED.SPLIT_STORES / tma_info_core_core_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization", "MetricName": "tma_split_stores", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_stores > 20 & tma_store_bound > 20 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_UOPS_RETIRED.SPLIT_STORES." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( ( 8 ) * DTLB_STORE_MISSES.STLB_HIT + DTLB_STORE_MISSES.WALK_DURATION ) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", + "MetricThreshold": "tma_dtlb_store > 5 & tma_store_bound > 20 & tma_memory_bound > 20 & tma_backend_bound > 20", "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_UOPS_RETIRED.STLB_MISS_STORES." }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "tma_backend_bound - tma_memory_bound", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_core_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_core_bound > 10 & tma_backend_bound > 20", "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations)." }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "10 * ARITH.DIVIDER_UOPS / tma_info_core_core_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", "ScaleUnit": "100%", + "MetricThreshold": "tma_divider > 20 & tma_core_bound > 10 & tma_backend_bound > 20", "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIVIDER_UOPS." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( ( ( ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.CYCLES_NO_EXECUTE ) ) + ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ - ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ if ( tma_info_thread_ipc > 1.8 ) else cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ ) ) / 2 - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) + RESOURCE_STALLS.SB ) if #SMT_on else ( ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.CYCLES_NO_EXECUTE ) ) + cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ - ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ if ( tma_info_thread_ipc > 1.8 ) else cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ ) - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) + RESOURCE_STALLS.SB ) ) - RESOURCE_STALLS.SB - ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.STALLS_LDM_PENDING ) ) ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilization > 15 & tma_core_bound > 10 & tma_backend_bound > 20", "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,inv\\=0x1\\,cmask\\=0x1@ ) / 2 if #SMT_on else ( ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.CYCLES_NO_EXECUTE ) ) - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) ) ) / tma_info_core_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_0 > 20 & tma_ports_utilization > 15 & tma_core_bound > 10 & tma_backend_bound > 20", "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ ) / 2 if #SMT_on else ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ ) ) / tma_info_core_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_1", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_1 > 20 & tma_ports_utilization > 15 & tma_core_bound > 10 & tma_backend_bound > 20", "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ ) / 2 if #SMT_on else ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ ) ) / tma_info_core_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_2 > 15 & tma_ports_utilization > 15 & tma_core_bound > 10 & tma_backend_bound > 20", "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ ) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_3m > 40 & tma_ports_utilization > 15 & tma_core_bound > 10 & tma_backend_bound > 20" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT.PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_alu_op_utilization > 40" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_0 / tma_info_core_core_clks", "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", "MetricName": "tma_port_0", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_0 > 60", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_1 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", "MetricName": "tma_port_1", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_1 > 60", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_5 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", "MetricName": "tma_port_5", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_5 > 60", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_6 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", "MetricName": "tma_port_6", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_6 > 60", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "( UOPS_DISPATCHED_PORT.PORT_2 + UOPS_DISPATCHED_PORT.PORT_3 + UOPS_DISPATCHED_PORT.PORT_7 - UOPS_DISPATCHED_PORT.PORT_4 ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_op_utilization > 60" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_2 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_group;Core_Clocks", "MetricName": "tma_port_2", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_2 > 60", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads). Sample with: UOPS_DISPATCHED_PORT.PORT_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_3 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_group;Core_Clocks", "MetricName": "tma_port_3", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_3 > 60", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads). Sample with: UOPS_DISPATCHED_PORT.PORT_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_op_utilization > 60" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks", "MetricName": "tma_port_4", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_4 > 60", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data). Sample with: UOPS_DISPATCHED_PORT.PORT_4." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_7 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks", "MetricName": "tma_port_7", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_7 > 60", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address). Sample with: UOPS_DISPATCHED_PORT.PORT_7." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / tma_info_thread_slots", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_retiring", "ScaleUnit": "100%", + "MetricThreshold": "tma_retiring > 70 | tma_heavy_operations > 10", "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.RETIRE_SLOTS." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_light_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_light_operations > 60", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "tma_microcode_sequencer", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_heavy_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_heavy_operations > 10", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+])." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "( ( UOPS_RETIRED.RETIRE_SLOTS ) / UOPS_ISSUED.ANY ) * IDQ.MS_UOPS / tma_info_thread_slots", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", + "MetricThreshold": "tma_microcode_sequencer > 5 & tma_heavy_operations > 10", "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( 66 ) * OTHER_ASSISTS.ANY_WB_ASSIST / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_assists > 10 & tma_microcode_sequencer > 5 & tma_heavy_operations > 10", "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY_WB_ASSIST." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", "ScaleUnit": "100%", + "MetricThreshold": "tma_cisc > 10 & tma_microcode_sequencer > 5 & tma_heavy_operations > 10", "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources." }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", - "MetricName": "tma_info_thread_uoppi" + "MetricName": "tma_info_thread_uoppi", + "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", - "MetricName": "tma_info_thread_uptb" + "MetricName": "tma_info_thread_uptb", + "MetricThreshold": "tma_info_thread_uptb < 4 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "( 4 ) * tma_info_core_core_clks", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "( UOPS_EXECUTED.CORE / 2 / ( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ ) ) if #SMT_on else UOPS_EXECUTED.CORE / ( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ )", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "( CPU_CLK_UNHALTED.THREAD_ANY / 2 ) if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_UOPS_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipload" + "MetricName": "tma_info_inst_mix_ipload", + "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_UOPS_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipstore" + "MetricName": "tma_info_inst_mix_ipstore", + "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipbranch" + "MetricName": "tma_info_inst_mix_ipbranch", + "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipcall" + "MetricName": "tma_info_inst_mix_ipcall", + "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_iptb" + "MetricName": "tma_info_inst_mix_iptb", + "MetricThreshold": "tma_info_inst_mix_iptb < 4 * 2 + 1" }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / cpu@UOPS_RETIRED.RETIRE_SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( ( IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS ) )", "MetricGroup": "DSB;Fed;FetchBW;Metric", - "MetricName": "tma_info_frontend_dsb_coverage" + "MetricName": "tma_info_frontend_dsb_coverage", + "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 4 > 0.35" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmispredict" + "MetricName": "tma_info_bad_spec_ipmispredict", + "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / ( ( ( UOPS_RETIRED.RETIRE_SLOTS ) / UOPS_ISSUED.ANY ) * BR_MISP_EXEC.INDIRECT )", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_indirect" + "MetricName": "tma_info_bad_spec_ipmisp_indirect", + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_UOPS_RETIRED.L1_MISS + MEM_LOAD_UOPS_RETIRED.HIT_LFB )", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_UOPS_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_UOPS_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * OFFCORE_REQUESTS.DEMAND_RFO / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_UOPS_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_DURATION + DTLB_LOAD_MISSES.WALK_DURATION + DTLB_STORE_MISSES.WALK_DURATION ) / tma_info_core_core_clks", "MetricGroup": "Mem;MemoryTLB;Core_Metric", - "MetricName": "tma_info_memory_tlb_page_walks_utilization" + "MetricName": "tma_info_memory_tlb_page_walks_utilization", + "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / ( CPU_CLK_UNHALTED.REF_XCLK_ANY / 2 ) if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", - "MetricName": "tma_info_system_kernel_utilization" + "MetricName": "tma_info_system_kernel_utilization", + "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "64 * ( UNC_ARB_TRK_REQUESTS.ALL + UNC_ARB_COH_TRK_REQUESTS.ALL ) / ( 1000000 ) / tma_info_system_time / 1000", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec", "MetricName": "tma_info_system_dram_bw_use" }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "power@energy\\-pkg@ * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", - "MetricName": "tma_info_system_mux" + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "UNC_CLOCK.SOCKET", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", - "MetricName": "tma_info_system_ipfarbranch" + "MetricName": "tma_info_system_ipfarbranch", + "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" } ] \ No newline at end of file diff --git a/HSX/metrics/perf/haswellx_metrics_perf.json b/HSX/metrics/perf/haswellx_metrics_perf.json index 5ca86af5..e1e4c1ad 100644 --- a/HSX/metrics/perf/haswellx_metrics_perf.json +++ b/HSX/metrics/perf/haswellx_metrics_perf.json @@ -1,139 +1,139 @@ [ { - "BriefDescription": "CPU operating frequency (in GHz).", - "MetricExpr": "(( CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000)", + "BriefDescription": "CPU operating frequency (in GHz)", + "MetricExpr": "((CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000)", "MetricGroup": "", "MetricName": "cpu_operating_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", - "MetricExpr": "( cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x182@ + cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x192@ ) / INST_RETIRED.ANY", + "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", + "MetricExpr": "(cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x182@ + cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x192@) / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_data_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", - "MetricExpr": "( cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x181@ + cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x191@ ) / INST_RETIRED.ANY", + "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", + "MetricExpr": "(cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x181@ + cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x191@) / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_code_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Percentage of time spent in the active CPU power state C0.", + "BriefDescription": "Percentage of time spent in the active CPU power state C0", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "", "MetricName": "cpu_utilization", "ScaleUnit": "100%" }, { - "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ / ( cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ + cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@ )", + "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ / (cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ + cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_local_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@ / ( cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ + cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@ )", + "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@ / (cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ + cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_remote_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles.", + "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles", "MetricExpr": "CPU_CLK_UNHALTED.THREAD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions", "MetricExpr": "MEM_UOPS_RETIRED.ALL_LOADS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "loads_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions", "MetricExpr": "MEM_UOPS_RETIRED.ALL_STORES / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "stores_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L1D.REPLACEMENT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_UOPS_RETIRED.L1_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions", "MetricExpr": "L2_RQSTS.ALL_CODE_RD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1_i_code_read_misses_with_prefetches_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_UOPS_RETIRED.L2_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L2_LINES_IN.ALL / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_UOPS_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions", "MetricExpr": "L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_code_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) in nano seconds.", - "MetricExpr": "( 1000000000 * ( cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x182@ ) / ( UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) in nano seconds", + "MetricExpr": "( 1000000000 * (cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x182@) / (UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_data_read_demand_plus_prefetch_miss_latency", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to local memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( cbox@UNC_C_TOR_OCCUPANCY.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ ) / ( UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to local memory in nano seconds", + "MetricExpr": "( 1000000000 * (cbox@UNC_C_TOR_OCCUPANCY.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_LOCAL_OPCODE\\,filter_opc\\=0x182@) / (UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_data_read_demand_plus_prefetch_miss_latency_for_local_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to remote memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( cbox@UNC_C_TOR_OCCUPANCY.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@ ) / ( UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to remote memory in nano seconds", + "MetricExpr": "( 1000000000 * (cbox@UNC_C_TOR_OCCUPANCY.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_REMOTE_OPCODE\\,filter_opc\\=0x182@) / (UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_data_read_demand_plus_prefetch_miss_latency_for_remote_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_mpi", @@ -141,7 +141,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the ITLB (Instruction TLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED_2M_4M / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_large_page_mpi", @@ -149,7 +149,7 @@ "PublicDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the Instruction Translation Lookaside Buffer (ITLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions", "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_load_mpi", @@ -157,7 +157,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions", "MetricExpr": "DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_store_mpi", @@ -165,84 +165,84 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Uncore operating frequency in GHz.", - "MetricExpr": "( UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) / 1000000000) / duration_time", + "BriefDescription": "Uncore operating frequency in GHz", + "MetricExpr": "(UNC_C_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) / 1000000000) / duration_time", "MetricGroup": "", "MetricName": "uncore_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Intel(R) Quick Path Interconnect (QPI) data transmit bandwidth (MB/sec).", - "MetricExpr": "( UNC_Q_TxL_FLITS_G0.DATA * 8 / 1000000) / duration_time", + "BriefDescription": "Intel(R) Quick Path Interconnect (QPI) data transmit bandwidth (MB/sec)", + "MetricExpr": "(UNC_Q_TxL_FLITS_G0.DATA * 8 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "qpi_data_transmit_bw", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory read bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory read bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory write bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory write bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory bandwidth (MB/sec)", + "MetricExpr": "((UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_total", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU.", - "MetricExpr": "( cbox@UNC_C_TOR_INSERTS.OPCODE\\,filter_opc\\=0x19e@ * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU", + "MetricExpr": "(cbox@UNC_C_TOR_INSERTS.OPCODE\\,filter_opc\\=0x19e@ * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU.", - "MetricExpr": "( cbox@UNC_C_TOR_INSERTS.OPCODE\\,filter_opc\\=0x1c8\\,filter_tid\\=0x3e@ * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU", + "MetricExpr": "(cbox@UNC_C_TOR_INSERTS.OPCODE\\,filter_opc\\=0x1c8\\,filter_tid\\=0x3e@ * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.DSB_UOPS / UOPS_ISSUED.ANY )", + "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.DSB_UOPS / UOPS_ISSUED.ANY)", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_decoded_icache", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MITE_UOPS / UOPS_ISSUED.ANY )", + "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MITE_UOPS / UOPS_ISSUED.ANY)", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_legacy_decode_pipeline", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MS_UOPS / UOPS_ISSUED.ANY )", + "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MS_UOPS / UOPS_ISSUED.ANY)", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_microcode_sequencer", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from loop stream detector(LSD) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( UOPS_ISSUED.ANY - IDQ.MITE_UOPS - IDQ.MS_UOPS - IDQ.DSB_UOPS ) / UOPS_ISSUED.ANY", + "BriefDescription": "Uops delivered from loop stream detector(LSD) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(UOPS_ISSUED.ANY - IDQ.MITE_UOPS - IDQ.MS_UOPS - IDQ.DSB_UOPS) / UOPS_ISSUED.ANY", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_loop_stream_detector", "ScaleUnit": "100%" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "IDQ_UOPS_NOT_DELIVERED.CORE / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_frontend_bound", @@ -251,7 +251,7 @@ "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( 4 ) * ( min( CPU_CLK_UNHALTED.THREAD , IDQ_UOPS_NOT_DELIVERED.CYCLES_0_UOPS_DELIV.CORE ) ) / tma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", "MetricName": "tma_fetch_latency", @@ -260,7 +260,7 @@ "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: RS_EVENTS.EMPTY_END." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE.IFDATA_STALL / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", @@ -268,7 +268,7 @@ "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "( ( 14 * ITLB_MISSES.STLB_HIT + ITLB_MISSES.WALK_DURATION ) ) / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", @@ -277,7 +277,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: ITLB_MISSES.WALK_COMPLETED." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "( 12 ) * ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT + BACLEARS.ANY ) / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", @@ -286,7 +286,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 2 ) * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", @@ -295,34 +295,34 @@ "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "ILD_STALL.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_info_frontend_dsb_coverage, tma_fetch_bandwidth, tma_info_inst_mix_iptb." + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Related metrics: tma_lcp, tma_info_frontend_dsb_coverage, tma_fetch_bandwidth, tma_info_inst_mix_iptb." + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Related metrics: tma_fetch_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "tma_frontend_bound - tma_fetch_latency", "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", "MetricThreshold": "tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Related metrics: tma_dsb_switches, tma_lcp, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Related metrics: tma_dsb_switches, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.ALL_MITE_CYCLES_ANY_UOPS - IDQ.ALL_MITE_CYCLES_4_UOPS ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", @@ -331,7 +331,7 @@ "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.ALL_DSB_CYCLES_ANY_UOPS - IDQ.ALL_DSB_CYCLES_4_UOPS ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", @@ -340,7 +340,7 @@ "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "( UOPS_ISSUED.ANY - ( UOPS_RETIRED.RETIRE_SLOTS ) + ( 4 ) * ( ( INT_MISC.RECOVERY_CYCLES_ANY / 2 ) if #SMT_on else INT_MISC.RECOVERY_CYCLES ) ) / tma_info_thread_slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_bad_speculation", @@ -349,7 +349,7 @@ "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example." }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * tma_bad_speculation", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", @@ -358,16 +358,16 @@ "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "tma_bad_speculation - tma_branch_mispredicts", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_ms_switches, tma_microcode_sequencer, tma_false_sharing, tma_remote_cache, tma_l1_bound, tma_data_sharing, tma_contested_accesses." + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches, tma_remote_cache." }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "1 - ( tma_frontend_bound + tma_bad_speculation + tma_retiring )", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_backend_bound", @@ -376,7 +376,7 @@ "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound." }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( ( ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.STALLS_LDM_PENDING ) ) + RESOURCE_STALLS.SB ) / ( ( ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.CYCLES_NO_EXECUTE ) ) + ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ - ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ if ( tma_info_thread_ipc > 1.8 ) else cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ ) ) / 2 - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) + RESOURCE_STALLS.SB ) if #SMT_on else ( ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.CYCLES_NO_EXECUTE ) ) + cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ - ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ if ( tma_info_thread_ipc > 1.8 ) else cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ ) - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) + RESOURCE_STALLS.SB ) ) ) * tma_backend_bound", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_memory_bound", @@ -385,16 +385,16 @@ "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two)." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.STALLS_LDM_PENDING ) ) - CYCLE_ACTIVITY.STALLS_L1D_PENDING ) / tma_info_thread_clks , 0 )", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT. Related metrics: tma_microcode_sequencer, tma_machine_clears, tma_ms_switches, tma_ports_utilized_1." + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT. Related metrics: tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "( ( 8 ) * DTLB_LOAD_MISSES.STLB_HIT + DTLB_LOAD_MISSES.WALK_DURATION ) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", @@ -403,7 +403,7 @@ "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_UOPS_RETIRED.STLB_MISS_LOADS. Related metrics: tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", @@ -412,7 +412,7 @@ "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_STORES ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", @@ -421,7 +421,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", @@ -430,7 +430,7 @@ "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_UOPS_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset.", + "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset", "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_4k_aliasing", @@ -439,16 +439,16 @@ "PublicDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset. False match is possible; which incur a few cycles load re-issue. However; the short re-issue duration is often hidden by the out-of-order core and HW optimizations; hence a user may safely ignore a high value of this metric unless it manages to propagate up into parent nodes of the hierarchy (e.g. to L1_Bound)." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PEND_MISS.REQUEST_FB_FULL\\,cmask\\=0x1@ / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", "MetricThreshold": "tma_fb_full > 0.3", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_info_system_dram_bw_use, tma_sq_full, tma_mem_bandwidth, tma_store_latency." + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L1D_PENDING - CYCLE_ACTIVITY.STALLS_L2_PENDING ) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", @@ -457,7 +457,7 @@ "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( MEM_LOAD_UOPS_RETIRED.L3_HIT / ( MEM_LOAD_UOPS_RETIRED.L3_HIT + ( 7 ) * MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) * CYCLE_ACTIVITY.STALLS_L2_PENDING / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", @@ -466,43 +466,43 @@ "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( 60 ) * ( MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) + ( 43 ) * ( MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) ) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_machine_clears, tma_false_sharing, tma_remote_cache, tma_data_sharing." + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_data_sharing, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( 43 ) * ( MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_machine_clears, tma_false_sharing, tma_remote_cache, tma_contested_accesses." + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_contested_accesses, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( 41 ) * ( MEM_LOAD_UOPS_RETIRED.L3_HIT * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT. Related metrics: tma_store_latency, tma_mem_latency, tma_branch_resteers." + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT. Related metrics: tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "( ( OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 ) if #SMT_on else OFFCORE_REQUESTS_BUFFER.SQ_FULL ) / tma_info_core_core_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", "ScaleUnit": "100%", "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_info_system_dram_bw_use, tma_fb_full, tma_mem_bandwidth." + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( 1 - ( MEM_LOAD_UOPS_RETIRED.L3_HIT / ( MEM_LOAD_UOPS_RETIRED.L3_HIT + ( 7 ) * MEM_LOAD_UOPS_RETIRED.L3_MISS ) ) ) * CYCLE_ACTIVITY.STALLS_L2_PENDING / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", @@ -511,7 +511,7 @@ "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x6@ ) ) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", @@ -520,7 +520,7 @@ "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", @@ -529,7 +529,7 @@ "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory", "MetricExpr": "( 200 ) * ( MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_local_mem", @@ -538,7 +538,7 @@ "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory. Caching will improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory", "MetricExpr": "( 310 ) * ( MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_remote_mem", @@ -547,16 +547,16 @@ "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues", "MetricExpr": "( ( 200 ) * ( MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) + ( 180 ) * ( MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD * ( 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / ( ( MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS ) + MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD ) ) ) ) / tma_info_thread_clks", "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_remote_cache", "ScaleUnit": "100%", "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_machine_clears, tma_false_sharing, tma_data_sharing, tma_contested_accesses." + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "RESOURCE_STALLS.SB / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", @@ -565,25 +565,25 @@ "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_UOPS_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( L2_RQSTS.RFO_HIT * ( 9 ) * ( 1 - ( MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_l3_hit_latency, tma_fb_full, tma_lock_latency, tma_branch_resteers." + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( ( 200 ) * OFFCORE_RESPONSE.DEMAND_RFO.LLC_MISS.REMOTE_HITM + ( 60 ) * OFFCORE_RESPONSE.DEMAND_RFO.LLC_HIT.HITM_OTHER_CORE ) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM, OFFCORE_RESPONSE.DEMAND_RFO.LLC_HIT.HITM_OTHER_CORE, OFFCORE_RESPONSE.DEMAND_RFO.LLC_MISS.REMOTE_HITM. Related metrics: tma_machine_clears, tma_remote_cache, tma_data_sharing, tma_contested_accesses." + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM, OFFCORE_RESPONSE.DEMAND_RFO.LLC_HIT.HITM_OTHER_CORE, OFFCORE_RESPONSE.DEMAND_RFO.LLC_MISS.REMOTE_HITM. Related metrics: tma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "2 * MEM_UOPS_RETIRED.SPLIT_STORES / tma_info_core_core_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", @@ -592,7 +592,7 @@ "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_UOPS_RETIRED.SPLIT_STORES. Related metrics: tma_port_4." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( ( 8 ) * DTLB_STORE_MISSES.STLB_HIT + DTLB_STORE_MISSES.WALK_DURATION ) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", @@ -601,7 +601,7 @@ "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_UOPS_RETIRED.STLB_MISS_STORES. Related metrics: tma_dtlb_load." }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "tma_backend_bound - tma_memory_bound", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_core_bound", @@ -610,7 +610,7 @@ "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations)." }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "10 * ARITH.DIVIDER_UOPS / tma_info_core_core_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", @@ -619,7 +619,7 @@ "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIVIDER_UOPS." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( ( ( ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.CYCLES_NO_EXECUTE ) ) + ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ - ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ if ( tma_info_thread_ipc > 1.8 ) else cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ ) ) / 2 - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) + RESOURCE_STALLS.SB ) if #SMT_on else ( ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.CYCLES_NO_EXECUTE ) ) + cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ - ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ if ( tma_info_thread_ipc > 1.8 ) else cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ ) - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) + RESOURCE_STALLS.SB ) ) - RESOURCE_STALLS.SB - ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.STALLS_LDM_PENDING ) ) ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", @@ -628,7 +628,7 @@ "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,inv\\=0x1\\,cmask\\=0x1@ ) / 2 if #SMT_on else ( ( min( CPU_CLK_UNHALTED.THREAD , CYCLE_ACTIVITY.CYCLES_NO_EXECUTE ) ) - ( RS_EVENTS.EMPTY_CYCLES if ( tma_fetch_latency > 0.1 ) else 0 ) ) ) / tma_info_core_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", @@ -637,7 +637,7 @@ "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ ) / 2 if #SMT_on else ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ ) ) / tma_info_core_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", @@ -646,16 +646,16 @@ "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ ) / 2 if #SMT_on else ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x2@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ ) ) / tma_info_core_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Related metrics: tma_port_5, tma_port_0, tma_port_1, tma_port_6." + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Related metrics: tma_port_0, tma_port_1, tma_port_5, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x3@ ) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", @@ -663,7 +663,7 @@ "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT.PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", @@ -671,43 +671,43 @@ "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_0 / tma_info_core_core_clks", "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", "ScaleUnit": "100%", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_port_5, tma_ports_utilized_2, tma_port_1, tma_port_6." + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_1 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", "ScaleUnit": "100%", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_port_5, tma_ports_utilized_2, tma_port_0, tma_port_6." + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_5 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_5", "ScaleUnit": "100%", "MetricThreshold": "tma_port_5 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_ports_utilized_2, tma_port_0, tma_port_1, tma_port_6." + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_6 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", "ScaleUnit": "100%", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_port_5, tma_ports_utilized_2, tma_port_0, tma_port_1." + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "( UOPS_DISPATCHED_PORT.PORT_2 + UOPS_DISPATCHED_PORT.PORT_3 + UOPS_DISPATCHED_PORT.PORT_7 - UOPS_DISPATCHED_PORT.PORT_4 ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", @@ -715,7 +715,7 @@ "MetricThreshold": "tma_load_op_utilization > 0.6" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_2 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_group;Core_Clocks", "MetricName": "tma_port_2", @@ -724,7 +724,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads). Sample with: UOPS_DISPATCHED_PORT.PORT_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_3 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_group;Core_Clocks", "MetricName": "tma_port_3", @@ -733,7 +733,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads). Sample with: UOPS_DISPATCHED_PORT.PORT_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", @@ -741,7 +741,7 @@ "MetricThreshold": "tma_store_op_utilization > 0.6" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks;tma_issueSpSt", "MetricName": "tma_port_4", @@ -750,7 +750,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data). Sample with: UOPS_DISPATCHED_PORT.PORT_4. Related metrics: tma_split_stores." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_7 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks", "MetricName": "tma_port_7", @@ -759,7 +759,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address). Sample with: UOPS_DISPATCHED_PORT.PORT_7." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / tma_info_thread_slots", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_retiring", @@ -768,7 +768,7 @@ "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.RETIRE_SLOTS." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_light_operations", @@ -777,7 +777,7 @@ "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "tma_microcode_sequencer", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_heavy_operations", @@ -786,7 +786,7 @@ "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+])." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "( ( UOPS_RETIRED.RETIRE_SLOTS ) / UOPS_ISSUED.ANY ) * IDQ.MS_UOPS / tma_info_thread_slots", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", @@ -795,7 +795,7 @@ "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( 66 ) * OTHER_ASSISTS.ANY_WB_ASSIST / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", @@ -804,7 +804,7 @@ "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY_WB_ASSIST." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", @@ -813,347 +813,348 @@ "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources." }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", "MetricName": "tma_info_thread_uoppi", "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", "MetricName": "tma_info_thread_uptb", "MetricThreshold": "tma_info_thread_uptb < 4 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "( 4 ) * tma_info_core_core_clks", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "( UOPS_EXECUTED.CORE / 2 / ( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ ) ) if #SMT_on else UOPS_EXECUTED.CORE / ( ( cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=0x1@ )", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "( CPU_CLK_UNHALTED.THREAD_ANY / 2 ) if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_UOPS_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipload", "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_UOPS_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipstore", "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipbranch", "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", "MetricName": "tma_info_inst_mix_ipcall", "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", "MetricThreshold": "tma_info_inst_mix_iptb < 4 * 2 + 1", - "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_lcp, tma_info_frontend_dsb_coverage, tma_fetch_bandwidth." + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / cpu@UOPS_RETIRED.RETIRE_SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( ( IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS ) )", "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", "MetricName": "tma_info_frontend_dsb_coverage", "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 4 > 0.35", - "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_lcp, tma_fetch_bandwidth, tma_info_inst_mix_iptb." + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmispredict", "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / ( ( ( UOPS_RETIRED.RETIRE_SLOTS ) / UOPS_ISSUED.ANY ) * BR_MISP_EXEC.INDIRECT )", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_indirect", "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_UOPS_RETIRED.L1_MISS + MEM_LOAD_UOPS_RETIRED.HIT_LFB )", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_UOPS_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_UOPS_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * OFFCORE_REQUESTS.DEMAND_RFO / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_UOPS_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_DURATION + DTLB_LOAD_MISSES.WALK_DURATION + DTLB_STORE_MISSES.WALK_DURATION ) / tma_info_core_core_clks", "MetricGroup": "Mem;MemoryTLB;Core_Metric", "MetricName": "tma_info_memory_tlb_page_walks_utilization", "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz].", + "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz]", "MetricExpr": "tma_info_system_socket_clks / 1e9 / tma_info_system_time", "MetricGroup": "SoC;System_Metric", "MetricName": "tma_info_system_uncore_frequency" }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / ( CPU_CLK_UNHALTED.REF_XCLK_ANY / 2 ) if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_utilization", "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "( 64 * ( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", "MetricName": "tma_info_system_dram_bw_use", "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_fb_full, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Average latency of data read request to external memory (in nanoseconds).", + "BriefDescription": "Average latency of data read request to external memory (in nanoseconds)", "MetricExpr": "( 1000000000 ) * ( cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=0x182@ ) / ( tma_info_system_socket_clks / tma_info_system_time )", "MetricGroup": "Mem;MemoryLat;SoC;NanoSeconds", "MetricName": "tma_info_system_mem_read_latency", "PublicDescription": "Average latency of data read request to external memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetches. ([RKL+]memory-controller only)." }, { - "BriefDescription": "Average number of parallel data read requests to external memory.", + "BriefDescription": "Average number of parallel data read requests to external memory", "MetricExpr": "cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filter_opc\\=0x182@ / cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filter_opc\\=0x182@", "MetricGroup": "Mem;MemoryBW;SoC;System_Metric", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches." }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "( power@energy\\-pkg@ * ( 61 ) + 15.6 * power@energy\\-ram@ ) / ( ( duration_time ) * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", "MetricName": "tma_info_system_mux", "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "cbox_0@event\\=0x0@", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", "MetricName": "tma_info_system_ipfarbranch", diff --git a/ICL/metrics/icelake_metrics.json b/ICL/metrics/icelake_metrics.json index a8c896ca..e5980af7 100644 --- a/ICL/metrics/icelake_metrics.json +++ b/ICL/metrics/icelake_metrics.json @@ -2,7 +2,7 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for 10th Generation Intel(R) Core(TM) Processor0", - "DatePublished": "11/12/2024", + "DatePublished": "11/15/2024", "Version": "1.0", "Legend": "", "TmaVersion": "5.01", @@ -107,7 +107,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Mispredictions > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Mispredictions" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Mispredictions > 20", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP", @@ -171,7 +179,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Big_Code > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Big_Code" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Big_Code > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB", @@ -332,7 +348,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Instruction_Fetch_BW" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", @@ -541,7 +565,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore", @@ -782,7 +814,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Latency" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;Mem;MemoryLat;Offcore", @@ -995,7 +1035,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Data_TLBs" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore", @@ -1188,7 +1236,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Synchronization > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Synchronization" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Synchronization > 10", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;LockCont;Mem;Offcore", @@ -1272,7 +1328,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Compute_Bound_Est" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20", + "ThresholdIssues": "$issueComp" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Cor", @@ -1473,7 +1537,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Irregular_Overhead > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Irregular_Overhead" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Irregular_Overhead > 10", + "ThresholdIssues": "$issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BvIO;Cor;Ret", @@ -1850,7 +1922,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Other_Bottlenecks" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;Cor;Offcore", @@ -1886,7 +1966,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Branching_Overhead > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Branching_Overhead" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Bottleneck_Branching_Overhead > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Ret", @@ -1966,7 +2054,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Useful_Work > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Useful_Work" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Useful_Work > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;Ret", @@ -2010,7 +2106,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;BvIO;TmaL1;PGO", @@ -2043,7 +2147,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Frontend;TmaL2", @@ -2072,7 +2188,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss", @@ -2105,7 +2237,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -2134,7 +2286,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -2163,7 +2335,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB", @@ -2196,7 +2384,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2225,7 +2433,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2262,7 +2490,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2299,7 +2551,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2332,7 +2608,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -2369,7 +2661,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Mispredicts_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP", @@ -2406,7 +2718,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Clears_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;MachineClears", @@ -2435,7 +2767,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Unknown_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat", @@ -2464,7 +2816,23 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC, $issueMS, $issueMV, $issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MicroSeq", @@ -2493,7 +2861,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LCP(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -2522,7 +2906,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchLat", @@ -2571,7 +2971,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;Frontend;TmaL2", @@ -2617,7 +3025,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -2663,7 +3083,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Decoder0_Alone(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -2696,7 +3132,23 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_......MITE_4wide(%) > 5 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MITE_4wide(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_......MITE_4wide(%) > 5 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -2742,7 +3194,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;FetchBW", @@ -2788,7 +3252,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....LSD(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LSD(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....LSD(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;LSD", @@ -2830,7 +3306,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 5 & b > 20", + "BaseFormula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -2878,7 +3366,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", @@ -2935,7 +3431,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2", @@ -2992,7 +3500,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;BrMispredicts", @@ -3049,7 +3573,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueMC, $issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2", @@ -3110,7 +3646,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Nukes(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;Machine_Clears", @@ -3154,7 +3706,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;TmaL1", @@ -3219,7 +3779,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20", + "BaseFormula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2", @@ -3252,7 +3824,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1, $issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -3293,7 +3881,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -3334,7 +3942,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3363,7 +3995,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3404,7 +4060,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3445,7 +4129,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3486,7 +4198,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3515,7 +4255,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Fwd_Blk(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3565,7 +4325,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L1_Latency_Dependency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -3610,7 +4390,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Lock_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Offcore", @@ -3651,7 +4451,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......Split_Loads(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Loads(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......Split_Loads(%) > 30", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3680,7 +4488,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......4K_Aliasing(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......4K_Aliasing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......4K_Aliasing(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3709,7 +4537,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......FB_Full(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FB_Full(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......FB_Full(%) > 30", + "ThresholdIssues": "$issueBW, $issueSL, $issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW", @@ -3758,7 +4594,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem", @@ -3808,7 +4660,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L2_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryLat", @@ -3841,7 +4713,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -3895,7 +4783,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Contested_Accesses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -3945,7 +4853,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Data_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;Offcore;Snoop", @@ -3995,7 +4923,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L3_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -4024,7 +4972,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......SQ_Full(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 30 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -4077,7 +5045,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -4106,7 +5090,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Bandwidth(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -4139,7 +5143,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat;Offcore", @@ -4168,7 +5192,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -4209,7 +5249,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO, $issueSL, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore", @@ -4251,7 +5311,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......False_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -4293,7 +5373,27 @@ "Category": "TMA", "CountDomain": "Core_Utilization", "Threshold": { - "Formula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4322,7 +5422,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Streaming_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBW;Offcore", @@ -4368,7 +5488,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -4414,7 +5554,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4456,7 +5620,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4510,7 +5698,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4564,7 +5780,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4618,7 +5862,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4683,7 +5955,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2;Compute", @@ -4712,7 +5996,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB", @@ -4741,7 +6041,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4774,7 +6094,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......INT_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4803,7 +6143,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;PortsUtil", @@ -4832,7 +6188,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Slow_Pause(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4897,7 +6273,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4926,7 +6318,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_0(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4955,7 +6367,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_........Mixing_Vectors(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Mixing_Vectors(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Mixing_Vectors(%) > 5", + "ThresholdIssues": "$issueMV" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4984,7 +6404,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_1(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -5013,7 +6453,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_2(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -5042,7 +6502,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_3m(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 40 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;PortsUtil", @@ -5096,7 +6576,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........ALU_Op_Utilization(%) > 40" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........ALU_Op_Utilization(%)" + } + ], + "Formula": "a > 40", + "BaseFormula": "metric_TMA_........ALU_Op_Utilization(%) > 40", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5138,7 +6626,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_0(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_0(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_0(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -5180,7 +6676,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_1(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_1(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_1(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5222,7 +6726,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_5(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_5(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_5(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5264,7 +6776,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_6(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_6(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_6(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5306,7 +6826,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Load_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Load_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5352,7 +6880,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Store_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Store_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5388,7 +6924,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Retiring(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 70 | b > 10", + "BaseFormula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;TmaL1", @@ -5453,7 +7001,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -5510,7 +7066,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 20 & b > 60", + "BaseFormula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -5555,7 +7123,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......X87_Use(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -5600,7 +7184,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Scalar(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5645,7 +7245,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5694,7 +7310,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5743,7 +7379,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5792,7 +7448,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_512b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5865,7 +7541,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Memory_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -5934,7 +7622,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Branch_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Branch_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -6027,7 +7727,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 30 & b > 60", + "BaseFormula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -6096,7 +7808,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Nop_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Pipeline", @@ -6161,7 +7889,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -6226,7 +7962,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Few_Uops_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6263,7 +8011,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueMC, $issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -6292,7 +8052,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Assists(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO", @@ -6321,7 +8097,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........FP_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........FP_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -6362,7 +8146,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......CISC(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6455,7 +8255,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_Core_Bound_Likely > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_Core_Bound_Likely" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Botlnk_Core_Bound_Likely > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;SMT", @@ -6482,6 +8290,11 @@ "BaseFormula": " inst_retired.any / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Summary", "LocateWith": "" @@ -6524,7 +8337,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UopPI > 1.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UopPI" + } + ], + "Formula": "a > 1.05", + "BaseFormula": "metric_TMA_Info_Thread_UopPI > 1.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret;Retire", @@ -6568,7 +8389,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UpTB < 5 * 1.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UpTB" + } + ], + "Formula": "a < 5 * 1.5", + "BaseFormula": "metric_TMA_Info_Thread_UpTB < 5 * 1.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW", @@ -6595,6 +8424,11 @@ "BaseFormula": " 1 / tma_info_thread_ipc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Mem", "LocateWith": "" @@ -6616,6 +8450,11 @@ "BaseFormula": " cpu_clk_unhalted.thread", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", "LocateWith": "" @@ -6637,6 +8476,11 @@ "BaseFormula": " topdown.slots:perf_metrics", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", "LocateWith": "" @@ -6671,6 +8515,11 @@ "BaseFormula": " tma_info_thread_slots / ( topdown.slots:percore / 2 ) if smt_on else 1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "SMT;TmaL1", "LocateWith": "" @@ -6696,6 +8545,11 @@ "BaseFormula": " uops_executed.thread / uops_issued.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline", "LocateWith": "" @@ -6734,6 +8588,11 @@ "BaseFormula": " inst_retired.any / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;SMT;TmaL1", "LocateWith": "" @@ -6788,6 +8647,11 @@ "BaseFormula": " ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.8_flops + 16 * fp_arith_inst_retired.512b_packed_single ) / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Flops", "LocateWith": "" @@ -6830,6 +8694,11 @@ "BaseFormula": " ( ( fp_arith_inst_retired.scalar ) + ( fp_arith_inst_retired.vector ) ) / ( 2 * tma_info_core_core_clks )", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -6855,6 +8724,11 @@ "BaseFormula": " uops_executed.thread / uops_executed.thread:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "LocateWith": "" @@ -6880,6 +8754,11 @@ "BaseFormula": " uops_executed.thread / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -6914,6 +8793,11 @@ "BaseFormula": " cpu_clk_unhalted.distributed if smt_on else tma_info_thread_clks", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "SMT", "LocateWith": "" @@ -6940,7 +8824,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpLoad < 3" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpLoad" + } + ], + "Formula": "a < 3", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpLoad < 3", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -6968,7 +8860,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpStore < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpStore" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpStore < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -6996,7 +8896,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpBranch < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpBranch" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpBranch < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;InsType", @@ -7024,7 +8932,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpCall < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpCall" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpCall < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", @@ -7052,7 +8968,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpTB < 5 * 2 + 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpTB" + } + ], + "Formula": "a < 5 * 2 + 1", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpTB < 5 * 2 + 1", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO", @@ -7079,6 +9003,11 @@ "BaseFormula": " br_inst_retired.all_branches / br_inst_retired.near_taken", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", "LocateWith": "" @@ -7121,7 +9050,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpFLOP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -7153,7 +9090,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -7181,7 +9126,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -7209,7 +9162,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -7241,7 +9202,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX128" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -7273,7 +9242,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX256" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -7305,7 +9282,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX512 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX512" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX512 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -7332,6 +9317,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / misc_retired.pause_inst", "Category": "TMA", "CountDomain": "Inst_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", "LocateWith": "" @@ -7358,7 +9348,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpSWPF" + } + ], + "Formula": "a < 100", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -7381,6 +9379,11 @@ "BaseFormula": " inst_retired.any", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;TmaL1", "LocateWith": "INST_RETIRED.PREC_DIST" @@ -7422,6 +9425,11 @@ "BaseFormula": " ( tma_retiring * tma_info_thread_slots ) / uops_retired.slots:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret", "LocateWith": "" @@ -7448,7 +9456,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Pipeline_IpAssist < 100000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Pipeline_IpAssist" + } + ], + "Formula": "a < 100000", + "BaseFormula": "metric_TMA_Info_Pipeline_IpAssist < 100000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", @@ -7488,6 +9504,11 @@ "BaseFormula": " uops_executed.thread / ( ( uops_executed.core_cycles_ge_1 / 2 ) if smt_on else uops_executed.thread:c1 )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "LocateWith": "" @@ -7513,6 +9534,11 @@ "BaseFormula": " lsd.uops / lsd.cycles_active", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7538,6 +9564,11 @@ "BaseFormula": " idq.dsb_uops / idq.dsb_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7563,6 +9594,11 @@ "BaseFormula": " idq.mite_uops / idq.mite_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7588,6 +9624,11 @@ "BaseFormula": " uops_issued.any / uops_issued.any:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7613,6 +9654,11 @@ "BaseFormula": " lsd.uops / ( uops_issued.any )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;LSD", "LocateWith": "" @@ -7639,7 +9685,19 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 5 > 0.35" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_DSB_Coverage" + }, + { + "Alias": "b", + "Value": "IPC" + } + ], + "Formula": "a < 0.7 & b / 5 > 0.35", + "BaseFormula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 5 > 0.35", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -7666,6 +9724,11 @@ "BaseFormula": " dsb2mite_switches.penalty_cycles / dsb2mite_switches.penalty_cycles:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss", "LocateWith": "" @@ -7691,6 +9754,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -7716,6 +9784,11 @@ "BaseFormula": " icache_16b.ifdata_stall / icache_16b.ifdata_stall:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", "LocateWith": "" @@ -7742,7 +9815,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret" + } + ], + "Formula": "a < 50", + "BaseFormula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -7769,6 +9850,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / baclears.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed", "LocateWith": "" @@ -7794,6 +9880,11 @@ "BaseFormula": " 1000 * frontend_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -7819,6 +9910,11 @@ "BaseFormula": " 1000 * l2_rqsts.code_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -7938,7 +10034,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Misses > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Misses" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Misses > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -8031,7 +10135,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Bandwidth" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -8095,7 +10207,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_IC_Misses > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_IC_Misses" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Info_Botlnk_IC_Misses > 5", + "ThresholdIssues": "$issueFL" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", @@ -8123,7 +10243,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMispredict" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts", @@ -8151,7 +10279,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8179,7 +10315,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8207,7 +10351,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Ret" + } + ], + "Formula": "a < 500", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8235,7 +10387,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect" + } + ], + "Formula": "a < 1000", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8338,6 +10498,11 @@ "BaseFormula": " tma_mispredictions * tma_info_thread_slots / ( 5 ) / br_misp_retired.all_branches / 100", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "$issueBM" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", "LocateWith": "" @@ -8367,6 +10532,11 @@ "BaseFormula": " int_misc.clears_count / ( br_misp_retired.all_branches + machine_clears.count )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", "LocateWith": "" @@ -8392,6 +10562,11 @@ "BaseFormula": " br_inst_retired.cond_ntaken / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", "LocateWith": "" @@ -8417,6 +10592,11 @@ "BaseFormula": " br_inst_retired.cond_taken / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", "LocateWith": "" @@ -8446,6 +10626,11 @@ "BaseFormula": " ( br_inst_retired.near_call + br_inst_retired.near_return ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8479,6 +10664,11 @@ "BaseFormula": " ( br_inst_retired.near_taken - br_inst_retired.cond_taken - 2 * br_inst_retired.near_call ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8520,6 +10710,11 @@ "BaseFormula": " 1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8549,6 +10744,11 @@ "BaseFormula": " l1d_pend_miss.pending / ( mem_load_retired.l1_miss + mem_load_retired.fb_hit )", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryLat", "LocateWith": "" @@ -8574,6 +10774,11 @@ "BaseFormula": " l1d_pend_miss.pending / l1d_pend_miss.pending_cycles", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryBW", "LocateWith": "" @@ -8599,6 +10804,11 @@ "BaseFormula": " 1000 * mem_load_retired.l1_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8624,6 +10834,11 @@ "BaseFormula": " 1000 * l2_rqsts.all_demand_data_rd / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8649,6 +10864,11 @@ "BaseFormula": " 1000 * mem_load_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;Backend;CacheHits", "LocateWith": "" @@ -8686,6 +10906,11 @@ "BaseFormula": " 1000 * ( ( offcore_requests.all_data_rd - offcore_requests.demand_data_rd ) + l2_rqsts.all_demand_miss + l2_rqsts.swpf_miss ) / tma_info_inst_mix_instructions", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem;Offcore", "LocateWith": "" @@ -8711,6 +10936,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8736,6 +10966,11 @@ "BaseFormula": " 1000 * l2_rqsts.rfo_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheMisses;Offcore", "LocateWith": "" @@ -8761,6 +10996,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8786,6 +11026,11 @@ "BaseFormula": " 1000 * mem_load_retired.l3_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -8811,6 +11056,11 @@ "BaseFormula": " 1000 * mem_load_retired.fb_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8837,6 +11087,11 @@ "BaseFormula": " 64 * l1d.replacement / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8863,6 +11118,11 @@ "BaseFormula": " 64 * l2_lines_in.all / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8889,6 +11149,11 @@ "BaseFormula": " 64 * longest_lat_cache.miss / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8915,6 +11180,11 @@ "BaseFormula": " 64 * offcore_requests.all_requests / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW;Offcore", "LocateWith": "" @@ -8962,7 +11232,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Page_Walks_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -8989,6 +11267,11 @@ "BaseFormula": " 1000 * itlb_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;MemoryTLB", "LocateWith": "" @@ -9014,6 +11297,11 @@ "BaseFormula": " 1000 * dtlb_load_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -9039,6 +11327,11 @@ "BaseFormula": " 1000 * dtlb_store_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -9065,6 +11358,11 @@ "BaseFormula": " tma_info_memory_l1d_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9091,6 +11389,11 @@ "BaseFormula": " tma_info_memory_l2_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9117,6 +11420,11 @@ "BaseFormula": " tma_info_memory_l3_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9143,6 +11451,11 @@ "BaseFormula": " tma_info_memory_l3_cache_access_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW;Offcore", "LocateWith": "" @@ -9173,7 +11486,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Useless_HWPF" + } + ], + "Formula": "a > 0.15", + "BaseFormula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -9200,6 +11521,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests.demand_data_rd", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Memory_Lat;Offcore", "LocateWith": "" @@ -9225,6 +11551,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests_outstanding.demand_data_rd:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -9250,6 +11581,11 @@ "BaseFormula": " offcore_requests_outstanding.all_data_rd / offcore_requests_outstanding.cycles_with_data_rd", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -9275,6 +11611,11 @@ "BaseFormula": " 1000 * mem_load_misc_retired.uc / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -9300,6 +11641,11 @@ "BaseFormula": " 1000 * sq_misc.bus_lock / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -9330,6 +11676,11 @@ "BaseFormula": " tma_info_system_cpus_utilized / num_cpus", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC;Summary", "LocateWith": "" @@ -9356,6 +11707,11 @@ "BaseFormula": " cpu_clk_unhalted.ref_tsc / tsc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", "LocateWith": "" @@ -9390,6 +11746,11 @@ "BaseFormula": " tma_info_system_turbo_utilization * tsc / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;Power", "LocateWith": "" @@ -9432,6 +11793,11 @@ "BaseFormula": " ( ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.8_flops + 16 * fp_arith_inst_retired.512b_packed_single ) / ( 1000000000 ) ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -9457,6 +11823,11 @@ "BaseFormula": " tma_info_thread_clks / cpu_clk_unhalted.ref_tsc", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -9495,6 +11866,11 @@ "BaseFormula": " core_power.lvl0_turbo_license / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -9534,7 +11910,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Power_License1_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Power_License1_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_System_Power_License1_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", @@ -9575,7 +11959,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Power_License2_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Power_License2_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_System_Power_License2_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", @@ -9611,6 +12003,11 @@ "BaseFormula": " 1 - cpu_clk_unhalted.one_thread_active / cpu_clk_unhalted.ref_distributed if smt_on else 0", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "SMT", "LocateWith": "" @@ -9637,7 +12034,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Kernel_Utilization > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Kernel_Utilization" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_System_Kernel_Utilization > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "OS", @@ -9664,6 +12069,11 @@ "BaseFormula": " cpu_clk_unhalted.thread_p:sup / inst_retired.any_p:sup", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "OS", "LocateWith": "" @@ -9694,6 +12104,11 @@ "BaseFormula": " 64 * ( unc_arb_trk_requests.all + unc_arb_coh_trk_requests.all ) / ( 1000000 ) / tma_info_system_time / 1000", "Category": "TMA", "CountDomain": "GB/sec", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "$issueBW" + }, "ResolutionLevels": "ARB", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC", "LocateWith": "" @@ -9720,6 +12135,11 @@ "BaseFormula": " unc_pkg_energy_status * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "PKG", "MetricGroup": "Power;SoC", "LocateWith": "" @@ -9742,7 +12162,15 @@ "Category": "TMA", "CountDomain": "Seconds", "Threshold": { - "Formula": "metric_TMA_Info_System_Time < 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Time" + } + ], + "Formula": "a < 1", + "BaseFormula": "metric_TMA_Info_System_Time < 1", + "ThresholdIssues": "" }, "ResolutionLevels": "SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -9770,7 +12198,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_MUX" + } + ], + "Formula": "a > 1.1 | a < 0.9", + "BaseFormula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -9793,6 +12229,11 @@ "BaseFormula": " unc_clock.socket", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CLOCK", "MetricGroup": "SoC", "LocateWith": "" @@ -9819,7 +12260,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_IpFarBranch < 1000000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_IpFarBranch" + } + ], + "Formula": "a < 1000000", + "BaseFormula": "metric_TMA_Info_System_IpFarBranch < 1000000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;OS", diff --git a/ICL/metrics/perf/icelake_metrics_perf.json b/ICL/metrics/perf/icelake_metrics_perf.json index e8060bb9..a39df88d 100644 --- a/ICL/metrics/perf/icelake_metrics_perf.json +++ b/ICL/metrics/perf/icelake_metrics_perf.json @@ -1,1580 +1,1744 @@ [ { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", + "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", + "MetricThreshold": "tma_bottleneck_mispredictions > 20", + "PublicDescription": "Total pipeline cost of Branch Misprediction related bottlenecks. Related metrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", + "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_big_code", + "MetricThreshold": "tma_bottleneck_big_code > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", + "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_mispredicts_resteers ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) ) ) - tma_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_instruction_fetch_bw", + "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) )", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", + "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", + "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks. Related metrics: tma_l3_hit_latency, tma_mem_latency.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", + "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs)", "MetricExpr": "100 * ( tma_memory_bound * ( tma_l1_bound / max( tma_memory_bound , ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) ) * ( tma_dtlb_load / max( tma_l1_bound , ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", + "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", + "PublicDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs). Related metrics: tma_dtlb_load, tma_dtlb_store.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", + "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors)", "MetricExpr": "100 * ( tma_memory_bound * ( ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", - "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", + "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", + "PublicDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", + "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation", "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", + "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", "PublicDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", + "BriefDescription": "Total pipeline cost of irregular execution (e.g", "MetricExpr": "100 * ( ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_mispredicts_resteers ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + tma_core_bound * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", - "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments).", + "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", + "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments). Related metrics: tma_microcode_sequencer, tma_ms_switches.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end.", + "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end", "MetricExpr": "100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", "MetricGroup": "BvOB;Cor;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_other_bottlenecks", + "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", "PublicDescription": "Total pipeline cost of remaining bottlenecks in the back-end. Examples include data-dependencies (Core Bound when Low ILP) and other unlisted memory-related stalls.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", + "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA", "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_branching_overhead", + "MetricThreshold": "tma_bottleneck_branching_overhead > 5", "PublicDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound).", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", + "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead", "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_useful_work", + "MetricThreshold": "tma_bottleneck_useful_work > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "topdown\\-fe\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_frontend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_frontend_bound > 0.15", "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound. Sample with: FRONTEND_RETIRED.LATENCY_GE_4.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( ( 5 ) * IDQ_UOPS_NOT_DELIVERED.CYCLES_0_UOPS_DELIV.CORE - INT_MISC.UOP_DROPPING ) / tma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", "MetricName": "tma_fetch_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_16, FRONTEND_RETIRED.LATENCY_GE_8.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache", "MetricExpr": "max( 0 , tma_icache_misses - tma_code_l2_miss )", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_CODE_RD / tma_info_thread_clks", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_itlb_misses - tma_code_stlb_miss )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + ( 10 ) * BACLEARS.ANY / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "( 1 - ( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "( 10 ) * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: BACLEARS.ANY." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 3 ) * IDQ.MS_SWITCHES / tma_info_thread_clks", - "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated", + "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES." + "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "DECODE.LCP / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." + "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", - "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS." + "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "max( 0 , tma_frontend_bound - tma_fetch_latency )", - "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", + "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2.", + "MetricThreshold": "tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.MITE_CYCLES_ANY - IDQ.MITE_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", "ScaleUnit": "100%", + "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder.", + "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder", "MetricExpr": "( cpu@INST_DECODED.DECODERS\\,cmask\\=0x1@ - cpu@INST_DECODED.DECODERS\\,cmask\\=0x2@ ) / tma_info_core_core_clks / 2", - "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated", + "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated;tma_issueD0", "MetricName": "tma_decoder0_alone", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder. Related metrics: tma_few_uops_instructions." }, { - "BriefDescription": "This metric represents fraction of cycles where (only) 4 uops were delivered by the MITE pipeline.", + "BriefDescription": "This metric represents fraction of cycles where (only) 4 uops were delivered by the MITE pipeline", "MetricExpr": "( cpu@IDQ.MITE_UOPS\\,cmask\\=0x4@ - cpu@IDQ.MITE_UOPS\\,cmask\\=0x5@ ) / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Core_Clocks", "MetricName": "tma_mite_4wide", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_mite_4wide > 0.05 & tma_mite > 0.1 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.DSB_CYCLES_ANY - IDQ.DSB_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", "ScaleUnit": "100%", + "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit", "MetricExpr": "( LSD.CYCLES_ACTIVE - LSD.CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "FetchBW;LSD;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_lsd", "ScaleUnit": "100%", + "MetricThreshold": "tma_lsd > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit. LSD typically does well sustaining Uop supply. However; in some rare cases; optimal uop-delivery could not be reached for small loops whose size (in terms of number of uops) does not suit well the LSD structure." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details", "MetricExpr": "cpu@IDQ.MS_UOPS\\,cmask\\=0x1@ / tma_info_core_core_clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_ms", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "max( 1 - ( tma_frontend_bound + tma_backend_bound + tma_retiring ) , 0 )", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_bad_speculation", "ScaleUnit": "100%", + "MetricThreshold": "tma_bad_speculation > 0.15", "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * tma_bad_speculation", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES.", + "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", + "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types)", "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Slots", "MetricName": "tma_other_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "max( 0 , tma_bad_speculation - tma_branch_mispredicts )", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", + "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT.", + "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering", "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_machine_clears_group;Slots", "MetricName": "tma_other_nukes", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "topdown\\-be\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) + ( ( 5 ) * INT_MISC.CLEARS_COUNT ) / tma_info_thread_slots", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_backend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_backend_bound > 0.2", "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( ( CYCLE_ACTIVITY.STALLS_MEM_ANY + EXE_ACTIVITY.BOUND_ON_STORES ) / ( CYCLE_ACTIVITY.STALLS_TOTAL + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) + EXE_ACTIVITY.BOUND_ON_STORES ) ) * tma_backend_bound", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_memory_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two).", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", + "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT." + "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "min( ( 7 ) * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS." + "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", "MetricName": "tma_load_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", "MetricName": "tma_load_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", + "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache", "MetricExpr": "min( 2 * ( MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETIRED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS ) * 20 / 100 , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_l1_latency_dependency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( 16 * max( 0 , MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS.ALL_RFO ) + ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) * ( ( 10 ) * L2_RQSTS.RFO_HIT + ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) ) / tma_info_thread_clks", - "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks", + "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS." + "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset.", + "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset", "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_4k_aliasing", "ScaleUnit": "100%", + "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset. False match is possible; which incur a few cycles load re-issue. However; the short re-issue duration is often hidden by the out-of-order core and HW optimizations; hence a user may safely ignore a high value of this metric unless it manages to propagate up into parent nodes of the hierarchy (e.g. to L1_Bound)." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "L1D_PEND_MISS.FB_FULL / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", + "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory)." + "MetricThreshold": "tma_fb_full > 0.3", + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( ( MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) ) ) / ( ( MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) ) ) + L1D_PEND_MISS.FB_FULL_PERIODS ) ) * ( ( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks )", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", + "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited)", "MetricExpr": "( 3.5 * tma_info_system_core_frequency ) * MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", "MetricName": "tma_l2_hit_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L2_MISS - CYCLE_ACTIVITY.STALLS_L3_MISS ) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( ( 32.5 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM ) + ( ( 27 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS." + "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( ( 27 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT." + "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( ( 12.5 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_RETIRED.L3_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT." + "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_bottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "L1D_PEND_MISS.L2_STALL / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clks + ( ( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks ) - tma_l2_bound )", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that)." + "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that)." + "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( L2_RQSTS.RFO_HIT * ( 10 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)." + "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( 32.5 * tma_info_system_core_frequency ) * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM." + "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization", + "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores.", + "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores", "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread_clks", - "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSmSt", "MetricName": "tma_streaming_stores", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE." + "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma_fb_full." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( ( 7 ) * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_STORE_MISSES.WALK_ACTIVE ) / tma_info_core_core_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES." + "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_load." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", "MetricName": "tma_store_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", "MetricName": "tma_store_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "max( 0 , tma_backend_bound - tma_memory_bound )", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_core_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations).", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", "ScaleUnit": "100%", + "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIVIDER_ACTIVE." }, { - "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active", "MetricExpr": "ARITH.FP_DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_fp_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_int_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations", "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks", - "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Clocks;tma_issueSO", "MetricName": "tma_serializing_operation", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD." + "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions", "MetricExpr": "140 * MISC_RETIRED.PAUSE_INST / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_slow_pause", "ScaleUnit": "100%", + "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions. Sample with: MISC_RETIRED.PAUSE_INST." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( tma_ports_utilized_0 * tma_info_thread_clks + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIVIDER_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - CYCLE_ACTIVITY.STALLS_MEM_ANY ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "cpu@EXE_ACTIVITY.3_PORTS_UTIL\\,umask\\=0x80@ / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", + "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles)", "MetricExpr": "UOPS_ISSUED.VECTOR_WIDTH_MISMATCH / UOPS_ISSUED.ANY", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks;tma_issueMV", "MetricName": "tma_mixing_vectors", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic." + "MetricThreshold": "tma_mixing_vectors > 0.05", + "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Sample with: UOPS_EXECUTED.CYCLES_GE_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + UOPS_DISPATCHED.PORT_5 + UOPS_DISPATCHED.PORT_6 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED.PORT_0 / tma_info_core_core_clks", - "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0." + "MetricThreshold": "tma_port_0 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_1 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1." + "MetricThreshold": "tma_port_1 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_5 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_5", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED.PORT_5." + "MetricThreshold": "tma_port_5 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED.PORT_5. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_6 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1." + "MetricThreshold": "tma_port_6 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "UOPS_DISPATCHED.PORT_2_3 / ( 2 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_load_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations. Sample with: UOPS_DISPATCHED.PORT_2_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_4_9 + UOPS_DISPATCHED.PORT_7_8 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations. Sample with: UOPS_DISPATCHED.PORT_7_8." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( topdown\\-retiring / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_retiring", "ScaleUnit": "100%", + "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.1", "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "max( 0 , tma_retiring - tma_heavy_operations )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_light_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", "ScaleUnit": "100%", + "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_512b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses", "MetricExpr": "tma_light_operations * MEM_INST_RETIRED.ANY / INST_RETIRED.ANY", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_memory_operations", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_operations > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions", "MetricExpr": "tma_light_operations * BR_INST_RETIRED.ALL_BRANCHES / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_branch_instructions", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_branch_instructions > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", + "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes", "MetricExpr": "max( 0 , tma_light_operations - ( tma_fp_arith + tma_memory_operations + tma_branch_instructions ) )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_other_light_ops", "ScaleUnit": "100%", + "MetricThreshold": "tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_light_operations_group;Slots", "MetricName": "tma_nop_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "tma_microcode_sequencer + tma_retiring * ( UOPS_DECODED.DEC0 - cpu@UOPS_DECODED.DEC0\\,cmask\\=0x1@ ) / IDQ.MITE_UOPS", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_heavy_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+]).", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops", "MetricExpr": "tma_heavy_operations - tma_microcode_sequencer", - "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueD0", "MetricName": "tma_few_uops_instructions", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions." + "MetricThreshold": "tma_few_uops_instructions > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions. Related metrics: tma_decoder0_alone." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) * IDQ.MS_UOPS / tma_info_thread_slots", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS." + "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( 34 ) * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists", "MetricExpr": "34 * ASSISTS.FP / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_fp_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_assists > 0.1", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", "ScaleUnit": "100%", + "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources." }, { - "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts.", + "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts", "MetricExpr": "100 * ( 1 - tma_core_bound / tma_ports_utilization if tma_core_bound < tma_ports_utilization else 1 ) if tma_info_system_smt_2t_utilization > 0.5 else 0", "MetricGroup": "Cor;SMT;Metric", - "MetricName": "tma_info_botlnk_l0_core_bound_likely" + "MetricName": "tma_info_botlnk_l0_core_bound_likely", + "MetricThreshold": "tma_info_botlnk_core_bound_likely > 0.5" }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", - "MetricName": "tma_info_thread_uoppi" + "MetricName": "tma_info_thread_uoppi", + "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", - "MetricName": "tma_info_thread_uptb" + "MetricName": "tma_info_thread_uptb", + "MetricThreshold": "tma_info_thread_uptb < 5 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor.", + "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor", "MetricExpr": "tma_info_thread_slots / ( TOPDOWN.SLOTS / 2 ) if #SMT_on else 1", "MetricGroup": "SMT;TmaL1;TopdownL1;tma_L1_group;Metric", "MetricName": "tma_info_thread_slots_utilization" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / tma_info_core_core_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "uops Executed per Cycle.", + "BriefDescription": "uops Executed per Cycle", "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", "MetricGroup": "Power;Metric", "MetricName": "tma_info_core_epc" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipload" + "MetricName": "tma_info_inst_mix_ipload", + "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipstore" + "MetricName": "tma_info_inst_mix_ipstore", + "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipbranch" + "MetricName": "tma_info_inst_mix_ipbranch", + "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipcall" + "MetricName": "tma_info_inst_mix_ipcall", + "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", - "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_iptb" + "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", + "MetricName": "tma_info_inst_mix_iptb", + "MetricThreshold": "tma_info_inst_mix_iptb < 5 * 2 + 1", + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", "MetricGroup": "Flops;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipflop" + "MetricName": "tma_info_inst_mix_ipflop", + "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", + "MetricThreshold": "tma_info_inst_mix_iparith < 10", "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", + "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", + "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx512", + "MetricThreshold": "tma_info_inst_mix_iparith_avx512 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / MISC_RETIRED.PAUSE_INST", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ippause" }, { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / SW_PREFETCH_ACCESS.ANY", "MetricGroup": "Prefetches;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipswpf" + "MetricName": "tma_info_inst_mix_ipswpf", + "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Instructions per a microcode Assist invocation.", + "BriefDescription": "Instructions per a microcode Assist invocation", "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", "MetricName": "tma_info_pipeline_ipassist", + "MetricThreshold": "tma_info_pipeline_ipassist < 100000", "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core.", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core", "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@ )", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Average number of uops fetched from LSD per cycle.", + "BriefDescription": "Average number of uops fetched from LSD per cycle", "MetricExpr": "LSD.UOPS / LSD.CYCLES_ACTIVE", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_lsd" }, { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", + "BriefDescription": "Average number of uops fetched from DSB per cycle", "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_dsb" }, { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", + "BriefDescription": "Average number of uops fetched from MITE per cycle", "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_mite" }, { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", + "BriefDescription": "Average number of Uops issued by front-end when it issued something", "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_frontend_fetch_upc" }, { - "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache).", + "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache)", "MetricExpr": "LSD.UOPS / ( UOPS_ISSUED.ANY )", "MetricGroup": "Fed;LSD;Metric", "MetricName": "tma_info_frontend_lsd_coverage" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( UOPS_ISSUED.ANY )", - "MetricGroup": "DSB;Fed;FetchBW;Metric", - "MetricName": "tma_info_frontend_dsb_coverage" + "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", + "MetricName": "tma_info_frontend_dsb_coverage", + "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 5 > 0.35", + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", + "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "DSBmiss;Metric", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Average Latency for L1 instruction cache misses.", + "BriefDescription": "Average Latency for L1 instruction cache misses", "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STALL\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "Fed;FetchLat;IcMiss;Metric", "MetricName": "tma_info_frontend_icache_miss_latency" }, { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", "MetricGroup": "DSBmiss;Fed;Inst_Metric", - "MetricName": "tma_info_frontend_ipdsb_miss_ret" + "MetricName": "tma_info_frontend_ipdsb_miss_ret", + "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache true code cacheline misses per kilo instruction", "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code" }, { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction", "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code_all" }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) )", - "MetricGroup": "DSBmiss;Fed;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_misses" + "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_misses", + "MetricThreshold": "tma_info_botlnk_dsb_misses > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) )", - "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_bandwidth" + "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", + "MetricThreshold": "tma_info_botlnk_dsb_bandwidth > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", + "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_ic_misses" + "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", + "MetricName": "tma_info_botlnk_l2_ic_misses", + "MetricThreshold": "tma_info_botlnk_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmispredict" + "MetricName": "tma_info_bad_spec_ipmispredict", + "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken" + "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", + "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_cond_taken" + "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", + "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_ret" + "MetricName": "tma_info_bad_spec_ipmisp_ret", + "MetricThreshold": "tma_info_bad_spec_ipmisp_ret < 500" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_indirect" + "MetricName": "tma_info_bad_spec_ipmisp_indirect", + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 5 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", - "MetricGroup": "Bad;BrMispredicts;Core_Metric", - "MetricName": "tma_info_bad_spec_branch_misprediction_cost" + "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", + "MetricName": "tma_info_bad_spec_branch_misprediction_cost", + "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_mispredicts_resteers." }, { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", + "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", "MetricGroup": "BrMispredicts;Metric", "MetricName": "tma_info_bad_spec_spec_clears_ratio" }, { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", + "BriefDescription": "Fraction of branches that are non-taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_nt" }, { - "BriefDescription": "Fraction of branches that are taken conditionals.", + "BriefDescription": "Fraction of branches that are taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_tk" }, { - "BriefDescription": "Fraction of branches that are CALL or RET.", + "BriefDescription": "Fraction of branches that are CALL or RET", "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_callret" }, { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", + "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_jump" }, { - "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group).", + "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group)", "MetricExpr": "1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_other_branches" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_RETIRED.L1_MISS + MEM_LOAD_RETIRED.FB_HIT )", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki_load" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( ( OFFCORE_REQUESTS.ALL_DATA_RD - OFFCORE_REQUESTS.DEMAND_DATA_RD ) + L2_RQSTS.ALL_DEMAND_MISS + L2_RQSTS.SWPF_MISS ) / tma_info_inst_mix_instructions", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * L2_RQSTS.RFO_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", + "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_fb_hpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Offcore;Metric", "MetricName": "tma_info_memory_l3_cache_access_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", - "MetricName": "tma_info_memory_tlb_page_walks_utilization" + "MetricName": "tma_info_memory_tlb_page_walks_utilization", + "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Fed;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_code_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_load_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_access_bw", "MetricGroup": "Mem;MemoryBW;Offcore;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_access_bw_2t" }, { - "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses.", + "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses", "MetricExpr": "L2_LINES_OUT.USELESS_HWPF / ( L2_LINES_OUT.SILENT + L2_LINES_OUT.NON_SILENT )", "MetricGroup": "Prefetches;Metric", - "MetricName": "tma_info_memory_prefetches_useless_hwpf" + "MetricName": "tma_info_memory_prefetches_useless_hwpf", + "MetricThreshold": "tma_info_memory_useless_hwpf > 0.15" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=0x1@", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", + "BriefDescription": "Un-cacheable retired load per kilo instruction", "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_uc_load_pki" }, { - "BriefDescription": "\"Bus lock\" per kilo instruction.", + "BriefDescription": "\"Bus lock\" per kilo instruction", "MetricExpr": "1000 * SQ_MISC.BUS_LOCK / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_bus_lock_pki" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0.", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0", "MetricExpr": "CORE_POWER.LVL0_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license0_utilization", "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0. This includes non-AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes." }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1.", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1", "MetricExpr": "CORE_POWER.LVL1_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license1_utilization", + "MetricThreshold": "tma_info_system_power_license1_utilization > 0.5", "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1. This includes high current AVX 256-bit instructions as well as low current AVX 512-bit instructions." }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX).", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX)", "MetricExpr": "CORE_POWER.LVL2_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license2_utilization", + "MetricThreshold": "tma_info_system_power_license2_utilization > 0.5", "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX). This includes high current AVX 512-bit instructions." }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", - "MetricName": "tma_info_system_kernel_utilization" + "MetricName": "tma_info_system_kernel_utilization", + "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "64 * ( UNC_ARB_TRK_REQUESTS.ALL + UNC_ARB_COH_TRK_REQUESTS.ALL ) / ( 1000000 ) / tma_info_system_time / 1000", - "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec", - "MetricName": "tma_info_system_dram_bw_use" + "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", + "MetricName": "tma_info_system_dram_bw_use", + "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "power@energy\\-pkg@ * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", - "MetricName": "tma_info_system_mux" + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "UNC_CLOCK.SOCKET", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", - "MetricName": "tma_info_system_ipfarbranch" + "MetricName": "tma_info_system_ipfarbranch", + "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" } ] \ No newline at end of file diff --git a/ICX/metrics/icelakex_metrics.json b/ICX/metrics/icelakex_metrics.json index 4f7bc59e..fcc4db46 100644 --- a/ICX/metrics/icelakex_metrics.json +++ b/ICX/metrics/icelakex_metrics.json @@ -2,7 +2,7 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for 3rd Generation Intel(R) Xeon(R) Processor Scalable Family based on Ice Lake microarchitecture0", - "DatePublished": "11/06/2024", + "DatePublished": "11/15/2024", "Version": "1.0", "Legend": "", "TmaVersion": "5.01", @@ -1227,7 +1227,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Mispredictions > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Mispredictions" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Mispredictions > 20", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP", @@ -1291,7 +1299,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Big_Code > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Big_Code" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Big_Code > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB", @@ -1444,7 +1460,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Instruction_Fetch_BW" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", @@ -1661,7 +1685,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore", @@ -1914,7 +1946,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Latency" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;Mem;MemoryLat;Offcore", @@ -2131,7 +2171,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Data_TLBs" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore", @@ -2360,7 +2408,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Synchronization > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Synchronization" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Synchronization > 10", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;LockCont;Mem;Offcore", @@ -2444,7 +2500,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Compute_Bound_Est" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20", + "ThresholdIssues": "$issueComp" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Cor", @@ -2637,7 +2701,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Irregular_Overhead > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Irregular_Overhead" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Irregular_Overhead > 10", + "ThresholdIssues": "$issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BvIO;Cor;Ret", @@ -3034,7 +3106,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Other_Bottlenecks" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;Cor;Offcore", @@ -3070,7 +3150,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Branching_Overhead > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Branching_Overhead" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Bottleneck_Branching_Overhead > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Ret", @@ -3150,7 +3238,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Useful_Work > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Useful_Work" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Useful_Work > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;Ret", @@ -3325,7 +3421,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -3354,7 +3470,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -3432,7 +3568,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3461,7 +3617,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3498,7 +3674,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3535,7 +3735,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -4208,7 +4432,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 5 & b > 20", + "BaseFormula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -4950,7 +5186,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4991,7 +5255,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5032,7 +5324,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5131,7 +5451,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L1_Latency_Dependency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -5446,7 +5786,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L2_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryLat", @@ -6714,7 +7074,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -6768,7 +7156,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -6822,7 +7238,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -6973,7 +7417,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -7006,7 +7470,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......INT_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -10396,6 +10880,11 @@ "BaseFormula": " uops_executed.thread / ( ( uops_executed.core_cycles_ge_1 / 2 ) if smt_on else uops_executed.thread:c1 )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "LocateWith": "" @@ -10581,6 +11070,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -12352,7 +12846,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Useless_HWPF" + } + ], + "Formula": "a > 0.15", + "BaseFormula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -13232,7 +13734,15 @@ "Category": "TMA", "CountDomain": "Seconds", "Threshold": { - "Formula": "metric_TMA_Info_System_Time < 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Time" + } + ], + "Formula": "a < 1", + "BaseFormula": "metric_TMA_Info_System_Time < 1", + "ThresholdIssues": "" }, "ResolutionLevels": "SOCKET, SYSTEM", "MetricGroup": "Summary", diff --git a/ICX/metrics/perf/icelakex_metrics_perf.json b/ICX/metrics/perf/icelakex_metrics_perf.json index 6d7f32ca..98c54963 100644 --- a/ICX/metrics/perf/icelakex_metrics_perf.json +++ b/ICX/metrics/perf/icelakex_metrics_perf.json @@ -1,139 +1,139 @@ [ { - "BriefDescription": "CPU operating frequency (in GHz).", - "MetricExpr": "( CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000", + "BriefDescription": "CPU operating frequency (in GHz)", + "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000", "MetricGroup": "", "MetricName": "cpu_operating_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Percentage of time spent in the active CPU power state C0.", + "BriefDescription": "Percentage of time spent in the active CPU power state C0", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "", "MetricName": "cpu_utilization", "ScaleUnit": "100%" }, { - "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles.", + "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles", "MetricExpr": "CPU_CLK_UNHALTED.THREAD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions", "MetricExpr": "MEM_INST_RETIRED.ALL_LOADS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "loads_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions", "MetricExpr": "MEM_INST_RETIRED.ALL_STORES / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "stores_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L1D.REPLACEMENT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L1_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions", "MetricExpr": "L2_RQSTS.ALL_CODE_RD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1_i_code_read_misses_with_prefetches_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L2_LINES_IN.ALL / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions", "MetricExpr": "L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_code_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_LLCPREFDATA + UNC_CHA_TOR_INSERTS.IA_MISS_DRD + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF ) / INST_RETIRED.ANY", + "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IA_MISS_LLCPREFDATA + UNC_CHA_TOR_INSERTS.IA_MISS_DRD + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF) / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_data_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_CRD + UNC_CHA_TOR_INSERTS.IA_MISS_CRD_PREF ) / INST_RETIRED.ANY", + "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IA_MISS_CRD + UNC_CHA_TOR_INSERTS.IA_MISS_CRD_PREF) / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_code_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_latency", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to local memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to local memory in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_latency_for_local_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to remote memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to remote memory in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_latency_for_remote_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to Intel(R) Optane(TM) Persistent Memory(PMEM) in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PMM ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to Intel(R) Optane(TM) Persistent Memory(PMEM) in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PMM) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_to_pmem_latency", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to DRAM in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to DRAM in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_to_dram_latency", "ScaleUnit": "1ns" }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_2nd_level_mpi", @@ -141,7 +141,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the ITLB (Instruction TLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED_2M_4M / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_2nd_level_large_page_mpi", @@ -149,7 +149,7 @@ "PublicDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the Instruction Translation Lookaside Buffer (ITLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions", "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_2nd_level_load_mpi", @@ -157,7 +157,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions", "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_2nd_level_2mb_large_page_load_mpi", @@ -165,7 +165,7 @@ "PublicDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the Data Translation Lookaside Buffer (DTLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions", "MetricExpr": "DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_2nd_level_store_mpi", @@ -173,282 +173,299 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL ) / ( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE )", + "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL) / (UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_local_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE ) / ( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE )", + "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE) / (UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_remote_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Uncore operating frequency in GHz.", - "MetricExpr": "( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) / 1000000000) / duration_time", + "BriefDescription": "Uncore operating frequency in GHz", + "MetricExpr": "(UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) / 1000000000) / duration_time", "MetricGroup": "", "MetricName": "uncore_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data transmit bandwidth (MB/sec).", - "MetricExpr": "( UNC_UPI_TxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", + "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data transmit bandwidth (MB/sec)", + "MetricExpr": "(UNC_UPI_TxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", "MetricGroup": "", "MetricName": "upi_data_transmit_bw", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory read bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory read bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory write bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory write bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory bandwidth (MB/sec)", + "MetricExpr": "((UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_total", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Memory write bandwidth (MB/sec) caused by directory updates; includes DDR and Intel(R) Optane(TM) Persistent Memory(PMEM).", - "MetricExpr": "(( UNC_CHA_DIR_UPDATE.HA + UNC_CHA_DIR_UPDATE.TOR + UNC_M2M_DIRECTORY_UPDATE.ANY ) * 64 / 1000000) / duration_time", + "BriefDescription": "Memory write bandwidth (MB/sec) caused by directory updates; includes DDR and Intel(R) Optane(TM) Persistent Memory(PMEM)", + "MetricExpr": "((UNC_CHA_DIR_UPDATE.HA + UNC_CHA_DIR_UPDATE.TOR + UNC_M2M_DIRECTORY_UPDATE.ANY) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_extra_write_bw_due_to_directory_updates", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory read bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_PMM_RPQ_INSERTS * 64 / 1000000) / duration_time", + "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory read bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_PMM_RPQ_INSERTS * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "pmem_memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory write bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_PMM_WPQ_INSERTS * 64 / 1000000) / duration_time", + "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory write bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_PMM_WPQ_INSERTS * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "pmem_memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_PMM_RPQ_INSERTS + UNC_M_PMM_WPQ_INSERTS ) * 64 / 1000000) / duration_time", + "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory bandwidth (MB/sec)", + "MetricExpr": "((UNC_M_PMM_RPQ_INSERTS + UNC_M_PMM_WPQ_INSERTS) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "pmem_memory_bandwidth_total", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_HIT_PCIRDCUR + UNC_CHA_TOR_INSERTS.IO_MISS_PCIRDCUR ) * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_HIT_PCIRDCUR + UNC_CHA_TOR_INSERTS.IO_MISS_PCIRDCUR) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_HIT_ITOM + UNC_CHA_TOR_INSERTS.IO_MISS_ITOM + UNC_CHA_TOR_INSERTS.IO_HIT_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_MISS_ITOMCACHENEAR ) * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_HIT_ITOM + UNC_CHA_TOR_INSERTS.IO_MISS_ITOM + UNC_CHA_TOR_INSERTS.IO_HIT_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_MISS_ITOMCACHENEAR) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.DSB_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.DSB_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_decoded_icache", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MITE_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MITE_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_legacy_decode_pipeline", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MS_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MS_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_microcode_sequencer", "ScaleUnit": "100%" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the local CPU socket.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_LOCAL * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the local CPU socket", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_LOCAL * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read_local", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from a remote CPU socket.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_REMOTE * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from a remote CPU socket", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_REMOTE * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read_remote", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the local CPU socket.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_ITOM_LOCAL + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_LOCAL ) * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the local CPU socket", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_ITOM_LOCAL + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_LOCAL) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write_local", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to a remote CPU socket.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_ITOM_REMOTE + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_REMOTE ) * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to a remote CPU socket", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_ITOM_REMOTE + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_REMOTE) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write_remote", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to local memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.READS_LOCAL * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to local memory", + "MetricExpr": "(UNC_CHA_REQUESTS.READS_LOCAL * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_local_memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to local memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.WRITES_LOCAL * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to local memory", + "MetricExpr": "(UNC_CHA_REQUESTS.WRITES_LOCAL * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_local_memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to remote memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.READS_REMOTE * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to remote memory", + "MetricExpr": "(UNC_CHA_REQUESTS.READS_REMOTE * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_remote_memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to remote memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.WRITES_REMOTE * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to remote memory", + "MetricExpr": "(UNC_CHA_REQUESTS.WRITES_REMOTE * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_remote_memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data receive bandwidth (MB/sec).", - "MetricExpr": "( UNC_UPI_RxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", + "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data receive bandwidth (MB/sec)", + "MetricExpr": "(UNC_UPI_RxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", "MetricGroup": "", "MetricName": "upi_data_receive_bw", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", + "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", + "MetricThreshold": "tma_bottleneck_mispredictions > 20", + "PublicDescription": "Total pipeline cost of Branch Misprediction related bottlenecks. Related metrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", + "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_big_code", + "MetricThreshold": "tma_bottleneck_big_code > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", + "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_mispredicts_resteers ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_ms ) ) ) ) - tma_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_instruction_fetch_bw", + "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) )", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", + "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", + "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks. Related metrics: tma_l3_hit_latency, tma_mem_latency.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", + "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs)", "MetricExpr": "100 * ( tma_memory_bound * ( tma_l1_bound / max( tma_memory_bound , ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) ) * ( tma_dtlb_load / max( tma_l1_bound , ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", + "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", + "PublicDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs). Related metrics: tma_dtlb_load, tma_dtlb_store.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", + "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors)", "MetricExpr": "100 * ( tma_memory_bound * ( ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) * tma_remote_cache / ( tma_local_mem + tma_remote_mem + tma_remote_cache ) + ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", - "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", + "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", + "PublicDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears, tma_remote_cache.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", + "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation", "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", + "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", "PublicDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", + "BriefDescription": "Total pipeline cost of irregular execution (e.g", "MetricExpr": "100 * ( ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_mispredicts_resteers ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_ms ) ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + tma_core_bound * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", - "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments).", + "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", + "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments). Related metrics: tma_microcode_sequencer, tma_ms_switches.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end.", + "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end", "MetricExpr": "100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", "MetricGroup": "BvOB;Cor;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_other_bottlenecks", + "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", "PublicDescription": "Total pipeline cost of remaining bottlenecks in the back-end. Examples include data-dependencies (Core Bound when Low ILP) and other unlisted memory-related stalls.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", + "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA", "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_branching_overhead", + "MetricThreshold": "tma_bottleneck_branching_overhead > 5", "PublicDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound).", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", + "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead", "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_useful_work", + "MetricThreshold": "tma_bottleneck_useful_work > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "topdown\\-fe\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_frontend_bound", @@ -459,7 +476,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( ( 5 ) * IDQ_UOPS_NOT_DELIVERED.CYCLES_0_UOPS_DELIV.CORE - INT_MISC.UOP_DROPPING ) / tma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", "MetricName": "tma_fetch_latency", @@ -469,7 +486,7 @@ "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", @@ -478,21 +495,23 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache", "MetricExpr": "max( 0 , tma_icache_misses - tma_code_l2_miss )", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_CODE_RD / tma_info_thread_clks", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", @@ -501,53 +520,57 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_itlb_misses - tma_code_stlb_miss )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + ( 10 ) * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", "ScaleUnit": "100%", "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_store_latency, tma_l3_hit_latency." + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", "ScaleUnit": "100%", "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "( 1 - ( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", @@ -556,7 +579,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "( 10 ) * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", @@ -565,44 +588,44 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: BACLEARS.ANY." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 3 ) * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_clears_resteers, tma_mixing_vectors, tma_l1_bound, tma_serializing_operation, tma_machine_clears, tma_microcode_sequencer." + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "DECODE.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_info_botlnk_l2_dsb_bandwidth, tma_fetch_bandwidth, tma_info_inst_mix_iptb, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_dsb_switches." + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_inst_mix_iptb, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp, tma_info_botlnk_l2_dsb_bandwidth." + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "max( 0 , tma_frontend_bound - tma_fetch_latency )", "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", "MetricThreshold": "tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_info_botlnk_l2_dsb_bandwidth, tma_info_inst_mix_iptb, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp, tma_dsb_switches.", + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.MITE_CYCLES_ANY - IDQ.MITE_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", @@ -611,7 +634,7 @@ "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder.", + "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder", "MetricExpr": "( cpu@INST_DECODED.DECODERS\\,cmask\\=0x1@ - cpu@INST_DECODED.DECODERS\\,cmask\\=0x2@ ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated;tma_issueD0", "MetricName": "tma_decoder0_alone", @@ -620,7 +643,7 @@ "PublicDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder. Related metrics: tma_few_uops_instructions." }, { - "BriefDescription": "This metric represents fraction of cycles where (only) 4 uops were delivered by the MITE pipeline.", + "BriefDescription": "This metric represents fraction of cycles where (only) 4 uops were delivered by the MITE pipeline", "MetricExpr": "( cpu@IDQ.MITE_UOPS\\,cmask\\=0x4@ - cpu@IDQ.MITE_UOPS\\,cmask\\=0x5@ ) / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Core_Clocks", "MetricName": "tma_mite_4wide", @@ -628,7 +651,7 @@ "MetricThreshold": "tma_mite_4wide > 0.05 & tma_mite > 0.1 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.DSB_CYCLES_ANY - IDQ.DSB_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", @@ -637,14 +660,15 @@ "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details", "MetricExpr": "cpu@IDQ.MS_UOPS\\,cmask\\=0x1@ / tma_info_core_core_clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_ms", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "max( 1 - ( tma_frontend_bound + tma_backend_bound + tma_retiring ) , 0 )", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_bad_speculation", @@ -655,17 +679,17 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * tma_bad_speculation", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", + "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types)", "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Slots", "MetricName": "tma_other_mispredicts", @@ -673,17 +697,17 @@ "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "max( 0 , tma_bad_speculation - tma_branch_mispredicts )", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_clears_resteers, tma_l1_bound, tma_remote_cache, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_microcode_sequencer, tma_ms_switches.", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches, tma_remote_cache.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering", "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_machine_clears_group;Slots", "MetricName": "tma_other_nukes", @@ -691,7 +715,7 @@ "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "topdown\\-be\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) + ( ( 5 ) * INT_MISC.CLEARS_COUNT ) / tma_info_thread_slots", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_backend_bound", @@ -702,7 +726,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( ( CYCLE_ACTIVITY.STALLS_MEM_ANY + EXE_ACTIVITY.BOUND_ON_STORES ) / ( CYCLE_ACTIVITY.STALLS_TOTAL + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) + EXE_ACTIVITY.BOUND_ON_STORES ) ) * tma_backend_bound", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_memory_bound", @@ -712,7 +736,7 @@ "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", @@ -721,16 +745,16 @@ "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "min( ( 7 ) * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_dtlb_store." + "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", "MetricName": "tma_load_stlb_hit", @@ -738,7 +762,7 @@ "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", "MetricName": "tma_load_stlb_miss", @@ -746,28 +770,31 @@ "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", @@ -776,15 +803,16 @@ "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", + "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache", "MetricExpr": "min( 2 * ( MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETIRED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS ) * 20 / 100 , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_l1_latency_dependency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( 16 * max( 0 , MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS.ALL_RFO ) + ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) * ( ( 10 ) * L2_RQSTS.RFO_HIT + ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) ) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", @@ -793,7 +821,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", @@ -802,7 +830,7 @@ "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset.", + "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset", "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_4k_aliasing", @@ -811,16 +839,16 @@ "PublicDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset. False match is possible; which incur a few cycles load re-issue. However; the short re-issue duration is often hidden by the out-of-order core and HW optimizations; hence a user may safely ignore a high value of this metric unless it manages to propagate up into parent nodes of the hierarchy (e.g. to L1_Bound)." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "L1D_PEND_MISS.FB_FULL / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", "MetricThreshold": "tma_fb_full > 0.3", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_store_latency, tma_streaming_stores, tma_info_system_dram_bw_use, tma_sq_full, tma_mem_bandwidth." + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( ( MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) ) ) / ( ( MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) ) ) + L1D_PEND_MISS.FB_FULL_PERIODS ) ) * ( ( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks )", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", @@ -829,15 +857,16 @@ "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", + "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited)", "MetricExpr": "( 4 * tma_info_system_core_frequency ) * MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", "MetricName": "tma_l2_hit_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L2_MISS - CYCLE_ACTIVITY.STALLS_L3_MISS ) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", @@ -846,43 +875,43 @@ "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( ( 48 * tma_info_system_core_frequency ) - ( 4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) + ( ( 47.5 * tma_info_system_core_frequency ) - ( 4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_data_sharing, tma_machine_clears, tma_remote_cache, tma_false_sharing." + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( ( 47.5 * tma_info_system_core_frequency ) - ( 4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * ( 1 - ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_false_sharing, tma_machine_clears, tma_remote_cache, tma_contested_accesses." + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( ( 23 * tma_info_system_core_frequency ) - ( 4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_RETIRED.L3_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_store_latency, tma_branch_resteers, tma_mem_latency." + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_bottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "L1D_PEND_MISS.L2_STALL / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", "ScaleUnit": "100%", "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_mem_bandwidth, tma_info_system_dram_bw_use, tma_fb_full." + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clks + ( ( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks ) - tma_l2_bound )", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", @@ -891,25 +920,25 @@ "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_sq_full, tma_info_system_dram_bw_use, tma_fb_full." + "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_l3_hit_latency." + "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory", "MetricExpr": "( ( 66.5 * tma_info_system_core_frequency ) - ( 23 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_local_mem", @@ -918,7 +947,7 @@ "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory. Caching will improve the latency and increase performance. Sample with: MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory", "MetricExpr": "( ( 131 * tma_info_system_core_frequency ) - ( 23 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_remote_mem", @@ -927,16 +956,16 @@ "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues", "MetricExpr": "( ( ( 120 * tma_info_system_core_frequency ) - ( 23 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM + ( ( 120 * tma_info_system_core_frequency ) - ( 23 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_remote_cache", "ScaleUnit": "100%", "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_data_sharing, tma_machine_clears, tma_false_sharing, tma_contested_accesses." + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", @@ -945,25 +974,25 @@ "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( L2_RQSTS.RFO_HIT * ( 10 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_lock_latency, tma_branch_resteers, tma_fb_full, tma_l3_hit_latency." + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( ( 120 * tma_info_system_core_frequency ) * cpu@OCR.DEMAND_RFO.L3_MISS\\,offcore_rsp\\=0x103b800002@ + ( 48 * tma_info_system_core_frequency ) * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM ) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related metrics: tma_data_sharing, tma_machine_clears, tma_remote_cache, tma_contested_accesses." + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", @@ -972,7 +1001,7 @@ "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores.", + "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores", "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread_clks", "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSmSt", "MetricName": "tma_streaming_stores", @@ -981,16 +1010,16 @@ "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma_fb_full." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( ( 7 ) * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_STORE_MISSES.WALK_ACTIVE ) / tma_info_core_core_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_dtlb_load." + "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_load." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", "MetricName": "tma_store_stlb_hit", @@ -998,7 +1027,7 @@ "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", "MetricName": "tma_store_stlb_miss", @@ -1006,28 +1035,31 @@ "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "max( 0 , tma_backend_bound - tma_memory_bound )", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_core_bound", @@ -1037,7 +1069,7 @@ "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", @@ -1046,21 +1078,23 @@ "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIVIDER_ACTIVE." }, { - "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active", "MetricExpr": "ARITH.FP_DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_fp_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_int_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations", "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Clocks;tma_issueSO", "MetricName": "tma_serializing_operation", @@ -1069,7 +1103,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions", "MetricExpr": "37 * MISC_RETIRED.PAUSE_INST / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_slow_pause", @@ -1078,7 +1112,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions. Sample with: MISC_RETIRED.PAUSE_INST." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( tma_ports_utilized_0 * tma_info_thread_clks + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIVIDER_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - CYCLE_ACTIVITY.STALLS_MEM_ANY ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", @@ -1087,7 +1121,7 @@ "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "cpu@EXE_ACTIVITY.3_PORTS_UTIL\\,umask\\=0x80@ / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", @@ -1096,7 +1130,7 @@ "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", + "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles)", "MetricExpr": "UOPS_ISSUED.VECTOR_WIDTH_MISMATCH / UOPS_ISSUED.ANY", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks;tma_issueMV", "MetricName": "tma_mixing_vectors", @@ -1105,7 +1139,7 @@ "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", @@ -1114,16 +1148,16 @@ "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_port_1, tma_fp_vector_512b, tma_port_5, tma_port_0, tma_fp_vector, tma_fp_scalar, tma_fp_vector_256b, tma_port_6, tma_fp_vector_128b." + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", @@ -1132,7 +1166,7 @@ "PublicDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Sample with: UOPS_EXECUTED.CYCLES_GE_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + UOPS_DISPATCHED.PORT_5 + UOPS_DISPATCHED.PORT_6 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", @@ -1140,43 +1174,43 @@ "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED.PORT_0 / tma_info_core_core_clks", "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", "ScaleUnit": "100%", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_port_1, tma_fp_vector_512b, tma_port_5, tma_fp_vector, tma_fp_scalar, tma_fp_vector_256b, tma_port_6, tma_ports_utilized_2, tma_fp_vector_128b." + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_1 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", "ScaleUnit": "100%", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_vector_512b, tma_port_5, tma_port_0, tma_fp_vector, tma_fp_scalar, tma_fp_vector_256b, tma_port_6, tma_ports_utilized_2, tma_fp_vector_128b." + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_5 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_5", "ScaleUnit": "100%", "MetricThreshold": "tma_port_5 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED.PORT_5. Related metrics: tma_port_1, tma_fp_vector_512b, tma_port_0, tma_fp_vector, tma_fp_scalar, tma_fp_vector_256b, tma_port_6, tma_ports_utilized_2, tma_fp_vector_128b." + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED.PORT_5. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_6 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", "ScaleUnit": "100%", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_port_1, tma_fp_vector_512b, tma_port_5, tma_port_0, tma_fp_vector, tma_fp_scalar, tma_fp_vector_256b, tma_ports_utilized_2, tma_fp_vector_128b." + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "UOPS_DISPATCHED.PORT_2_3 / ( 2 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", @@ -1185,7 +1219,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations. Sample with: UOPS_DISPATCHED.PORT_2_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_4_9 + UOPS_DISPATCHED.PORT_7_8 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", @@ -1194,7 +1228,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations. Sample with: UOPS_DISPATCHED.PORT_7_8." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( topdown\\-retiring / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_retiring", @@ -1205,7 +1239,7 @@ "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "max( 0 , tma_retiring - tma_heavy_operations )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_light_operations", @@ -1215,7 +1249,7 @@ "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", @@ -1224,7 +1258,7 @@ "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", @@ -1233,52 +1267,52 @@ "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", "ScaleUnit": "100%", "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_port_1, tma_fp_vector_512b, tma_port_5, tma_port_0, tma_fp_vector, tma_fp_vector_256b, tma_port_6, tma_ports_utilized_2, tma_fp_vector_128b." + "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", "ScaleUnit": "100%", "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_port_1, tma_fp_vector_512b, tma_port_5, tma_port_0, tma_fp_scalar, tma_fp_vector_256b, tma_port_6, tma_ports_utilized_2, tma_fp_vector_128b." + "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_port_1, tma_fp_vector_512b, tma_port_5, tma_port_0, tma_fp_vector, tma_fp_scalar, tma_fp_vector_256b, tma_port_6, tma_ports_utilized_2." + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", "ScaleUnit": "100%", "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_port_1, tma_fp_vector_512b, tma_port_5, tma_port_0, tma_fp_vector, tma_fp_scalar, tma_port_6, tma_ports_utilized_2, tma_fp_vector_128b." + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_512b", "ScaleUnit": "100%", "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting. Related metrics: tma_port_1, tma_port_5, tma_port_0, tma_fp_vector, tma_fp_scalar, tma_fp_vector_256b, tma_port_6, tma_ports_utilized_2, tma_fp_vector_128b." + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses", "MetricExpr": "tma_light_operations * MEM_INST_RETIRED.ANY / INST_RETIRED.ANY", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_memory_operations", @@ -1286,7 +1320,7 @@ "MetricThreshold": "tma_memory_operations > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions", "MetricExpr": "tma_light_operations * BR_INST_RETIRED.ALL_BRANCHES / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_branch_instructions", @@ -1294,7 +1328,7 @@ "MetricThreshold": "tma_branch_instructions > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", + "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes", "MetricExpr": "max( 0 , tma_light_operations - ( tma_fp_arith + tma_memory_operations + tma_branch_instructions ) )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_other_light_ops", @@ -1303,7 +1337,7 @@ "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_light_operations_group;Slots", "MetricName": "tma_nop_instructions", @@ -1312,7 +1346,7 @@ "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "tma_microcode_sequencer + tma_retiring * ( UOPS_DECODED.DEC0 - cpu@UOPS_DECODED.DEC0\\,cmask\\=0x1@ ) / IDQ.MITE_UOPS", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_heavy_operations", @@ -1322,7 +1356,7 @@ "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops", "MetricExpr": "tma_heavy_operations - tma_microcode_sequencer", "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueD0", "MetricName": "tma_few_uops_instructions", @@ -1331,16 +1365,16 @@ "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions. Related metrics: tma_decoder0_alone." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) * IDQ.MS_UOPS / tma_info_thread_slots", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_l1_bound, tma_machine_clears, tma_ms_switches, tma_clears_resteers." + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( 34 ) * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", @@ -1349,7 +1383,7 @@ "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists", "MetricExpr": "34 * ASSISTS.FP / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_fp_assists", @@ -1358,7 +1392,7 @@ "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", @@ -1367,152 +1401,152 @@ "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources." }, { - "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts.", + "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts", "MetricExpr": "100 * ( 1 - tma_core_bound / tma_ports_utilization if tma_core_bound < tma_ports_utilization else 1 ) if tma_info_system_smt_2t_utilization > 0.5 else 0", "MetricGroup": "Cor;SMT;Metric", "MetricName": "tma_info_botlnk_l0_core_bound_likely", "MetricThreshold": "tma_info_botlnk_core_bound_likely > 0.5" }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", "MetricName": "tma_info_thread_uoppi", "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", "MetricName": "tma_info_thread_uptb", "MetricThreshold": "tma_info_thread_uptb < 5 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor.", + "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor", "MetricExpr": "tma_info_thread_slots / ( TOPDOWN.SLOTS / 2 ) if #SMT_on else 1", "MetricGroup": "SMT;TmaL1;TopdownL1;tma_L1_group;Metric", "MetricName": "tma_info_thread_slots_utilization" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / tma_info_core_core_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "uops Executed per Cycle.", + "BriefDescription": "uops Executed per Cycle", "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", "MetricGroup": "Power;Metric", "MetricName": "tma_info_core_epc" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipload", "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipstore", "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipbranch", "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", "MetricName": "tma_info_inst_mix_ipcall", "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", "MetricThreshold": "tma_info_inst_mix_iptb < 5 * 2 + 1", - "PublicDescription": "Instructions per taken branch. Related metrics: tma_info_botlnk_l2_dsb_bandwidth, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp, tma_dsb_switches." + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipflop", "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", @@ -1520,7 +1554,7 @@ "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", @@ -1528,7 +1562,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", @@ -1536,7 +1570,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", @@ -1544,7 +1578,7 @@ "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", @@ -1552,7 +1586,7 @@ "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx512", @@ -1560,33 +1594,33 @@ "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / MISC_RETIRED.PAUSE_INST", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ippause" }, { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / SW_PREFETCH_ACCESS.ANY", "MetricGroup": "Prefetches;Inst_Metric", "MetricName": "tma_info_inst_mix_ipswpf", "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Instructions per a microcode Assist invocation.", + "BriefDescription": "Instructions per a microcode Assist invocation", "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", "MetricName": "tma_info_pipeline_ipassist", @@ -1594,98 +1628,98 @@ "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core.", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core", "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@ )", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", + "BriefDescription": "Average number of uops fetched from DSB per cycle", "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_dsb" }, { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", + "BriefDescription": "Average number of uops fetched from MITE per cycle", "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_mite" }, { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", + "BriefDescription": "Average number of Uops issued by front-end when it issued something", "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_frontend_fetch_upc" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( UOPS_ISSUED.ANY )", "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", "MetricName": "tma_info_frontend_dsb_coverage", "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 5 > 0.35", - "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_info_botlnk_l2_dsb_bandwidth, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp, tma_dsb_switches." + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", + "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "DSBmiss;Metric", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Average Latency for L1 instruction cache misses.", + "BriefDescription": "Average Latency for L1 instruction cache misses", "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STALL\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "Fed;FetchLat;IcMiss;Metric", "MetricName": "tma_info_frontend_icache_miss_latency" }, { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", "MetricGroup": "DSBmiss;Fed;Inst_Metric", "MetricName": "tma_info_frontend_ipdsb_miss_ret", "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache true code cacheline misses per kilo instruction", "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code" }, { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction", "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code_all" }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb + tma_ms ) )", "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_dsb_misses > 10", - "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_info_botlnk_l2_dsb_bandwidth, tma_fetch_bandwidth, tma_info_inst_mix_iptb, tma_info_frontend_dsb_coverage, tma_lcp, tma_dsb_switches." + "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb + tma_ms ) ) )", "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_dsb_bandwidth > 10", - "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_fetch_bandwidth, tma_info_inst_mix_iptb, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp, tma_dsb_switches." + "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", + "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", @@ -1693,317 +1727,318 @@ "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmispredict", "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_ret", "MetricThreshold": "tma_info_bad_spec_ipmisp_ret < 500" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_indirect", "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 5 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", "MetricName": "tma_info_bad_spec_branch_misprediction_cost", - "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_branch_mispredicts, tma_mispredicts_resteers." + "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_mispredicts_resteers." }, { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", + "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", "MetricGroup": "BrMispredicts;Metric", "MetricName": "tma_info_bad_spec_spec_clears_ratio" }, { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", + "BriefDescription": "Fraction of branches that are non-taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_nt" }, { - "BriefDescription": "Fraction of branches that are taken conditionals.", + "BriefDescription": "Fraction of branches that are taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_tk" }, { - "BriefDescription": "Fraction of branches that are CALL or RET.", + "BriefDescription": "Fraction of branches that are CALL or RET", "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_callret" }, { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", + "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_jump" }, { - "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group).", + "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group)", "MetricExpr": "1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_other_branches" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_RETIRED.L1_MISS + MEM_LOAD_RETIRED.FB_HIT )", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki_load" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( ( OFFCORE_REQUESTS.ALL_DATA_RD - OFFCORE_REQUESTS.DEMAND_DATA_RD ) + L2_RQSTS.ALL_DEMAND_MISS + L2_RQSTS.SWPF_MISS ) / tma_info_inst_mix_instructions", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * L2_RQSTS.RFO_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", + "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_fb_hpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Offcore;Metric", "MetricName": "tma_info_memory_l3_cache_access_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", "MetricName": "tma_info_memory_tlb_page_walks_utilization", "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Fed;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_code_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_load_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_access_bw", "MetricGroup": "Mem;MemoryBW;Offcore;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_access_bw_2t" }, { - "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory).", + "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory)", "MetricExpr": "1000 * L2_LINES_OUT.SILENT / tma_info_inst_mix_instructions", "MetricGroup": "L2Evicts;Mem;Server;Core_Metric", "MetricName": "tma_info_memory_core_l2_evictions_silent_pki" }, { - "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction.", + "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction", "MetricExpr": "1000 * L2_LINES_OUT.NON_SILENT / tma_info_inst_mix_instructions", "MetricGroup": "L2Evicts;Mem;Server;Core_Metric", "MetricName": "tma_info_memory_core_l2_evictions_nonsilent_pki" }, { - "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses.", + "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses", "MetricExpr": "L2_LINES_OUT.USELESS_HWPF / ( L2_LINES_OUT.SILENT + L2_LINES_OUT.NON_SILENT )", "MetricGroup": "Prefetches;Metric", - "MetricName": "tma_info_memory_prefetches_useless_hwpf" + "MetricName": "tma_info_memory_prefetches_useless_hwpf", + "MetricThreshold": "tma_info_memory_useless_hwpf > 0.15" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=0x1@", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", + "BriefDescription": "Un-cacheable retired load per kilo instruction", "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_uc_load_pki" }, { - "BriefDescription": "\"Bus lock\" per kilo instruction.", + "BriefDescription": "\"Bus lock\" per kilo instruction", "MetricExpr": "1000 * SQ_MISC.BUS_LOCK / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_bus_lock_pki" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz].", + "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz]", "MetricExpr": "tma_info_system_socket_clks / 1e9 / tma_info_system_time", "MetricGroup": "SoC;System_Metric", "MetricName": "tma_info_system_uncore_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0.", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0", "MetricExpr": "CORE_POWER.LVL0_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license0_utilization", "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0. This includes non-AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes." }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1.", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1", "MetricExpr": "CORE_POWER.LVL1_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license1_utilization", @@ -2011,7 +2046,7 @@ "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1. This includes high current AVX 256-bit instructions as well as low current AVX 512-bit instructions." }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX).", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX)", "MetricExpr": "CORE_POWER.LVL2_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license2_utilization", @@ -2019,94 +2054,95 @@ "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX). This includes high current AVX 512-bit instructions." }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_utilization", "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "( 64 * ( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", "MetricName": "tma_info_system_dram_bw_use", - "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_sq_full, tma_mem_bandwidth, tma_fb_full." + "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Average latency of data read request to external memory (in nanoseconds).", + "BriefDescription": "Average latency of data read request to external memory (in nanoseconds)", "MetricExpr": "( 1000000000 ) * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( tma_info_system_socket_clks / tma_info_system_time )", "MetricGroup": "Mem;MemoryLat;SoC;NanoSeconds", "MetricName": "tma_info_system_mem_read_latency", "PublicDescription": "Average latency of data read request to external memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetches. ([RKL+]memory-controller only)." }, { - "BriefDescription": "Average number of parallel data read requests to external memory.", + "BriefDescription": "Average number of parallel data read requests to external memory", "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD\\,thresh\\=0x1@", "MetricGroup": "Mem;MemoryBW;SoC;System_Metric", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches." }, { - "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds].", + "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]", "MetricExpr": "( 1000000000 ) * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR ) / cha_0@event\\=0x0@", "MetricGroup": "MemOffcore;MemoryLat;SoC;Server;NanoSeconds", "MetricName": "tma_info_system_mem_dram_read_latency", "PublicDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]. Accounts for demand loads and L1/L2 data-read prefetches." }, { - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec].", + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]", "MetricExpr": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "IoBW;MemOffcore;SoC;Server;GB/sec", "MetricName": "tma_info_system_io_read_bw", "PublicDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]. Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU." }, { - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec].", + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]", "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_HIT_ITOM + UNC_CHA_TOR_INSERTS.IO_MISS_ITOM + UNC_CHA_TOR_INSERTS.IO_HIT_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_MISS_ITOMCACHENEAR ) * 64 / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "IoBW;MemOffcore;SoC;Server;GB/sec", "MetricName": "tma_info_system_io_write_bw", "PublicDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]. Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU." }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "( power@energy\\-pkg@ * ( 61 ) + 15.6 * power@energy\\-ram@ ) / ( ( duration_time ) * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", "MetricName": "tma_info_system_mux", "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "cha_0@event\\=0x0@", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", "MetricName": "tma_info_system_ipfarbranch", diff --git a/LNL/metrics/lunarlake_metrics_lioncove_core.json b/LNL/metrics/lunarlake_metrics_lioncove_core.json index 80ce6d6c..6f16bdc2 100644 --- a/LNL/metrics/lunarlake_metrics_lioncove_core.json +++ b/LNL/metrics/lunarlake_metrics_lioncove_core.json @@ -2,8 +2,8 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for Intel(R) Core(TM) processors based on Lunar Lake performance hybrid architecture0", - "DatePublished": "11/12/2024", - "Version": "0", + "DatePublished": "11/15/2024", + "Version": "1.0", "Legend": "", "TmaVersion": "5.01", "TmaFlavor": "Full" @@ -99,7 +99,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Mispredictions > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Mispredictions" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Mispredictions > 20", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP", @@ -171,7 +179,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Big_Code > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Big_Code" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Big_Code > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB", @@ -307,7 +323,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Instruction_Fetch_BW" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", @@ -504,7 +528,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore", @@ -741,7 +773,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Latency" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;Mem;MemoryLat;Offcore", @@ -914,7 +954,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Data_TLBs" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore", @@ -1103,7 +1151,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Synchronization > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Synchronization" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Synchronization > 10", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;LockCont;Mem;Offcore", @@ -1187,7 +1243,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Compute_Bound_Est" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20", + "ThresholdIssues": "$issueComp" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Cor", @@ -1375,7 +1439,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Irregular_Overhead > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Irregular_Overhead" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Irregular_Overhead > 10", + "ThresholdIssues": "$issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BvIO;Cor;Ret", @@ -1776,7 +1848,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Other_Bottlenecks" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;Cor;Offcore", @@ -1812,7 +1892,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Branching_Overhead > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Branching_Overhead" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Bottleneck_Branching_Overhead > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Ret", @@ -1876,7 +1964,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Useful_Work > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Useful_Work" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Useful_Work > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;Ret", @@ -1912,7 +2008,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;BvIO;TmaL1;PGO", @@ -1953,7 +2057,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Frontend;TmaL2", @@ -1982,7 +2098,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss", @@ -2023,7 +2155,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -2056,7 +2208,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -2085,7 +2257,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB", @@ -2126,7 +2314,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2159,7 +2367,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2196,7 +2424,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2233,7 +2485,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2266,7 +2542,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -2315,7 +2607,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Mispredicts_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP", @@ -2364,7 +2676,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Clears_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;MachineClears", @@ -2393,7 +2725,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Unknown_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat", @@ -2422,7 +2774,23 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC, $issueMS, $issueMV, $issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MicroSeq", @@ -2451,7 +2819,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LCP(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -2480,7 +2864,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchLat", @@ -2521,7 +2921,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;Frontend;TmaL2", @@ -2566,7 +2974,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -2611,7 +3031,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;FetchBW", @@ -2640,7 +3072,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....LSD(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LSD(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....LSD(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;LSD", @@ -2669,7 +3113,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 5 & b > 20", + "BaseFormula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -2705,7 +3161,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", @@ -2746,7 +3210,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2", @@ -2779,7 +3255,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Cond_NT_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Cond_NT_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Cond_NT_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -2812,7 +3304,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Cond_TK_Bwd_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Cond_TK_Bwd_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Cond_TK_Bwd_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -2845,7 +3353,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Cond_TK_Fwd_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Cond_TK_Fwd_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Cond_TK_Fwd_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -2878,7 +3402,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Ind_Call_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ind_Call_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Ind_Call_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -2919,7 +3459,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Ind_Jump_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ind_Jump_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Ind_Jump_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -2952,7 +3508,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Ret_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ret_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Ret_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -3005,7 +3577,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;BrMispredicts", @@ -3046,7 +3634,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueMC, $issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2", @@ -3095,7 +3695,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Nukes(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;Machine_Clears", @@ -3131,7 +3747,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;TmaL1", @@ -3172,7 +3796,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20", + "BaseFormula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2", @@ -3201,7 +3837,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1, $issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -3238,7 +3890,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -3275,7 +3947,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3304,7 +4000,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3345,7 +4065,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3386,7 +4134,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3427,7 +4203,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3456,7 +4260,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Fwd_Blk(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3485,7 +4309,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L1_Latency_Dependency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -3518,7 +4362,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Capacity(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L1_Latency_Capacity(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L1_Latency_Capacity(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -3551,7 +4415,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Lock_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Offcore", @@ -3592,7 +4476,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......Split_Loads(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Loads(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......Split_Loads(%) > 30", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3621,7 +4513,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......FB_Full(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FB_Full(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......FB_Full(%) > 30", + "ThresholdIssues": "$issueBW, $issueSL, $issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW", @@ -3650,7 +4550,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem", @@ -3704,7 +4620,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L2_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryLat", @@ -3733,7 +4669,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -3795,7 +4747,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Contested_Accesses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -3857,7 +4829,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Data_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;Offcore;Snoop", @@ -3911,7 +4903,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L3_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -3944,7 +4956,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......SQ_Full(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 30 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -3973,7 +5005,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -4002,7 +5050,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Bandwidth(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -4035,7 +5103,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat;Offcore", @@ -4064,7 +5152,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -4105,7 +5209,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO, $issueSL, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore", @@ -4147,7 +5271,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......False_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -4180,7 +5324,27 @@ "Category": "TMA", "CountDomain": "Core_Utilization", "Threshold": { - "Formula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4209,7 +5373,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Streaming_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBW;Offcore", @@ -4246,7 +5430,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -4283,7 +5487,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4312,7 +5540,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4353,7 +5605,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4394,7 +5674,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4435,7 +5743,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4476,7 +5812,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2;Compute", @@ -4505,7 +5853,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB", @@ -4534,7 +5898,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4567,7 +5951,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......INT_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4600,7 +6004,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;PortsUtil", @@ -4629,7 +6049,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Slow_Pause(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4658,7 +6098,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......C01_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......C01_Wait(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......C01_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", @@ -4687,7 +6147,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......C02_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......C02_Wait(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......C02_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", @@ -4716,7 +6196,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Memory_Fence(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Memory_Fence(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Memory_Fence(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4781,7 +6281,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4810,7 +6326,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_0(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4839,7 +6375,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_........Mixing_Vectors(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Mixing_Vectors(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Mixing_Vectors(%) > 5", + "ThresholdIssues": "$issueMV" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4868,7 +6412,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_1(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4897,7 +6461,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_2(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4926,7 +6510,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_3m(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 40 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;PortsUtil", @@ -4955,7 +6559,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........ALU_Op_Utilization(%) > 40" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........ALU_Op_Utilization(%)" + } + ], + "Formula": "a > 40", + "BaseFormula": "metric_TMA_........ALU_Op_Utilization(%) > 40", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4984,7 +6596,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Load_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Load_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5017,7 +6637,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Store_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Store_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5053,7 +6681,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Retiring(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 70 | b > 10", + "BaseFormula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;TmaL1", @@ -5094,7 +6734,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -5151,7 +6799,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 20 & b > 60", + "BaseFormula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -5196,7 +6856,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......X87_Use(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -5241,7 +6917,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Scalar(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5286,7 +6978,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5335,7 +7043,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5380,7 +7108,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5429,7 +7177,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -5474,7 +7234,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......Int_Vector_128b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Int_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 60", + "BaseFormula": "metric_TMA_......Int_Vector_128b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;IntVector;Pipeline", @@ -5519,7 +7295,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......Int_Vector_256b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Int_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 60", + "BaseFormula": "metric_TMA_......Int_Vector_256b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;IntVector;Pipeline", @@ -5568,7 +7360,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Memory_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -5617,7 +7421,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Fused_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -5670,7 +7486,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Non_Fused_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -5767,7 +7595,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 30 & b > 60", + "BaseFormula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -5816,7 +7656,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Nop_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Pipeline", @@ -5865,7 +7721,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Shuffles_256b(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Shuffles_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Shuffles_256b(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC;Pipeline", @@ -5906,7 +7778,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -5955,7 +7835,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Few_Uops_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5984,7 +7876,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueMC, $issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -6013,7 +7917,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Assists(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO", @@ -6042,7 +7962,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........Page_Faults(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Page_Faults(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Page_Faults(%) > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6071,7 +7999,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........FP_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........FP_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -6100,7 +8036,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........AVX_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........AVX_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........AVX_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -6133,7 +8077,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......CISC(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6160,6 +8120,11 @@ "BaseFormula": " inst_retired.any / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Summary", "LocateWith": "" @@ -6202,7 +8167,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UopPI > 1.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UopPI" + } + ], + "Formula": "a > 1.05", + "BaseFormula": "metric_TMA_Info_Thread_UopPI > 1.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret;Retire", @@ -6246,7 +8219,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UpTB < 8 * 1.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UpTB" + } + ], + "Formula": "a < 8 * 1.5", + "BaseFormula": "metric_TMA_Info_Thread_UpTB < 8 * 1.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW", @@ -6273,6 +8254,11 @@ "BaseFormula": " 1 / tma_info_thread_ipc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Mem", "LocateWith": "" @@ -6294,6 +8280,11 @@ "BaseFormula": " cpu_clk_unhalted.thread", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", "LocateWith": "" @@ -6315,6 +8306,11 @@ "BaseFormula": " topdown.slots:perf_metrics", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", "LocateWith": "" @@ -6340,6 +8336,11 @@ "BaseFormula": " uops_executed.thread / uops_issued.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline", "LocateWith": "" @@ -6377,6 +8378,11 @@ "BaseFormula": " ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.256b_packed_single ) / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Flops", "LocateWith": "" @@ -6414,6 +8420,11 @@ "BaseFormula": " ( fp_arith_dispatched.v0 + fp_arith_dispatched.v1 + fp_arith_dispatched.v2 + fp_arith_dispatched.v3 ) / ( 4 * tma_info_thread_clks )", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -6439,6 +8450,11 @@ "BaseFormula": " uops_executed.thread / uops_executed.thread:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "LocateWith": "" @@ -6464,6 +8480,11 @@ "BaseFormula": " uops_executed.thread / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -6490,7 +8511,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpLoad < 3" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpLoad" + } + ], + "Formula": "a < 3", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpLoad < 3", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -6518,7 +8547,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpStore < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpStore" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpStore < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -6546,7 +8583,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpBranch < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpBranch" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpBranch < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;InsType", @@ -6574,7 +8619,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpCall < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpCall" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpCall < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", @@ -6602,7 +8655,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpTB < 8 * 2 + 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpTB" + } + ], + "Formula": "a < 8 * 2 + 1", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpTB < 8 * 2 + 1", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO", @@ -6629,6 +8690,11 @@ "BaseFormula": " br_inst_retired.all_branches / br_inst_retired.near_taken", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", "LocateWith": "" @@ -6667,7 +8733,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpFLOP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -6699,7 +8773,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -6727,7 +8809,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -6755,7 +8845,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -6787,7 +8885,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX128" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -6819,7 +8925,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX256" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -6846,6 +8960,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / cpu_clk_unhalted.pause_inst", "Category": "TMA", "CountDomain": "Inst_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", "LocateWith": "" @@ -6872,7 +8991,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpSWPF" + } + ], + "Formula": "a < 100", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -6895,6 +9022,11 @@ "BaseFormula": " inst_retired.any", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;TmaL1", "LocateWith": "INST_RETIRED.PREC_DIST" @@ -6936,6 +9068,11 @@ "BaseFormula": " ( tma_retiring * tma_info_thread_slots ) / uops_retired.slots:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret", "LocateWith": "" @@ -6962,7 +9099,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Pipeline_Strings_Cycles > 0.1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Pipeline_Strings_Cycles" + } + ], + "Formula": "a > 0.1", + "BaseFormula": "metric_TMA_Info_Pipeline_Strings_Cycles > 0.1", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq;Pipeline;Ret", @@ -6990,7 +9135,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Pipeline_IpAssist < 100000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Pipeline_IpAssist" + } + ], + "Formula": "a < 100000", + "BaseFormula": "metric_TMA_Info_Pipeline_IpAssist < 100000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", @@ -7017,6 +9170,11 @@ "BaseFormula": " lsd.uops / lsd.cycles_active", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7042,6 +9200,11 @@ "BaseFormula": " idq.dsb_uops / idq.dsb_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7067,6 +9230,11 @@ "BaseFormula": " idq.mite_uops / idq.mite_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7092,6 +9260,11 @@ "BaseFormula": " uops_issued.any / uops_issued.any:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7117,6 +9290,11 @@ "BaseFormula": " lsd.uops / ( uops_issued.any )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;LSD", "LocateWith": "" @@ -7143,7 +9321,19 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 8 > 0.35" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_DSB_Coverage" + }, + { + "Alias": "b", + "Value": "IPC" + } + ], + "Formula": "a < 0.7 & b / 8 > 0.35", + "BaseFormula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 8 > 0.35", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -7170,6 +9360,11 @@ "BaseFormula": " int_misc.unknown_branch_cycles / int_misc.unknown_branch_cycles:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed", "LocateWith": "" @@ -7195,6 +9390,11 @@ "BaseFormula": " dsb2mite_switches.penalty_cycles / dsb2mite_switches.penalty_cycles:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss", "LocateWith": "" @@ -7220,6 +9420,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -7250,7 +9455,15 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_DSB_Switches_Ret > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_DSB_Switches_Ret" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_Frontend_DSB_Switches_Ret > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed;FetchLat", @@ -7282,7 +9495,15 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_MS_Latency_Ret > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_MS_Latency_Ret" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_Frontend_MS_Latency_Ret > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;MicroSeq", @@ -7313,6 +9534,11 @@ "BaseFormula": " ( frontend_retired.unknown_branch * frontend_retired.unknown_branch:retire_latency ) / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Clocks_Retired", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat", "LocateWith": "" @@ -7338,6 +9564,11 @@ "BaseFormula": " icache_data.stalls / icache_data.stall_periods", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", "LocateWith": "" @@ -7364,7 +9595,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret" + } + ], + "Formula": "a < 50", + "BaseFormula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -7391,6 +9630,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / baclears.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed", "LocateWith": "" @@ -7416,6 +9660,11 @@ "BaseFormula": " 1000 * frontend_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -7441,6 +9690,11 @@ "BaseFormula": " 1000 * l2_rqsts.code_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -7543,7 +9797,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Misses > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Misses" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Misses > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -7619,7 +9881,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Bandwidth" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -7691,7 +9961,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_IC_Misses > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_IC_Misses" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Info_Botlnk_IC_Misses > 5", + "ThresholdIssues": "$issueFL" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", @@ -7719,7 +9997,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMispredict" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts", @@ -7747,7 +10033,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -7774,6 +10068,11 @@ "BaseFormula": " inst_retired.any / br_misp_retired.cond_taken_bwd", "Category": "TMA", "CountDomain": "", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", "LocateWith": "" @@ -7799,6 +10098,11 @@ "BaseFormula": " inst_retired.any / br_misp_retired.cond_taken_fwd", "Category": "TMA", "CountDomain": "", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", "LocateWith": "" @@ -7825,7 +10129,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Ret" + } + ], + "Formula": "a < 500", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -7853,7 +10165,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect" + } + ], + "Formula": "a < 1000", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -7948,6 +10268,11 @@ "BaseFormula": " tma_mispredictions * tma_info_thread_slots / ( 8 ) / br_misp_retired.all_branches / 100", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "$issueBM" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", "LocateWith": "" @@ -7977,6 +10302,11 @@ "BaseFormula": " int_misc.clears_count / ( br_misp_retired.all_branches + machine_clears.count )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", "LocateWith": "" @@ -8002,6 +10332,11 @@ "BaseFormula": " br_inst_retired.cond_ntaken / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", "LocateWith": "" @@ -8028,7 +10363,15 @@ "Category": "TMA", "CountDomain": "Fraction", "Threshold": { - "Formula": "metric_TMA_Info_Branches_Cond_TK_Bwd > 0.3" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Branches_Cond_TK_Bwd" + } + ], + "Formula": "a > 0.3", + "BaseFormula": "metric_TMA_Info_Branches_Cond_TK_Bwd > 0.3", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", @@ -8056,7 +10399,15 @@ "Category": "TMA", "CountDomain": "Fraction", "Threshold": { - "Formula": "metric_TMA_Info_Branches_Cond_TK_Fwd > 0.2" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Branches_Cond_TK_Fwd" + } + ], + "Formula": "a > 0.2", + "BaseFormula": "metric_TMA_Info_Branches_Cond_TK_Fwd > 0.2", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", @@ -8087,6 +10438,11 @@ "BaseFormula": " ( br_inst_retired.near_call + br_inst_retired.near_return ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8124,6 +10480,11 @@ "BaseFormula": " ( br_inst_retired.near_taken - br_inst_retired.cond_taken_bwd - br_inst_retired.cond_taken_fwd - 2 * br_inst_retired.near_call ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8169,6 +10530,11 @@ "BaseFormula": " 1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk_bwd + tma_info_branches_cond_tk_fwd + tma_info_branches_callret + tma_info_branches_jump )", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8194,6 +10560,11 @@ "BaseFormula": " l1d_pending.load / l1d_miss.load", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryLat", "LocateWith": "" @@ -8219,6 +10590,11 @@ "BaseFormula": " l1d_pending.load / l1d_pending.load_cycles", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryBW", "LocateWith": "" @@ -8244,6 +10620,11 @@ "BaseFormula": " 1000 * mem_load_retired.l1_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8269,6 +10650,11 @@ "BaseFormula": " 1000 * l2_rqsts.all_demand_data_rd / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8294,6 +10680,11 @@ "BaseFormula": " 1000 * mem_load_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;Backend;CacheHits", "LocateWith": "" @@ -8319,6 +10710,11 @@ "BaseFormula": " 1000 * l2_rqsts.miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem;Offcore", "LocateWith": "" @@ -8344,6 +10740,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8369,6 +10770,11 @@ "BaseFormula": " 1000 * l2_rqsts.rfo_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheMisses;Offcore", "LocateWith": "" @@ -8398,6 +10804,11 @@ "BaseFormula": " 1000 * ( l2_rqsts.references - l2_rqsts.miss ) / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8423,6 +10834,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8448,6 +10864,11 @@ "BaseFormula": " 1000 * mem_load_retired.l3_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -8473,6 +10894,11 @@ "BaseFormula": " 1000 * mem_load_retired.fb_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8499,6 +10925,11 @@ "BaseFormula": " 64 * l1d.replacement / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8525,6 +10956,11 @@ "BaseFormula": " 64 * l1d.l0_replacement / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8551,6 +10987,11 @@ "BaseFormula": " 64 * l2_lines_in.all / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8577,6 +11018,11 @@ "BaseFormula": " 64 * longest_lat_cache.miss / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8603,6 +11049,11 @@ "BaseFormula": " 64 * offcore_requests.all_requests / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW;Offcore", "LocateWith": "" @@ -8637,7 +11088,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Page_Walks_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -8664,6 +11123,11 @@ "BaseFormula": " 1000 * itlb_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;MemoryTLB", "LocateWith": "" @@ -8689,6 +11153,11 @@ "BaseFormula": " 1000 * dtlb_load_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -8714,6 +11183,11 @@ "BaseFormula": " 1000 * dtlb_store_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -8744,7 +11218,15 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Load_STLB_Miss_Ret > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Load_STLB_Miss_Ret" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_Memory_Load_STLB_Miss_Ret > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -8776,7 +11258,15 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Store_STLB_Miss_Ret > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Store_STLB_Miss_Ret" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_Memory_Store_STLB_Miss_Ret > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -8808,7 +11298,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Useless_HWPF" + } + ], + "Formula": "a > 0.15", + "BaseFormula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -8835,6 +11333,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests.demand_data_rd", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Memory_Lat;Offcore", "LocateWith": "" @@ -8860,6 +11363,11 @@ "BaseFormula": " offcore_requests_outstanding.l3_miss_demand_data_rd / offcore_requests.l3_miss_demand_data_rd", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_Lat;Offcore", "LocateWith": "" @@ -8885,6 +11393,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests_outstanding.demand_data_rd:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -8910,6 +11423,11 @@ "BaseFormula": " offcore_requests_outstanding.data_rd / offcore_requests_outstanding.cycles_with_data_rd", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -8935,6 +11453,11 @@ "BaseFormula": " 1000 * mem_load_misc_retired.uc / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -8960,6 +11483,11 @@ "BaseFormula": " 1000 * sq_misc.bus_lock / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -8990,6 +11518,11 @@ "BaseFormula": " tma_info_system_cpus_utilized / num_cpus", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC;Summary", "LocateWith": "" @@ -9016,6 +11549,11 @@ "BaseFormula": " cpu_clk_unhalted.ref_tsc / tsc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", "LocateWith": "" @@ -9050,6 +11588,11 @@ "BaseFormula": " tma_info_system_turbo_utilization * tsc / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;Power", "LocateWith": "" @@ -9088,6 +11631,11 @@ "BaseFormula": " ( ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.256b_packed_single ) / ( 1000000000 ) ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -9113,6 +11661,11 @@ "BaseFormula": " tma_info_thread_clks / cpu_clk_unhalted.ref_tsc", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -9139,7 +11692,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Kernel_Utilization > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Kernel_Utilization" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_System_Kernel_Utilization > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "OS", @@ -9166,6 +11727,11 @@ "BaseFormula": " cpu_clk_unhalted.thread_p:sup / inst_retired.any_p:sup", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "OS", "LocateWith": "" @@ -9192,38 +11758,20 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_C0_Wait > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_C0_Wait" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_System_C0_Wait > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", "LocateWith": "" }, - { - "MetricName": "Info_System_DRAM_BW_Use", - "LegacyName": "metric_TMA_Info_System_DRAM_BW_Use", - "Level": 1, - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UNC_ARB_TRK_REQUESTS.ALL", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "64 * a / ( 1000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) / 1000", - "BaseFormula": " 64 * unc_arb_trk_requests.all / ( 1000000 ) / tma_info_system_time / 1000", - "Category": "TMA", - "CountDomain": "GB/sec", - "ResolutionLevels": "ARB", - "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC", - "LocateWith": "" - }, { "MetricName": "Info_System_Power", "LegacyName": "metric_TMA_Info_System_Power", @@ -9246,6 +11794,11 @@ "BaseFormula": " unc_pkg_energy_status * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "PKG", "MetricGroup": "Power;SoC", "LocateWith": "" @@ -9268,7 +11821,15 @@ "Category": "TMA", "CountDomain": "Seconds", "Threshold": { - "Formula": "metric_TMA_Info_System_Time < 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Time" + } + ], + "Formula": "a < 1", + "BaseFormula": "metric_TMA_Info_System_Time < 1", + "ThresholdIssues": "" }, "ResolutionLevels": "SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -9296,7 +11857,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_MUX" + } + ], + "Formula": "a > 1.1 | a < 0.9", + "BaseFormula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -9324,7 +11893,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_IpFarBranch < 1000000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_IpFarBranch" + } + ], + "Formula": "a < 1000000", + "BaseFormula": "metric_TMA_Info_System_IpFarBranch < 1000000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;OS", diff --git a/LNL/metrics/perf/lunarlake_metrics_lioncove_core_perf.json b/LNL/metrics/perf/lunarlake_metrics_lioncove_core_perf.json index 9cc34e11..f1944578 100644 --- a/LNL/metrics/perf/lunarlake_metrics_lioncove_core_perf.json +++ b/LNL/metrics/perf/lunarlake_metrics_lioncove_core_perf.json @@ -1,1635 +1,1805 @@ [ { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", + "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", + "MetricThreshold": "tma_bottleneck_mispredictions > 20", + "PublicDescription": "Total pipeline cost of Branch Misprediction related bottlenecks. Related metrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", + "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_big_code", + "MetricThreshold": "tma_bottleneck_big_code > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", + "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) ) ) - tma_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_instruction_fetch_bw", + "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_l1_latency_capacity + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) )", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", + "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_l1_latency_capacity + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_capacity / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_l1_latency_capacity + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_l1_latency_capacity + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_l1_latency_capacity + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", + "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks. Related metrics: tma_l3_hit_latency, tma_mem_latency.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", + "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs)", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_load / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_l1_latency_capacity + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", + "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", + "PublicDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs). Related metrics: tma_dtlb_load, tma_dtlb_store.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", + "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors)", "MetricExpr": "100 * ( tma_memory_bound * ( ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", - "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", + "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", + "PublicDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", + "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation", "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", + "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", "PublicDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", + "BriefDescription": "Total pipeline cost of irregular execution (e.g", "MetricExpr": "100 * ( ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + RS.EMPTY_RESOURCE / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_microcode_sequencer + max( 0 , tma_heavy_operations - tma_microcode_sequencer ) ) ) * ( ( ( 99 *3 + 63 + 30 ) / 5 ) * ASSISTS.ANY / tma_info_thread_slots / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", - "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments).", + "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", + "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments). Related metrics: tma_microcode_sequencer, tma_ms_switches.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end.", + "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end", "MetricExpr": "100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", "MetricGroup": "BvOB;Cor;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_other_bottlenecks", + "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", "PublicDescription": "Total pipeline cost of remaining bottlenecks in the back-end. Examples include data-dependencies (Core Bound when Low ILP) and other unlisted memory-related stalls.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", + "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA", "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_branching_overhead", + "MetricThreshold": "tma_bottleneck_branching_overhead > 5", "PublicDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound).", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", + "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead", "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_microcode_sequencer + max( 0 , tma_heavy_operations - tma_microcode_sequencer ) ) ) * ( ( ( 99 *3 + 63 + 30 ) / 5 ) * ASSISTS.ANY / tma_info_thread_slots / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_useful_work", + "MetricThreshold": "tma_bottleneck_useful_work > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "( topdown\\-fe\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_frontend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_frontend_bound > 0.15", "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( topdown\\-fetch\\-lat / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", "MetricName": "tma_fetch_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_16, FRONTEND_RETIRED.LATENCY_GE_8.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache", "MetricExpr": "max( 0 , ( FRONTEND_RETIRED.L1I_MISS * cpu_core@FRONTEND_RETIRED.L1I_MISS@R ) / tma_info_thread_clks - tma_code_l2_miss )", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache", "MetricExpr": "( FRONTEND_RETIRED.L2_MISS * cpu_core@FRONTEND_RETIRED.L2_MISS@R ) / tma_info_thread_clks", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , ( FRONTEND_RETIRED.ITLB_MISS * cpu_core@FRONTEND_RETIRED.ITLB_MISS@R ) / tma_info_thread_clks - tma_code_stlb_miss )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", "MetricExpr": "( FRONTEND_RETIRED.STLB_MISS * cpu_core@FRONTEND_RETIRED.STLB_MISS@R ) / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + tma_unknown_branches", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "( tma_branch_mispredicts / tma_bad_speculation ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "( 1 - ( tma_branch_mispredicts / tma_bad_speculation ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 3 ) * IDQ.MS_SWITCHES / tma_info_thread_clks", - "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated", + "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES." + "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "DECODE.LCP / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." + "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", - "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS." + "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "max( 0 , tma_frontend_bound - tma_fetch_latency )", - "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", + "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend.", + "MetricThreshold": "tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Related metrics: tma_dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( cpu@IDQ.MITE_UOPS\\,cmask\\=0x8\\,inv\\=0x1@ / tma_info_thread_clks + IDQ.MITE_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS ) * ( IDQ_BUBBLES.CYCLES_0_UOPS_DELIV.CORE - IDQ_BUBBLES.FETCH_LATENCY ) ) / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", "ScaleUnit": "100%", + "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( cpu@IDQ.DSB_UOPS\\,cmask\\=0x8\\,inv\\=0x1@ + IDQ.DSB_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS ) * ( IDQ_BUBBLES.CYCLES_0_UOPS_DELIV.CORE - IDQ_BUBBLES.FETCH_LATENCY ) ) / tma_info_thread_clks", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", "ScaleUnit": "100%", + "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit", "MetricExpr": "cpu@LSD.UOPS\\,cmask\\=0x8\\,inv\\=0x1@ / tma_info_thread_clks", "MetricGroup": "FetchBW;LSD;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_lsd", "ScaleUnit": "100%", + "MetricThreshold": "tma_lsd > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit. LSD typically does well sustaining Uop supply. However; in some rare cases; optimal uop-delivery could not be reached for small loops whose size (in terms of number of uops) does not suit well the LSD structure." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details", "MetricExpr": "IDQ.MS_CYCLES_ANY / tma_info_thread_clks", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_ms", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "( topdown\\-bad\\-spec / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_bad_speculation", "ScaleUnit": "100%", + "MetricThreshold": "tma_bad_speculation > 0.15", "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( topdown\\-br\\-mispredict / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS.", + "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by non-taken conditional branches.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by non-taken conditional branches", "MetricExpr": "( BR_MISP_RETIRED.COND_NTAKEN_COST * cpu_core@BR_MISP_RETIRED.COND_NTAKEN_COST@R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_cond_nt_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_cond_nt_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to misprediction by backward-taken conditional branches.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to misprediction by backward-taken conditional branches", "MetricExpr": "( BR_MISP_RETIRED.COND_TAKEN_BWD_COST * cpu_core@BR_MISP_RETIRED.COND_TAKEN_BWD_COST@R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_cond_tk_bwd_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_cond_tk_bwd_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to misprediction by forward-taken conditional branches.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to misprediction by forward-taken conditional branches", "MetricExpr": "( BR_MISP_RETIRED.COND_TAKEN_FWD_COST * cpu_core@BR_MISP_RETIRED.COND_TAKEN_FWD_COST@R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_cond_tk_fwd_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_cond_tk_fwd_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect CALL instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect CALL instructions", "MetricExpr": "( BR_MISP_RETIRED.INDIRECT_CALL_COST * cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_ind_call_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ind_call_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect JMP instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect JMP instructions", "MetricExpr": "max( ( ( BR_MISP_RETIRED.INDIRECT_COST * cpu_core@BR_MISP_RETIRED.INDIRECT_COST@R ) - ( BR_MISP_RETIRED.INDIRECT_CALL_COST * cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@R ) ) / tma_info_thread_clks , 0 )", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_ind_jump_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ind_jump_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by (indirect) RET instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by (indirect) RET instructions", "MetricExpr": "( BR_MISP_RETIRED.RET_COST * cpu_core@BR_MISP_RETIRED.RET_COST@R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_ret_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ret_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", + "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types)", "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Slots", "MetricName": "tma_other_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "max( 0 , tma_bad_speculation - tma_branch_mispredicts )", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots", + "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT.", + "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering", "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_machine_clears_group;Slots", "MetricName": "tma_other_nukes", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "( topdown\\-be\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_backend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_backend_bound > 0.2", "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( topdown\\-mem\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", "MetricName": "tma_memory_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two).", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "MEMORY_STALLS.L1 / tma_info_thread_clks", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", + "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT." + "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "( min( ( MEM_INST_RETIRED.STLB_HIT_LOADS * cpu_core@MEM_INST_RETIRED.STLB_HIT_LOADS@R ) , MEM_INST_RETIRED.STLB_HIT_LOADS * ( 7 ) ) if ( cpu_core@MEM_INST_RETIRED.STLB_HIT_LOADS@R >= 0 ) else ( MEM_INST_RETIRED.STLB_HIT_LOADS * ( 7 ) ) ) / tma_info_thread_clks + tma_load_stlb_miss", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS." + "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_dtlb_load - tma_load_stlb_miss )", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", "MetricName": "tma_load_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", "MetricName": "tma_load_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", + "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache", "MetricExpr": "4 * DEPENDENT_LOADS.ANY / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_l1_latency_dependency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: DEPENDENT_LOADS.ANY." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit Level 1 after missing Level 0 within the L1D cache.", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit Level 1 after missing Level 0 within the L1D cache", "MetricExpr": "( min( ( MEM_LOAD_RETIRED.L1_HIT_L1 * cpu_core@MEM_LOAD_RETIRED.L1_HIT_L1@R ) , MEM_LOAD_RETIRED.L1_HIT_L1 * 9 ) if ( cpu_core@MEM_LOAD_RETIRED.L1_HIT_L1@R >= 0 ) else ( MEM_LOAD_RETIRED.L1_HIT_L1 * 9 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Retired", "MetricName": "tma_l1_latency_capacity", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_latency_capacity > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( MEM_INST_RETIRED.LOCK_LOADS * cpu_core@MEM_INST_RETIRED.LOCK_LOADS@R ) / tma_info_thread_clks", - "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks", + "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS." + "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "( min( ( MEM_INST_RETIRED.SPLIT_LOADS * cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@R ) , MEM_INST_RETIRED.SPLIT_LOADS * tma_info_memory_load_miss_real_latency ) if ( cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@R >= 0 ) else ( MEM_INST_RETIRED.SPLIT_LOADS * tma_info_memory_load_miss_real_latency ) ) / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "L1D_MISS.FB_FULL / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", + "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory)." + "MetricThreshold": "tma_fb_full > 0.3", + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "MEMORY_STALLS.L2 / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", + "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited)", "MetricExpr": "( min( ( MEM_LOAD_RETIRED.L2_HIT * cpu_core@MEM_LOAD_RETIRED.L2_HIT@R ) , MEM_LOAD_RETIRED.L2_HIT * ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_RETIRED.L2_HIT@R >= 0 ) else ( MEM_LOAD_RETIRED.L2_HIT * ( 3 * tma_info_system_core_frequency ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", "MetricName": "tma_l2_hit_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "MEMORY_STALLS.L3 / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) + ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM@R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * ( 28 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM@R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * ( 28 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS." + "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) + ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 28 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 28 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD." + "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( min( ( MEM_LOAD_RETIRED.L3_HIT * cpu_core@MEM_LOAD_RETIRED.L3_HIT@R ) , MEM_LOAD_RETIRED.L3_HIT * ( 12 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_RETIRED.L3_HIT@R >= 0 ) else ( MEM_LOAD_RETIRED.L3_HIT * ( 12 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT." + "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_bottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "( XQ.FULL + L1D_MISS.L2_STALLS ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( MEMORY_STALLS.MEM / tma_info_thread_clks )", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that)." + "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that)." + "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( MEM_STORE_RETIRED.L2_HIT * ( 10 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)." + "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( 28 * tma_info_system_core_frequency ) * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM." + "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "( min( ( MEM_INST_RETIRED.SPLIT_STORES * cpu_core@MEM_INST_RETIRED.SPLIT_STORES@R ) , MEM_INST_RETIRED.SPLIT_STORES * 1 ) if ( cpu_core@MEM_INST_RETIRED.SPLIT_STORES@R >= 0 ) else ( MEM_INST_RETIRED.SPLIT_STORES * 1 ) ) / tma_info_thread_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization", + "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores.", + "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores", "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread_clks", - "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSmSt", "MetricName": "tma_streaming_stores", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE." + "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma_fb_full." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( min( ( MEM_INST_RETIRED.STLB_HIT_STORES * cpu_core@MEM_INST_RETIRED.STLB_HIT_STORES@R ) , MEM_INST_RETIRED.STLB_HIT_STORES * ( 7 ) ) if ( cpu_core@MEM_INST_RETIRED.STLB_HIT_STORES@R >= 0 ) else ( MEM_INST_RETIRED.STLB_HIT_STORES * ( 7 ) ) ) / tma_info_thread_clks + tma_store_stlb_miss", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES." + "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_load." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_dtlb_store - tma_store_stlb_miss )", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", "MetricName": "tma_store_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", "MetricName": "tma_store_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "max( 0 , tma_backend_bound - tma_memory_bound )", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", "MetricName": "tma_core_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations).", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.DIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", "ScaleUnit": "100%", + "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIV_ACTIVE." }, { - "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active", "MetricExpr": "ARITH.FPDIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_fp_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_int_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations", "MetricExpr": "( BE_STALLS.SCOREBOARD + CPU_CLK_UNHALTED.C02 ) / tma_info_thread_clks", - "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", + "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks;tma_issueSO", "MetricName": "tma_serializing_operation", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: BE_STALLS.SCOREBOARD." + "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: BE_STALLS.SCOREBOARD. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions", "MetricExpr": "CPU_CLK_UNHALTED.PAUSE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_slow_pause", "ScaleUnit": "100%", + "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.PAUSE_INST." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings).", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings)", "MetricExpr": "CPU_CLK_UNHALTED.C01 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_c01_wait", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_c01_wait > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings).", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings)", "MetricExpr": "CPU_CLK_UNHALTED.C02 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_c02_wait", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_c02_wait > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions", "MetricExpr": "13 * MISC2_RETIRED.LFENCE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_memory_fence", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_fence > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( EXE_ACTIVITY.EXE_BOUND_0_PORTS + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIV_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - EXE_ACTIVITY.BOUND_ON_LOADS ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.EXE_BOUND_0_PORTS / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", + "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles)", "MetricExpr": "160 * ASSISTS.SSE_AVX_MIX / tma_info_thread_clks", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks;tma_issueMV", "MetricName": "tma_mixing_vectors", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic." + "MetricThreshold": "tma_mixing_vectors > 0.05", + "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Sample with: UOPS_EXECUTED.CYCLES_GE_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "UOPS_DISPATCHED.ALU / ( 6 * tma_info_thread_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "UOPS_DISPATCHED.LOAD / ( 3 * tma_info_thread_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_load_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations. Sample with: UOPS_DISPATCHED.LOAD." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "( UOPS_DISPATCHED.STD + UOPS_DISPATCHED.STA ) / ( 7 * tma_info_thread_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations. Sample with: UOPS_DISPATCHED.STD, UOPS_DISPATCHED.STA." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( topdown\\-retiring / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_retiring", "ScaleUnit": "100%", + "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.1", "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "max( 0 , tma_retiring - tma_heavy_operations )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", "MetricName": "tma_light_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", "ScaleUnit": "100%", + "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "cpu@FP_ARITH_INST_RETIRED.VECTOR\\,umask\\=0x30@ / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_int_vector_128b, tma_int_vector_256b, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired)", "MetricExpr": "tma_int_vector_128b + tma_int_vector_256b", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_int_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_int_operations > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired). Vector/Matrix Int operations and shuffles are counted. Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain." }, { - "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", + "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired", "MetricExpr": "INT_VEC_RETIRED.128BIT / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops", + "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops;tma_issue2P", "MetricName": "tma_int_vector_128b", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_vector_128b > 0.1 & tma_int_operations > 0.1 & tma_light_operations > 0.6", + "PublicDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_256b, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", + "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired", "MetricExpr": "INT_VEC_RETIRED.256BIT / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops", + "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops;tma_issue2P", "MetricName": "tma_int_vector_256b", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_vector_256b > 0.1 & tma_int_operations > 0.1 & tma_light_operations > 0.6", + "PublicDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses", "MetricExpr": "tma_light_operations * MEM_UOP_RETIRED.ANY / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_memory_operations", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_operations > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.MACRO_FUSED / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_fused_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Other_Light_Ops in MTL!)}." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused", "MetricExpr": "tma_light_operations * ( BR_INST_RETIRED.ALL_BRANCHES - INST_RETIRED.BR_FUSED ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_non_fused_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused. Non-conditional branches like direct JMP or CALL would count here. Can be used to examine fusible conditional jumps that were not fused." }, { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", + "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes", "MetricExpr": "max( 0 , tma_light_operations - ( tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD + ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) ) / ( tma_retiring * tma_info_thread_slots ) + ( INT_VEC_RETIRED.ADD_128 + INT_VEC_RETIRED.VNNI_128 + INT_VEC_RETIRED.ADD_256 + INT_VEC_RETIRED.MUL_256 + INT_VEC_RETIRED.VNNI_256 ) / ( tma_retiring * tma_info_thread_slots ) + tma_memory_operations + tma_fused_instructions + tma_non_fused_branches ) )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_other_light_ops", "ScaleUnit": "100%", + "MetricThreshold": "tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_light_ops_group;Slots", "MetricName": "tma_nop_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer)", "MetricExpr": "tma_light_operations * INT_VEC_RETIRED.SHUFFLES / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_other_light_ops_group;Slots", "MetricName": "tma_shuffles_256b", "ScaleUnit": "100%", + "MetricThreshold": "tma_shuffles_256b > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer). Shuffles may incur slow cross \"vector lane\" data transfers." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "( topdown\\-heavy\\-ops / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", "MetricName": "tma_heavy_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+]). Sample with: UOPS_RETIRED.HEAVY.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops", "MetricExpr": "max( 0 , tma_heavy_operations - tma_microcode_sequencer )", - "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueD0", "MetricName": "tma_few_uops_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_few_uops_instructions > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "UOPS_RETIRED.MS / tma_info_thread_slots", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: UOPS_RETIRED.MS." + "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: UOPS_RETIRED.MS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( ( 99 *3 + 63 + 30 ) / 5 ) * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults", "MetricExpr": "99 * ASSISTS.PAGE_FAULT / tma_info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_page_faults", "ScaleUnit": "100%", + "MetricThreshold": "tma_page_faults > 0.05", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults. A Page Fault may apply on first application access to a memory page. Note operating system handling of page faults accounts for the majority of its cost." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists", "MetricExpr": "30 * ASSISTS.FP / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_fp_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_assists > 0.1", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists", "MetricExpr": "63 * ASSISTS.SSE_AVX_MIX / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_avx_assists", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_avx_assists > 0.1" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", "ScaleUnit": "100%", + "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources." }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", - "MetricName": "tma_info_thread_uoppi" + "MetricName": "tma_info_thread_uoppi", + "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", - "MetricName": "tma_info_thread_uptb" + "MetricName": "tma_info_thread_uptb", + "MetricThreshold": "tma_info_thread_uptb < 8 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / tma_info_thread_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( FP_ARITH_DISPATCHED.V0 + FP_ARITH_DISPATCHED.V1 + FP_ARITH_DISPATCHED.V2 + FP_ARITH_DISPATCHED.V3 ) / ( 4 * tma_info_thread_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "uops Executed per Cycle.", + "BriefDescription": "uops Executed per Cycle", "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", "MetricGroup": "Power;Metric", "MetricName": "tma_info_core_epc" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipload" + "MetricName": "tma_info_inst_mix_ipload", + "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipstore" + "MetricName": "tma_info_inst_mix_ipstore", + "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipbranch" + "MetricName": "tma_info_inst_mix_ipbranch", + "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipcall" + "MetricName": "tma_info_inst_mix_ipcall", + "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", - "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_iptb" + "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", + "MetricName": "tma_info_inst_mix_iptb", + "MetricThreshold": "tma_info_inst_mix_iptb < 8 * 2 + 1", + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipflop" + "MetricName": "tma_info_inst_mix_ipflop", + "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", + "MetricThreshold": "tma_info_inst_mix_iparith < 10", "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", + "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", + "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / CPU_CLK_UNHALTED.PAUSE_INST", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ippause" }, { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_SWPF", "MetricGroup": "Prefetches;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipswpf" + "MetricName": "tma_info_inst_mix_ipswpf", + "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions.", + "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions", "MetricExpr": "INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "MicroSeq;Pipeline;Ret;Metric", - "MetricName": "tma_info_pipeline_strings_cycles" + "MetricName": "tma_info_pipeline_strings_cycles", + "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1" }, { - "BriefDescription": "Instructions per a microcode Assist invocation.", + "BriefDescription": "Instructions per a microcode Assist invocation", "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", "MetricName": "tma_info_pipeline_ipassist", + "MetricThreshold": "tma_info_pipeline_ipassist < 100000", "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." }, { - "BriefDescription": "Average number of uops fetched from LSD per cycle.", + "BriefDescription": "Average number of uops fetched from LSD per cycle", "MetricExpr": "LSD.UOPS / LSD.CYCLES_ACTIVE", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_lsd" }, { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", + "BriefDescription": "Average number of uops fetched from DSB per cycle", "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_dsb" }, { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", + "BriefDescription": "Average number of uops fetched from MITE per cycle", "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_mite" }, { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", + "BriefDescription": "Average number of Uops issued by front-end when it issued something", "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_frontend_fetch_upc" }, { - "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache).", + "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache)", "MetricExpr": "LSD.UOPS / ( UOPS_ISSUED.ANY )", "MetricGroup": "Fed;LSD;Metric", "MetricName": "tma_info_frontend_lsd_coverage" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( UOPS_ISSUED.ANY )", - "MetricGroup": "DSB;Fed;FetchBW;Metric", - "MetricName": "tma_info_frontend_dsb_coverage" + "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", + "MetricName": "tma_info_frontend_dsb_coverage", + "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 8 > 0.35", + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection.", + "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection", "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / cpu@INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_unknown_branch_cost", "PublicDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection. See Unknown_Branches node." }, { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", + "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "DSBmiss;Metric", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired DSB misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired DSB misses", "MetricExpr": "( FRONTEND_RETIRED.ANY_DSB_MISS * cpu_core@FRONTEND_RETIRED.ANY_DSB_MISS@R ) / tma_info_thread_clks", "MetricGroup": "DSBmiss;Fed;FetchLat;Clocks_Retired", - "MetricName": "tma_info_frontend_dsb_switches_ret" + "MetricName": "tma_info_frontend_dsb_switches_ret", + "MetricThreshold": "tma_info_frontend_dsb_switches_ret > 0.05" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired operations that invoke the Microcode Sequencer.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired operations that invoke the Microcode Sequencer", "MetricExpr": "( FRONTEND_RETIRED.MS_FLOWS * cpu_core@FRONTEND_RETIRED.MS_FLOWS@R ) / tma_info_thread_clks", "MetricGroup": "Fed;FetchLat;MicroSeq;Clocks_Retired", - "MetricName": "tma_info_frontend_ms_latency_ret" + "MetricName": "tma_info_frontend_ms_latency_ret", + "MetricThreshold": "tma_info_frontend_ms_latency_ret > 0.05" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired branches who got branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired branches who got branch address clears", "MetricExpr": "( FRONTEND_RETIRED.UNKNOWN_BRANCH * cpu_core@FRONTEND_RETIRED.UNKNOWN_BRANCH@R ) / tma_info_thread_clks", "MetricGroup": "Fed;FetchLat;Clocks_Retired", "MetricName": "tma_info_frontend_unknown_branches_ret" }, { - "BriefDescription": "Average Latency for L1 instruction cache misses.", + "BriefDescription": "Average Latency for L1 instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / ICACHE_DATA.STALL_PERIODS", "MetricGroup": "Fed;FetchLat;IcMiss;Metric", "MetricName": "tma_info_frontend_icache_miss_latency" }, { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", "MetricGroup": "DSBmiss;Fed;Inst_Metric", - "MetricName": "tma_info_frontend_ipdsb_miss_ret" + "MetricName": "tma_info_frontend_ipdsb_miss_ret", + "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache true code cacheline misses per kilo instruction", "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code" }, { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction", "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code_all" }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) )", - "MetricGroup": "DSBmiss;Fed;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_misses" + "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_misses", + "MetricThreshold": "tma_info_botlnk_dsb_misses > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) )", - "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_bandwidth" + "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", + "MetricThreshold": "tma_info_botlnk_dsb_bandwidth > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", + "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_ic_misses" + "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", + "MetricName": "tma_info_botlnk_l2_ic_misses", + "MetricThreshold": "tma_info_botlnk_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmispredict" + "MetricName": "tma_info_bad_spec_ipmispredict", + "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken" + "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", + "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional backward-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional backward-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN_BWD", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken_bwd" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional forward-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional forward-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN_FWD", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken_fwd" }, { - "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_ret" + "MetricName": "tma_info_bad_spec_ipmisp_ret", + "MetricThreshold": "tma_info_bad_spec_ipmisp_ret < 500" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_indirect" + "MetricName": "tma_info_bad_spec_ipmisp_indirect", + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 8 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", - "MetricGroup": "Bad;BrMispredicts;Core_Metric", - "MetricName": "tma_info_bad_spec_branch_misprediction_cost" + "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", + "MetricName": "tma_info_bad_spec_branch_misprediction_cost", + "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_mispredicts_resteers." }, { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", + "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", "MetricGroup": "BrMispredicts;Metric", "MetricName": "tma_info_bad_spec_spec_clears_ratio" }, { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", + "BriefDescription": "Fraction of branches that are non-taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_nt" }, { - "BriefDescription": "Fraction of branches that are forward taken conditionals.", + "BriefDescription": "Fraction of branches that are forward taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_TAKEN_BWD / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", - "MetricName": "tma_info_branches_cond_tk_bwd" + "MetricName": "tma_info_branches_cond_tk_bwd", + "MetricThreshold": "tma_info_branches_cond_tk_bwd > 0.3" }, { - "BriefDescription": "Fraction of branches that are forward taken conditionals.", + "BriefDescription": "Fraction of branches that are forward taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_TAKEN_FWD / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", - "MetricName": "tma_info_branches_cond_tk_fwd" + "MetricName": "tma_info_branches_cond_tk_fwd", + "MetricThreshold": "tma_info_branches_cond_tk_fwd > 0.2" }, { - "BriefDescription": "Fraction of branches that are CALL or RET.", + "BriefDescription": "Fraction of branches that are CALL or RET", "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_callret" }, { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", + "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN_BWD - BR_INST_RETIRED.COND_TAKEN_FWD - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_jump" }, { - "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group).", + "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group)", "MetricExpr": "1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk_bwd + tma_info_branches_cond_tk_fwd + tma_info_branches_callret + tma_info_branches_jump )", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_other_branches" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PENDING.LOAD / L1D_MISS.LOAD", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PENDING.LOAD / L1D_PENDING.LOAD_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki_load" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * L2_RQSTS.RFO_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_all" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", + "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_fb_hpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the Level 0 within L1D cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the Level 0 within L1D cache [GB / sec]", "MetricExpr": "64 * L1D.L0_REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1dl0_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Offcore;Metric", "MetricName": "tma_info_memory_l3_cache_access_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 4 * tma_info_thread_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", - "MetricName": "tma_info_memory_tlb_page_walks_utilization" + "MetricName": "tma_info_memory_tlb_page_walks_utilization", + "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Fed;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_code_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_load_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand loads.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand loads", "MetricExpr": "( MEM_INST_RETIRED.STLB_MISS_LOADS * cpu_core@MEM_INST_RETIRED.STLB_MISS_LOADS@R ) / tma_info_thread_clks", "MetricGroup": "Mem;MemoryTLB;Clocks_Retired", - "MetricName": "tma_info_memory_tlb_load_stlb_miss_ret" + "MetricName": "tma_info_memory_tlb_load_stlb_miss_ret", + "MetricThreshold": "tma_info_memory_load_stlb_miss_ret > 0.05" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand stores.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand stores", "MetricExpr": "( MEM_INST_RETIRED.STLB_MISS_STORES * cpu_core@MEM_INST_RETIRED.STLB_MISS_STORES@R ) / tma_info_thread_clks", "MetricGroup": "Mem;MemoryTLB;Clocks_Retired", - "MetricName": "tma_info_memory_tlb_store_stlb_miss_ret" + "MetricName": "tma_info_memory_tlb_store_stlb_miss_ret", + "MetricThreshold": "tma_info_memory_store_stlb_miss_ret > 0.05" }, { - "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses.", + "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses", "MetricExpr": "L2_LINES_OUT.USELESS_HWPF / ( L2_LINES_OUT.SILENT + L2_LINES_OUT.NON_SILENT )", "MetricGroup": "Prefetches;Metric", - "MetricName": "tma_info_memory_prefetches_useless_hwpf" + "MetricName": "tma_info_memory_prefetches_useless_hwpf", + "MetricThreshold": "tma_info_memory_useless_hwpf > 0.15" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Latency for L3 cache miss demand Loads.", + "BriefDescription": "Average Latency for L3 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.L3_MISS_DEMAND_DATA_RD / OFFCORE_REQUESTS.L3_MISS_DEMAND_DATA_RD", "MetricGroup": "Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l3_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=0x1@", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", + "BriefDescription": "Un-cacheable retired load per kilo instruction", "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_uc_load_pki" }, { - "BriefDescription": "\"Bus lock\" per kilo instruction.", + "BriefDescription": "\"Bus lock\" per kilo instruction", "MetricExpr": "1000 * SQ_MISC.BUS_LOCK / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_bus_lock_pki" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", - "MetricName": "tma_info_system_kernel_utilization" + "MetricName": "tma_info_system_kernel_utilization", + "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states.", + "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states", "MetricExpr": "CPU_CLK_UNHALTED.C0_WAIT / tma_info_thread_clks", "MetricGroup": "C0Wait;Metric", - "MetricName": "tma_info_system_c0_wait" - }, - { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", - "MetricExpr": "64 * UNC_ARB_TRK_REQUESTS.ALL / ( 1000000 ) / tma_info_system_time / 1000", - "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec", - "MetricName": "tma_info_system_dram_bw_use" + "MetricName": "tma_info_system_c0_wait", + "MetricThreshold": "tma_info_system_c0_wait > 0.05" }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "power@energy\\-pkg@ * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", - "MetricName": "tma_info_system_mux" + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", - "MetricName": "tma_info_system_ipfarbranch" + "MetricName": "tma_info_system_ipfarbranch", + "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" } ] \ No newline at end of file diff --git a/MTL/metrics/meteorlake_metrics_redwoodcove_core.json b/MTL/metrics/meteorlake_metrics_redwoodcove_core.json index 062c0590..3a488b7b 100644 --- a/MTL/metrics/meteorlake_metrics_redwoodcove_core.json +++ b/MTL/metrics/meteorlake_metrics_redwoodcove_core.json @@ -2,8 +2,8 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for Intel(R) Core(TM) processors based on Meteor Lake performance hybrid architecture0", - "DatePublished": "11/12/2024", - "Version": "0", + "DatePublished": "11/18/2024", + "Version": "1.0", "Legend": "", "TmaVersion": "5.01", "TmaFlavor": "Full" @@ -111,7 +111,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Mispredictions > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Mispredictions" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Mispredictions > 20", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP", @@ -199,7 +207,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Big_Code > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Big_Code" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Big_Code > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB", @@ -356,7 +372,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Instruction_Fetch_BW" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", @@ -557,7 +581,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore", @@ -810,7 +842,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Latency" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;Mem;MemoryLat;Offcore", @@ -1007,7 +1047,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Data_TLBs" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore", @@ -1216,7 +1264,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Synchronization > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Synchronization" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Synchronization > 10", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;LockCont;Mem;Offcore", @@ -1300,7 +1356,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Compute_Bound_Est" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20", + "ThresholdIssues": "$issueComp" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Cor", @@ -1509,7 +1573,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Irregular_Overhead > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Irregular_Overhead" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Irregular_Overhead > 10", + "ThresholdIssues": "$issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BvIO;Cor;Ret", @@ -1930,7 +2002,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Other_Bottlenecks" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;Cor;Offcore", @@ -1966,7 +2046,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Branching_Overhead > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Branching_Overhead" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Bottleneck_Branching_Overhead > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Ret", @@ -2030,7 +2118,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Useful_Work > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Useful_Work" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Useful_Work > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;Ret", @@ -2074,7 +2170,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;BvIO;TmaL1;PGO", @@ -2123,7 +2227,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Frontend;TmaL2", @@ -2132,6 +2248,7 @@ { "MetricName": "ICache_Misses", "LegacyName": "metric_TMA_....ICache_Misses(%)", + "ParentCategory": "Fetch_Latency", "Level": 3, "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", "UnitOfMeasure": "percent", @@ -2310,7 +2427,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB", @@ -2351,7 +2484,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2384,7 +2537,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2421,7 +2594,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2458,7 +2655,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2467,6 +2688,7 @@ { "MetricName": "Branch_Resteers", "LegacyName": "metric_TMA_....Branch_Resteers(%)", + "ParentCategory": "Fetch_Latency", "Level": 3, "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings.", "UnitOfMeasure": "percent", @@ -2504,8 +2726,8 @@ "Value": "metric_TMA_Frontend_Bound(%)" } ], - "Formula": "a > 5 && b > 10 && c > 15", - "BaseFormula": "metric_TMA_....Branch_Resteers(%) > 5 && metric_TMA_..Fetch_Latency(%) > 10 && metric_TMA_Frontend_Bound(%) > 15", + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", "ThresholdIssues": "~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -2658,13 +2880,13 @@ "Value": "metric_TMA_Frontend_Bound(%)" } ], - "Formula": "a > 5 && b > 5 && c > 10 && d > 15", - "BaseFormula": "metric_TMA_......Clears_Resteers(%) > 5 && metric_TMA_....Branch_Resteers(%) > 5 && metric_TMA_..Fetch_Latency(%) > 10 && metric_TMA_Frontend_Bound(%) > 15", + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", "ThresholdIssues": "$issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;MachineClears", - "LocateWith": "INT_MISC.CLEAR_RESTEER_CYCLES" + "LocateWith": " INT_MISC.CLEAR_RESTEER_CYCLES" }, { "MetricName": "Unknown_Branches", @@ -2718,6 +2940,7 @@ { "MetricName": "MS_Switches", "LegacyName": "metric_TMA_....MS_Switches(%)", + "ParentCategory": "Fetch_Latency", "Level": 3, "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals.", "UnitOfMeasure": "percent", @@ -2759,13 +2982,13 @@ "Value": "metric_TMA_Frontend_Bound(%)" } ], - "Formula": "a > 5 && b > 10 && c > 15", - "BaseFormula": "metric_TMA_....MS_Switches(%) > 5 && metric_TMA_..Fetch_Latency(%) > 10 && metric_TMA_Frontend_Bound(%) > 15", + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", "ThresholdIssues": "$issueMC, $issueMS, $issueMV, $issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MicroSeq", - "LocateWith": "FRONTEND_RETIRED.MS_FLOWS" + "LocateWith": " FRONTEND_RETIRED.MS_FLOWS " }, { "MetricName": "LCP", @@ -2790,7 +3013,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LCP(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -2799,6 +3038,7 @@ { "MetricName": "DSB_Switches", "LegacyName": "metric_TMA_....DSB_Switches(%)", + "ParentCategory": "Fetch_Latency", "Level": 3, "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty.", "UnitOfMeasure": "percent", @@ -2832,13 +3072,13 @@ "Value": "metric_TMA_Frontend_Bound(%)" } ], - "Formula": "a > 5 && b > 10 && c > 15", - "BaseFormula": "metric_TMA_....DSB_Switches(%) > 5 && metric_TMA_..Fetch_Latency(%) > 10 && metric_TMA_Frontend_Bound(%) > 15", + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchLat", - "LocateWith": "FRONTEND_RETIRED.DSB_MISS_PS" + "LocateWith": " FRONTEND_RETIRED.DSB_MISS" }, { "MetricName": "Fetch_Bandwidth", @@ -2883,7 +3123,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;Frontend;TmaL2", @@ -2892,6 +3140,7 @@ { "MetricName": "MITE", "LegacyName": "metric_TMA_....MITE(%)", + "ParentCategory": "Fetch_Bandwidth", "Level": 3, "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck.", "UnitOfMeasure": "percent", @@ -2938,13 +3187,13 @@ "Value": "metric_TMA_..Fetch_Bandwidth(%)" } ], - "Formula": "a > 10 && b > 20", - "BaseFormula": "metric_TMA_....MITE(%) > 10 && metric_TMA_..Fetch_Bandwidth(%) > 20", + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", - "LocateWith": "FRONTEND_RETIRED.ANY_DSB_MISS" + "LocateWith": " FRONTEND_RETIRED.ANY_DSB_MISS" }, { "MetricName": "Decoder0_Alone", @@ -3011,6 +3260,7 @@ { "MetricName": "DSB", "LegacyName": "metric_TMA_....DSB(%)", + "ParentCategory": "Fetch_Bandwidth", "Level": 3, "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here.", "UnitOfMeasure": "percent", @@ -3057,8 +3307,8 @@ "Value": "metric_TMA_..Fetch_Bandwidth(%)" } ], - "Formula": "a > 15 && b > 20", - "BaseFormula": "metric_TMA_....DSB(%) > 15 && metric_TMA_..Fetch_Bandwidth(%) > 20", + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -3105,7 +3355,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....LSD(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LSD(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....LSD(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;LSD", @@ -3159,7 +3421,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 5 & b > 20", + "BaseFormula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -3203,7 +3477,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", @@ -3244,7 +3526,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2", @@ -3277,9 +3571,25 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Cond_NT_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Cond_NT_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Cond_NT_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" + }, + "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", "LocateWith": "" }, @@ -3310,7 +3620,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Cond_TK_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Cond_TK_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Cond_TK_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -3343,7 +3669,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Ind_Call_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ind_Call_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Ind_Call_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -3384,7 +3726,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Ind_Jump_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ind_Jump_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Ind_Jump_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -3417,7 +3775,23 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_....Ret_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ret_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Ret_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", @@ -3426,6 +3800,7 @@ { "MetricName": "Other_Mispredicts", "LegacyName": "metric_TMA_....Other_Mispredicts(%)", + "ParentCategory": "Branch_Mispredicts", "Level": 3, "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", "UnitOfMeasure": "percent", @@ -3483,8 +3858,8 @@ "Value": "metric_TMA_Bad_Speculation(%)" } ], - "Formula": "a > 5 && b > 10 && c > 15", - "BaseFormula": "metric_TMA_....Other_Mispredicts(%) > 5 && metric_TMA_..Branch_Mispredicts(%) > 10 && metric_TMA_Bad_Speculation(%) > 15", + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -3534,7 +3909,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueMC, $issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2", @@ -3543,6 +3930,7 @@ { "MetricName": "Other_Nukes", "LegacyName": "metric_TMA_....Other_Nukes(%)", + "ParentCategory": "Machine_Clears", "Level": 3, "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", "UnitOfMeasure": "percent", @@ -3604,8 +3992,8 @@ "Value": "metric_TMA_Bad_Speculation(%)" } ], - "Formula": "a > 5 && b > 10 && c > 15", - "BaseFormula": "metric_TMA_....Other_Nukes(%) > 5 && metric_TMA_..Machine_Clears(%) > 10 && metric_TMA_Bad_Speculation(%) > 15", + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -3642,7 +4030,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;TmaL1", @@ -3651,6 +4047,7 @@ { "MetricName": "Memory_Bound", "LegacyName": "metric_TMA_..Memory_Bound(%)", + "ParentCategory": "Backend_Bound", "Level": 2, "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two).", "UnitOfMeasure": "percent", @@ -3692,8 +4089,8 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 20 && b > 20", - "BaseFormula": "metric_TMA_..Memory_Bound(%) > 20 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 20 & b > 20", + "BaseFormula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -3752,6 +4149,7 @@ { "MetricName": "DTLB_Load", "LegacyName": "metric_TMA_......DTLB_Load(%)", + "ParentCategory": "L1_Bound", "Level": 4, "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss.", "UnitOfMeasure": "percent", @@ -3797,13 +4195,13 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 10 && b > 10 && c > 20 && d > 20", - "BaseFormula": "metric_TMA_......DTLB_Load(%) > 10 && metric_TMA_....L1_Bound(%) > 10 && metric_TMA_..Memory_Bound(%) > 20 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", - "LocateWith": "MEM_INST_RETIRED.STLB_MISS_LOADS_PS" + "LocateWith": " MEM_INST_RETIRED.STLB_MISS_LOADS" }, { "MetricName": "Load_STLB_Hit", @@ -3911,8 +4309,8 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 5 && b > 10 && c > 10 && d > 20 && e > 20", - "BaseFormula": "metric_TMA_........Load_STLB_Miss(%) > 5 && metric_TMA_......DTLB_Load(%) > 10 && metric_TMA_....L1_Bound(%) > 10 && metric_TMA_..Memory_Bound(%) > 20 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -4301,6 +4699,7 @@ { "MetricName": "Split_Loads", "LegacyName": "metric_TMA_......Split_Loads(%)", + "ParentCategory": "L1_Bound", "Level": 4, "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. ", "UnitOfMeasure": "percent", @@ -4336,27 +4735,15 @@ { "Alias": "a", "Value": "metric_TMA_......Split_Loads(%)" - }, - { - "Alias": "b", - "Value": "metric_TMA_....L1_Bound(%)" - }, - { - "Alias": "c", - "Value": "metric_TMA_..Memory_Bound(%)" - }, - { - "Alias": "d", - "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 20 && b > 10 && c > 20 && d > 20", - "BaseFormula": "metric_TMA_......Split_Loads(%) > 20 && metric_TMA_....L1_Bound(%) > 10 && metric_TMA_..Memory_Bound(%) > 20 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......Split_Loads(%) > 30", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", - "LocateWith": "MEM_INST_RETIRED.SPLIT_LOADS_PS" + "LocateWith": " MEM_INST_RETIRED.SPLIT_LOADS" }, { "MetricName": "FB_Full", @@ -4436,13 +4823,13 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 5 && b > 20 && c > 20", - "BaseFormula": "metric_TMA_....L2_Bound(%) > 5 && metric_TMA_..Memory_Bound(%) > 20 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem", - "LocateWith": "MEM_LOAD_RETIRED.L2_HIT_PS" + "LocateWith": " MEM_LOAD_RETIRED.L2_HIT" }, { "MetricName": "L2_Hit_Latency", @@ -4570,6 +4957,7 @@ { "MetricName": "Contested_Accesses", "LegacyName": "metric_TMA_......Contested_Accesses(%)", + "ParentCategory": "L3_Bound", "Level": 4, "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing.", "UnitOfMeasure": "percent", @@ -4648,13 +5036,13 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 5 && b > 5 && c > 20 && d > 20", - "BaseFormula": "metric_TMA_......Contested_Accesses(%) > 5 && metric_TMA_....L3_Bound(%) > 5 && metric_TMA_..Memory_Bound(%) > 20 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvMS;DataSharing;Offcore;Snoop", - "LocateWith": "MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS" + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", + "LocateWith": " MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS" }, { "MetricName": "Data_Sharing", @@ -4749,6 +5137,7 @@ { "MetricName": "L3_Hit_Latency", "LegacyName": "metric_TMA_......L3_Hit_Latency(%)", + "ParentCategory": "L3_Bound", "Level": 4, "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings.", "UnitOfMeasure": "percent", @@ -4811,13 +5200,13 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 10 && b > 5 && c > 20 && d > 20", - "BaseFormula": "metric_TMA_......L3_Hit_Latency(%) > 10 && metric_TMA_....L3_Bound(%) > 5 && metric_TMA_..Memory_Bound(%) > 20 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 10 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "$issueLat, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", - "LocateWith": "MEM_LOAD_RETIRED.L3_HIT_PS" + "LocateWith": " MEM_LOAD_RETIRED.L3_HIT" }, { "MetricName": "SQ_Full", @@ -4909,13 +5298,13 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 10 && b > 20 && c > 20", - "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 && metric_TMA_..Memory_Bound(%) > 20 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", - "LocateWith": "MEM_LOAD_RETIRED.L3_MISS_PS" + "LocateWith": " MEM_LOAD_RETIRED.L3_MISS" }, { "MetricName": "MEM_Bandwidth", @@ -5011,8 +5400,8 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 10 && b > 10 && c > 20 && d > 20", - "BaseFormula": "metric_TMA_......MEM_Latency(%) > 10 && metric_TMA_....DRAM_Bound(%) > 10 && metric_TMA_..Memory_Bound(%) > 20 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -5067,6 +5456,7 @@ { "MetricName": "Store_Latency", "LegacyName": "metric_TMA_......Store_Latency(%)", + "ParentCategory": "Store_Bound", "Level": 4, "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)", "UnitOfMeasure": "percent", @@ -5116,12 +5506,12 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 10 && b > 20 && c > 20 && d > 20", - "BaseFormula": "metric_TMA_......Store_Latency(%) > 10 && metric_TMA_....Store_Bound(%) > 20 && metric_TMA_..Memory_Bound(%) > 20 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 10 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "$issueRFO, $issueSL, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvML;MemoryLat;Offcore", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore", "LocateWith": "" }, { @@ -5189,6 +5579,7 @@ { "MetricName": "Split_Stores", "LegacyName": "metric_TMA_......Split_Stores(%)", + "ParentCategory": "Store_Bound", "Level": 4, "BriefDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity.", "UnitOfMeasure": "percent", @@ -5230,13 +5621,13 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 20 && b > 20 && c > 20 && d > 20", - "BaseFormula": "metric_TMA_......Split_Stores(%) > 20 && metric_TMA_....Store_Bound(%) > 20 && metric_TMA_..Memory_Bound(%) > 20 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", - "LocateWith": "MEM_INST_RETIRED.SPLIT_STORES_PS" + "LocateWith": " MEM_INST_RETIRED.SPLIT_STORES" }, { "MetricName": "Streaming_Stores", @@ -5290,6 +5681,7 @@ { "MetricName": "DTLB_Store", "LegacyName": "metric_TMA_......DTLB_Store(%)", + "ParentCategory": "Store_Bound", "Level": 4, "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data.", "UnitOfMeasure": "percent", @@ -5348,13 +5740,13 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 5 && b > 20 && c > 20 && d > 20", - "BaseFormula": "metric_TMA_......DTLB_Store(%) > 5 && metric_TMA_....Store_Bound(%) > 20 && metric_TMA_..Memory_Bound(%) > 20 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", - "LocateWith": "MEM_INST_RETIRED.STLB_MISS_STORES_PS" + "LocateWith": " MEM_INST_RETIRED.STLB_MISS_STORES" }, { "MetricName": "Store_STLB_Hit", @@ -5488,8 +5880,8 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 5 && b > 5 && c > 20 && d > 20 && e > 20", - "BaseFormula": "metric_TMA_........Store_STLB_Miss(%) > 5 && metric_TMA_......DTLB_Store(%) > 5 && metric_TMA_....Store_Bound(%) > 20 && metric_TMA_..Memory_Bound(%) > 20 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -5777,7 +6169,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2;Compute", @@ -5786,6 +6190,7 @@ { "MetricName": "Divider", "LegacyName": "metric_TMA_....Divider(%)", + "ParentCategory": "Core_Bound", "Level": 3, "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication.", "UnitOfMeasure": "percent", @@ -5819,13 +6224,13 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 20 && b > 10 && c > 20", - "BaseFormula": "metric_TMA_....Divider(%) > 20 && metric_TMA_..Core_Bound(%) > 10 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 20 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB", - "LocateWith": "ARITH.DIV_ACTIVE" + "LocateWith": " ARITH.DIV_ACTIVE" }, { "MetricName": "FP_Divider", @@ -5956,7 +6361,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;PortsUtil", @@ -5965,6 +6386,7 @@ { "MetricName": "Slow_Pause", "LegacyName": "metric_TMA_......Slow_Pause(%)", + "ParentCategory": "Serializing_Operation", "Level": 4, "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", "UnitOfMeasure": "percent", @@ -6002,13 +6424,13 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 5 && b > 10 && c > 10 && d > 20", - "BaseFormula": "metric_TMA_......Slow_Pause(%) > 5 && metric_TMA_....Serializing_Operation(%) > 10 && metric_TMA_..Core_Bound(%) > 10 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", - "LocateWith": "CPU_CLK_UNHALTED.PAUSE_INST" + "LocateWith": " CPU_CLK_UNHALTED.PAUSE_INST" }, { "MetricName": "C01_Wait", @@ -6033,7 +6455,76 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......C01_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......C01_Wait(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......C01_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" + }, + "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", + "MetricGroup": "C0Wait", + "LocateWith": "" + }, + { + "MetricName": "C02_Wait", + "LegacyName": "metric_TMA_......C02_Wait(%)", + "ParentCategory": "Serializing_Operation", + "Level": 4, + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings).", + "UnitOfMeasure": "percent", + "Events": [ + { + "Name": "CPU_CLK_UNHALTED.C02", + "Alias": "a" + }, + { + "Name": "CPU_CLK_UNHALTED.THREAD", + "Alias": "b" + } + ], + "Constants": [], + "Formula": "100 * ( a / ( b ) )", + "BaseFormula": " cpu_clk_unhalted.c02 / tma_info_thread_clks", + "Category": "TMA", + "CountDomain": "Clocks", + "Threshold": { + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......C02_Wait(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......C02_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", @@ -6062,7 +6553,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Memory_Fence(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Memory_Fence(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Memory_Fence(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6071,6 +6582,7 @@ { "MetricName": "Ports_Utilization", "LegacyName": "metric_TMA_....Ports_Utilization(%)", + "ParentCategory": "Core_Bound", "Level": 3, "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations.", "UnitOfMeasure": "percent", @@ -6144,8 +6656,8 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 15 && b > 10 && c > 20", - "BaseFormula": "metric_TMA_....Ports_Utilization(%) > 15 && metric_TMA_..Core_Bound(%) > 10 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 15 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -6208,6 +6720,7 @@ { "MetricName": "Mixing_Vectors", "LegacyName": "metric_TMA_........Mixing_Vectors(%)", + "ParentCategory": "Ports_Utilized_0", "Level": 5, "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic.", "UnitOfMeasure": "percent", @@ -6331,13 +6844,13 @@ "Value": "metric_TMA_Backend_Bound(%)" } ], - "Formula": "a > 15 && b > 15 && c > 10 && d > 20", - "BaseFormula": "metric_TMA_......Ports_Utilized_2(%) > 15 && metric_TMA_....Ports_Utilization(%) > 15 && metric_TMA_..Core_Bound(%) > 10 && metric_TMA_Backend_Bound(%) > 20", + "Formula": "a > 15 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", - "LocateWith": "EXE_ACTIVITY.2_PORTS_UTIL" + "LocateWith": " EXE_ACTIVITY.2_PORTS_UTIL" }, { "MetricName": "Ports_Utilized_3m", @@ -6391,6 +6904,7 @@ { "MetricName": "ALU_Op_Utilization", "LegacyName": "metric_TMA_........ALU_Op_Utilization(%)", + "ParentCategory": "Ports_Utilized_3m", "Level": 5, "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", "UnitOfMeasure": "percent", @@ -6547,7 +7061,7 @@ }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", - "LocateWith": "UOPS_DISPATCHED.PORT_1" + "LocateWith": " UOPS_DISPATCHED.PORT_1 " }, { "MetricName": "Port_6", @@ -6602,6 +7116,7 @@ { "MetricName": "Load_Op_Utilization", "LegacyName": "metric_TMA_........Load_Op_Utilization(%)", + "ParentCategory": "Ports_Utilized_3m", "Level": 5, "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "UnitOfMeasure": "percent", @@ -6646,7 +7161,7 @@ }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", - "LocateWith": "UOPS_DISPATCHED.PORT_2_3_10" + "LocateWith": " UOPS_DISPATCHED.PORT_2_3_10" }, { "MetricName": "Store_Op_Utilization", @@ -6732,7 +7247,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Retiring(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 70 | b > 10", + "BaseFormula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;TmaL1", @@ -6773,15 +7300,24 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Retire;TmaL2", - "LocateWith": "INST_RETIRED.PREC_DIST" - }, + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" + }, + "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", + "MetricGroup": "Retire;TmaL2", + "LocateWith": "INST_RETIRED.PREC_DIST" + }, { "MetricName": "FP_Arith", "LegacyName": "metric_TMA_....FP_Arith(%)", + "ParentCategory": "Light_Operations", "Level": 3, "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting.", "UnitOfMeasure": "percent", @@ -6839,8 +7375,8 @@ "Value": "metric_TMA_..Light_Operations(%)" } ], - "Formula": "a > 20 && b > 60", - "BaseFormula": "metric_TMA_....FP_Arith(%) > 20 && metric_TMA_..Light_Operations(%) > 60", + "Formula": "a > 20 & b > 60", + "BaseFormula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -6961,8 +7497,8 @@ "Value": "metric_TMA_..Light_Operations(%)" } ], - "Formula": "a > 10 && b > 20 && c > 60", - "BaseFormula": "metric_TMA_......FP_Scalar(%) > 10 && metric_TMA_....FP_Arith(%) > 20 && metric_TMA_..Light_Operations(%) > 60", + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -7033,8 +7569,9 @@ { "MetricName": "FP_Vector_128b", "LegacyName": "metric_TMA_........FP_Vector_128b(%)", + "ParentCategory": "FP_Vector", "Level": 5, - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL.", "UnitOfMeasure": "percent", "Events": [ { @@ -7090,8 +7627,8 @@ "Value": "metric_TMA_..Light_Operations(%)" } ], - "Formula": "a > 10 && b > 10 && c > 20 && d > 60", - "BaseFormula": "metric_TMA_........FP_Vector_128b(%) > 10 && metric_TMA_......FP_Vector(%) > 10 && metric_TMA_....FP_Arith(%) > 20 && metric_TMA_..Light_Operations(%) > 60", + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -7170,6 +7707,7 @@ { "MetricName": "Int_Operations", "LegacyName": "metric_TMA_....Int_Operations(%)", + "ParentCategory": "Light_Operations", "Level": 3, "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired). Vector/Matrix Int operations and shuffles are counted. Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain.", "UnitOfMeasure": "percent", @@ -7231,8 +7769,8 @@ "Value": "metric_TMA_..Light_Operations(%)" } ], - "Formula": "a > 10 && b > 60", - "BaseFormula": "metric_TMA_....Int_Operations(%) > 10 && metric_TMA_..Light_Operations(%) > 60", + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -7365,8 +7903,8 @@ "Value": "metric_TMA_..Light_Operations(%)" } ], - "Formula": "a > 10 && b > 10 && c > 60", - "BaseFormula": "metric_TMA_......Int_Vector_256b(%) > 10 && metric_TMA_....Int_Operations(%) > 10 && metric_TMA_..Light_Operations(%) > 60", + "Formula": "a > 10 & b > 10 & c > 60", + "BaseFormula": "metric_TMA_......Int_Vector_256b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -7416,7 +7954,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Memory_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -7425,6 +7975,7 @@ { "MetricName": "Fused_Instructions", "LegacyName": "metric_TMA_....Fused_Instructions(%)", + "ParentCategory": "Light_Operations", "Level": 3, "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Other_Light_Ops in MTL!)}", "UnitOfMeasure": "percent", @@ -7474,8 +8025,8 @@ "Value": "metric_TMA_..Light_Operations(%)" } ], - "Formula": "a > 10 && b > 60", - "BaseFormula": "metric_TMA_....Fused_Instructions(%) > 10 && metric_TMA_..Light_Operations(%) > 60", + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -7529,7 +8080,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Non_Fused_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -7538,6 +8101,7 @@ { "MetricName": "Other_Light_Ops", "LegacyName": "metric_TMA_....Other_Light_Ops(%)", + "ParentCategory": "Light_Operations", "Level": 3, "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting", "UnitOfMeasure": "percent", @@ -7631,8 +8195,8 @@ "Value": "metric_TMA_..Light_Operations(%)" } ], - "Formula": "a > 30 && b > 60", - "BaseFormula": "metric_TMA_....Other_Light_Ops(%) > 30 && metric_TMA_..Light_Operations(%) > 60", + "Formula": "a > 30 & b > 60", + "BaseFormula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -7761,8 +8325,8 @@ "Value": "metric_TMA_..Light_Operations(%)" } ], - "Formula": "a > 10 && b > 30 && c > 60", - "BaseFormula": "metric_TMA_......Shuffles_256b(%) > 10 && metric_TMA_....Other_Light_Ops(%) > 30 && metric_TMA_..Light_Operations(%) > 60", + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Shuffles_256b(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -7804,7 +8368,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -7813,8 +8385,9 @@ { "MetricName": "Few_Uops_Instructions", "LegacyName": "metric_TMA_....Few_Uops_Instructions(%)", + "ParentCategory": "Heavy_Operations", "Level": 3, - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or up to ([SNB+] four; [ADL+] five) uops. This highly-correlates with the number of uops in such instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions.", "UnitOfMeasure": "percent", "Events": [ { @@ -7862,8 +8435,8 @@ "Value": "metric_TMA_..Heavy_Operations(%)" } ], - "Formula": "a > 5 && b > 10", - "BaseFormula": "metric_TMA_....Few_Uops_Instructions(%) > 5 && metric_TMA_..Heavy_Operations(%) > 10", + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", @@ -7893,7 +8466,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueMC, $issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -7902,6 +8487,7 @@ { "MetricName": "Assists", "LegacyName": "metric_TMA_......Assists(%)", + "ParentCategory": "Microcode_Sequencer", "Level": 4, "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases.", "UnitOfMeasure": "percent", @@ -7935,13 +8521,13 @@ "Value": "metric_TMA_..Heavy_Operations(%)" } ], - "Formula": "a > 10 && b > 5 && c > 10", - "BaseFormula": "metric_TMA_......Assists(%) > 10 && metric_TMA_....Microcode_Sequencer(%) > 5 && metric_TMA_..Heavy_Operations(%) > 10", + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO", - "LocateWith": "ASSISTS.ANY" + "LocateWith": " ASSISTS.ANY" }, { "MetricName": "Page_Faults", @@ -8057,6 +8643,7 @@ { "MetricName": "CISC", "LegacyName": "metric_TMA_......CISC(%)", + "ParentCategory": "Microcode_Sequencer", "Level": 4, "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources.", "UnitOfMeasure": "percent", @@ -8185,7 +8772,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_Core_Bound_Likely > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_Core_Bound_Likely" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Botlnk_Core_Bound_Likely > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;SMT", @@ -8212,6 +8807,11 @@ "BaseFormula": " inst_retired.any / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Summary", "LocateWith": "" @@ -8254,7 +8854,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UopPI > 1.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UopPI" + } + ], + "Formula": "a > 1.05", + "BaseFormula": "metric_TMA_Info_Thread_UopPI > 1.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret;Retire", @@ -8298,7 +8906,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UpTB < 6 * 1.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UpTB" + } + ], + "Formula": "a < 6 * 1.5", + "BaseFormula": "metric_TMA_Info_Thread_UpTB < 6 * 1.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW", @@ -8325,6 +8941,11 @@ "BaseFormula": " 1 / tma_info_thread_ipc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Mem", "LocateWith": "" @@ -8346,6 +8967,11 @@ "BaseFormula": " cpu_clk_unhalted.thread", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", "LocateWith": "" @@ -8367,6 +8993,11 @@ "BaseFormula": " topdown.slots:perf_metrics", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", "LocateWith": "" @@ -8401,6 +9032,11 @@ "BaseFormula": " tma_info_thread_slots / ( topdown.slots:percore / 2 ) if smt_on else 1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "SMT;TmaL1", "LocateWith": "" @@ -8426,6 +9062,11 @@ "BaseFormula": " uops_executed.thread / uops_issued.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline", "LocateWith": "" @@ -8464,6 +9105,11 @@ "BaseFormula": " inst_retired.any / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;SMT;TmaL1", "LocateWith": "" @@ -8514,6 +9160,11 @@ "BaseFormula": " ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.256b_packed_single ) / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Flops", "LocateWith": "" @@ -8560,6 +9211,11 @@ "BaseFormula": " ( fp_arith_dispatched.port_0 + fp_arith_dispatched.port_1 + fp_arith_dispatched.port_5 ) / ( 2 * tma_info_core_core_clks )", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -8585,6 +9241,11 @@ "BaseFormula": " uops_executed.thread / uops_executed.thread:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "LocateWith": "" @@ -8610,6 +9271,11 @@ "BaseFormula": " uops_executed.thread / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -8644,6 +9310,11 @@ "BaseFormula": " cpu_clk_unhalted.distributed if smt_on else tma_info_thread_clks", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "SMT", "LocateWith": "" @@ -8670,7 +9341,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpLoad < 3" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpLoad" + } + ], + "Formula": "a < 3", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpLoad < 3", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -8698,7 +9377,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpStore < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpStore" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpStore < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -8726,7 +9413,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpBranch < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpBranch" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpBranch < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;InsType", @@ -8754,7 +9449,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpCall < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpCall" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpCall < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", @@ -8782,7 +9485,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpTB < 6 * 2 + 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpTB" + } + ], + "Formula": "a < 6 * 2 + 1", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpTB < 6 * 2 + 1", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO", @@ -8809,6 +9520,11 @@ "BaseFormula": " br_inst_retired.all_branches / br_inst_retired.near_taken", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", "LocateWith": "" @@ -8847,7 +9563,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpFLOP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -8879,7 +9603,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -8907,7 +9639,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -8935,7 +9675,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -8967,7 +9715,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX128" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -8999,7 +9755,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX256" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -9026,6 +9790,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / cpu_clk_unhalted.pause_inst", "Category": "TMA", "CountDomain": "Inst_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", "LocateWith": "" @@ -9052,7 +9821,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpSWPF" + } + ], + "Formula": "a < 100", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -9075,6 +9852,11 @@ "BaseFormula": " inst_retired.any", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;TmaL1", "LocateWith": "INST_RETIRED.PREC_DIST" @@ -9116,6 +9898,11 @@ "BaseFormula": " ( tma_retiring * tma_info_thread_slots ) / uops_retired.slots:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret", "LocateWith": "" @@ -9142,7 +9929,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Pipeline_Strings_Cycles > 0.1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Pipeline_Strings_Cycles" + } + ], + "Formula": "a > 0.1", + "BaseFormula": "metric_TMA_Info_Pipeline_Strings_Cycles > 0.1", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq;Pipeline;Ret", @@ -9170,7 +9965,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Pipeline_IpAssist < 100000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Pipeline_IpAssist" + } + ], + "Formula": "a < 100000", + "BaseFormula": "metric_TMA_Info_Pipeline_IpAssist < 100000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", @@ -9180,7 +9983,7 @@ "MetricName": "Info_Pipeline_Execute", "LegacyName": "metric_TMA_Info_Pipeline_Execute", "Level": 1, - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core", + "BriefDescription": "", "UnitOfMeasure": "", "Events": [ { @@ -9210,6 +10013,11 @@ "BaseFormula": " uops_executed.thread / ( ( uops_executed.core_cycles_ge_1 / 2 ) if smt_on else uops_executed.thread:c1 )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "LocateWith": "" @@ -9235,6 +10043,11 @@ "BaseFormula": " lsd.uops / lsd.cycles_active", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -9260,6 +10073,11 @@ "BaseFormula": " idq.dsb_uops / idq.dsb_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -9285,6 +10103,11 @@ "BaseFormula": " idq.mite_uops / idq.mite_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -9310,6 +10133,11 @@ "BaseFormula": " uops_issued.any / uops_issued.any:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -9335,6 +10163,11 @@ "BaseFormula": " lsd.uops / ( uops_issued.any )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;LSD", "LocateWith": "" @@ -9361,7 +10194,19 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 && IPC / 6 > 0.35" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_DSB_Coverage" + }, + { + "Alias": "b", + "Value": "IPC" + } + ], + "Formula": "a < 0.7 & b / 6 > 0.35", + "BaseFormula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 6 > 0.35", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -9388,6 +10233,11 @@ "BaseFormula": " int_misc.unknown_branch_cycles / int_misc.unknown_branch_cycles:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed", "LocateWith": "" @@ -9413,6 +10263,11 @@ "BaseFormula": " dsb2mite_switches.penalty_cycles / dsb2mite_switches.penalty_cycles:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss", "LocateWith": "" @@ -9438,6 +10293,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -9468,7 +10328,15 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_DSB_Switches_Ret > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_DSB_Switches_Ret" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_Frontend_DSB_Switches_Ret > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed;FetchLat", @@ -9500,9 +10368,17 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_MS_Latency_Ret > 0.05" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_MS_Latency_Ret" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_Frontend_MS_Latency_Ret > 0.05", + "ThresholdIssues": "" + }, + "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;MicroSeq", "LocateWith": "" }, @@ -9531,6 +10407,11 @@ "BaseFormula": " ( frontend_retired.unknown_branch * frontend_retired.unknown_branch:retire_latency ) / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Clocks_Retired", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat", "LocateWith": "" @@ -9556,6 +10437,11 @@ "BaseFormula": " icache_data.stalls / icache_data.stall_periods", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", "LocateWith": "" @@ -9582,7 +10468,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret" + } + ], + "Formula": "a < 50", + "BaseFormula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -9609,6 +10503,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / baclears.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed", "LocateWith": "" @@ -9634,6 +10533,11 @@ "BaseFormula": " 1000 * frontend_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -9659,6 +10563,11 @@ "BaseFormula": " 1000 * l2_rqsts.code_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -9790,7 +10699,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Misses > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Misses" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Misses > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -9895,7 +10812,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Bandwidth" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -9983,7 +10908,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_IC_Misses > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_IC_Misses" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Info_Botlnk_IC_Misses > 5", + "ThresholdIssues": "$issueFL" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", @@ -10011,7 +10944,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMispredict" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts", @@ -10039,7 +10980,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -10067,7 +11016,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -10095,7 +11052,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Ret" + } + ], + "Formula": "a < 500", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -10123,7 +11088,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect" + } + ], + "Formula": "a < 1000", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -10133,7 +11106,7 @@ "MetricName": "Info_Bad_Spec_Branch_Misprediction_Cost", "LegacyName": "metric_TMA_Info_Bad_Spec_Branch_Misprediction_Cost", "Level": 1, - "BriefDescription": "Branch Misprediction Cost: Fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "UnitOfMeasure": "", "Events": [ { @@ -10230,6 +11203,11 @@ "BaseFormula": " tma_mispredictions * tma_info_thread_slots / ( 6 ) / br_misp_retired.all_branches / 100", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "$issueBM" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", "LocateWith": "" @@ -10259,6 +11237,11 @@ "BaseFormula": " int_misc.clears_count / ( br_misp_retired.all_branches + machine_clears.count )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", "LocateWith": "" @@ -10284,6 +11267,11 @@ "BaseFormula": " br_inst_retired.cond_ntaken / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", "LocateWith": "" @@ -10309,6 +11297,11 @@ "BaseFormula": " br_inst_retired.cond_taken / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", "LocateWith": "" @@ -10338,6 +11331,11 @@ "BaseFormula": " ( br_inst_retired.near_call + br_inst_retired.near_return ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -10371,6 +11369,11 @@ "BaseFormula": " ( br_inst_retired.near_taken - br_inst_retired.cond_taken - 2 * br_inst_retired.near_call ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -10412,6 +11415,11 @@ "BaseFormula": " 1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -10437,6 +11445,11 @@ "BaseFormula": " l1d_pend_miss.pending / mem_load_completed.l1_miss_any", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryLat", "LocateWith": "" @@ -10462,6 +11475,11 @@ "BaseFormula": " l1d_pend_miss.pending / l1d_pend_miss.pending_cycles", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryBW", "LocateWith": "" @@ -10487,6 +11505,11 @@ "BaseFormula": " 1000 * mem_load_retired.l1_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -10512,6 +11535,11 @@ "BaseFormula": " 1000 * l2_rqsts.all_demand_data_rd / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -10537,6 +11565,11 @@ "BaseFormula": " 1000 * mem_load_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;Backend;CacheHits", "LocateWith": "" @@ -10562,6 +11595,11 @@ "BaseFormula": " 1000 * l2_rqsts.miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem;Offcore", "LocateWith": "" @@ -10587,6 +11625,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -10612,6 +11655,11 @@ "BaseFormula": " 1000 * l2_rqsts.rfo_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheMisses;Offcore", "LocateWith": "" @@ -10641,6 +11689,11 @@ "BaseFormula": " 1000 * ( l2_rqsts.references - l2_rqsts.miss ) / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -10666,6 +11719,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -10691,6 +11749,11 @@ "BaseFormula": " 1000 * mem_load_retired.l3_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -10716,6 +11779,11 @@ "BaseFormula": " 1000 * mem_load_retired.fb_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -10742,6 +11810,11 @@ "BaseFormula": " 64 * l1d.replacement / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -10768,6 +11841,11 @@ "BaseFormula": " 64 * l2_lines_in.all / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -10794,6 +11872,11 @@ "BaseFormula": " 64 * longest_lat_cache.miss / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -10820,6 +11903,11 @@ "BaseFormula": " 64 * offcore_requests.all_requests / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW;Offcore", "LocateWith": "" @@ -10867,7 +11955,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Page_Walks_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -10894,6 +11990,11 @@ "BaseFormula": " 1000 * itlb_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;MemoryTLB", "LocateWith": "" @@ -10919,6 +12020,11 @@ "BaseFormula": " 1000 * dtlb_load_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -10944,6 +12050,11 @@ "BaseFormula": " 1000 * dtlb_store_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -10974,7 +12085,15 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Load_STLB_Miss_Ret > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Load_STLB_Miss_Ret" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_Memory_Load_STLB_Miss_Ret > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -11006,7 +12125,15 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Store_STLB_Miss_Ret > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Store_STLB_Miss_Ret" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_Memory_Store_STLB_Miss_Ret > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -11034,6 +12161,11 @@ "BaseFormula": " tma_info_memory_l1d_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -11060,6 +12192,11 @@ "BaseFormula": " tma_info_memory_l2_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -11086,6 +12223,11 @@ "BaseFormula": " tma_info_memory_l3_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -11112,6 +12254,11 @@ "BaseFormula": " tma_info_memory_l3_cache_access_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW;Offcore", "LocateWith": "" @@ -11142,7 +12289,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Useless_HWPF" + } + ], + "Formula": "a > 0.15", + "BaseFormula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -11169,6 +12324,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests.demand_data_rd", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Memory_Lat;Offcore", "LocateWith": "" @@ -11194,6 +12354,11 @@ "BaseFormula": " offcore_requests_outstanding.l3_miss_demand_data_rd / offcore_requests.l3_miss_demand_data_rd", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_Lat;Offcore", "LocateWith": "" @@ -11219,6 +12384,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests_outstanding.demand_data_rd:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -11244,6 +12414,11 @@ "BaseFormula": " offcore_requests_outstanding.data_rd / offcore_requests_outstanding.cycles_with_data_rd", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -11269,6 +12444,11 @@ "BaseFormula": " 1000 * mem_load_misc_retired.uc / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -11294,6 +12474,11 @@ "BaseFormula": " 1000 * sq_misc.bus_lock / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -11324,6 +12509,11 @@ "BaseFormula": " tma_info_system_cpus_utilized / num_cpus", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC;Summary", "LocateWith": "" @@ -11350,6 +12540,11 @@ "BaseFormula": " cpu_clk_unhalted.ref_tsc / tsc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", "LocateWith": "" @@ -11384,6 +12579,11 @@ "BaseFormula": " tma_info_system_turbo_utilization * tsc / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;Power", "LocateWith": "" @@ -11422,6 +12622,11 @@ "BaseFormula": " ( ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.256b_packed_single ) / ( 1000000000 ) ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -11447,6 +12652,11 @@ "BaseFormula": " tma_info_thread_clks / cpu_clk_unhalted.ref_tsc", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -11481,6 +12691,11 @@ "BaseFormula": " 1 - cpu_clk_unhalted.one_thread_active / cpu_clk_unhalted.ref_distributed if smt_on else 0", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "SMT", "LocateWith": "" @@ -11507,7 +12722,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Kernel_Utilization > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Kernel_Utilization" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_System_Kernel_Utilization > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "OS", @@ -11534,6 +12757,11 @@ "BaseFormula": " cpu_clk_unhalted.thread_p:sup / inst_retired.any_p:sup", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "OS", "LocateWith": "" @@ -11560,7 +12788,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_C0_Wait > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_C0_Wait" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_System_C0_Wait > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", @@ -11587,6 +12823,11 @@ "BaseFormula": " unc_arb_dat_occupancy.rd / unc_arb_dat_occupancy.rd:c1", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "ARB", "MetricGroup": "Mem;MemoryBW;SoC", "LocateWith": "" @@ -11612,33 +12853,14 @@ "Formula": "a * ( 61 ) / ( ( ( durationtimeinmilliseconds / 1000 ) ) * ( 1000000 ) )", "BaseFormula": " unc_pkg_energy_status * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "Category": "TMA", - "CountDomain": "SystemMetric", - "ResolutionLevels": "PKG", - "MetricGroup": "Power;SoC", - "LocateWith": "" - }, - { - "MetricName": "Info_System_Time", - "LegacyName": "metric_TMA_Info_System_Time", - "Level": 1, - "BriefDescription": "Run duration time in seconds", - "UnitOfMeasure": "", - "Events": [], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "( durationtimeinmilliseconds / 1000 )", - "BaseFormula": " duration_time", - "Category": "TMA", - "CountDomain": "Seconds", + "CountDomain": "System_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Time < 1" + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" }, - "ResolutionLevels": "SOCKET, SYSTEM", - "MetricGroup": "Summary", + "ResolutionLevels": "PKG", + "MetricGroup": "Power;SoC", "LocateWith": "" }, { @@ -11663,7 +12885,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_MUX" + } + ], + "Formula": "a > 1.1 | a < 0.9", + "BaseFormula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -11686,6 +12916,11 @@ "BaseFormula": " unc_clock.socket", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CLOCK", "MetricGroup": "SoC", "LocateWith": "" @@ -11712,7 +12947,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_IpFarBranch < 1000000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_IpFarBranch" + } + ], + "Formula": "a < 1000000", + "BaseFormula": "metric_TMA_Info_System_IpFarBranch < 1000000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;OS", diff --git a/MTL/metrics/perf/meteorlake_metrics_redwoodcove_core_perf.json b/MTL/metrics/perf/meteorlake_metrics_redwoodcove_core_perf.json index 567a2515..12820aa3 100644 --- a/MTL/metrics/perf/meteorlake_metrics_redwoodcove_core_perf.json +++ b/MTL/metrics/perf/meteorlake_metrics_redwoodcove_core_perf.json @@ -1,126 +1,145 @@ [ { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", + "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", + "MetricThreshold": "tma_bottleneck_mispredictions > 20", + "PublicDescription": "Total pipeline cost of Branch Misprediction related bottlenecks. Related metrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", + "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_big_code", + "MetricThreshold": "tma_bottleneck_big_code > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", + "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) ) ) - tma_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_instruction_fetch_bw", + "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) )", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", + "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_mem_bandwidth, tma_sq_full.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", + "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks. Related metrics: tma_l3_hit_latency, tma_mem_latency.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", + "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs)", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_load / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", + "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", + "PublicDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs). Related metrics: tma_dtlb_load, tma_dtlb_store.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", + "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors)", "MetricExpr": "100 * ( tma_memory_bound * ( ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", - "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", + "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", + "PublicDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", + "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation", "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", + "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", "PublicDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", + "BriefDescription": "Total pipeline cost of irregular execution (e.g", "MetricExpr": "100 * ( ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + RS.EMPTY_RESOURCE / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", - "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments).", + "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", + "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments). Related metrics: tma_microcode_sequencer, tma_ms_switches.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end.", + "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end", "MetricExpr": "100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", "MetricGroup": "BvOB;Cor;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_other_bottlenecks", + "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", "PublicDescription": "Total pipeline cost of remaining bottlenecks in the back-end. Examples include data-dependencies (Core Bound when Low ILP) and other unlisted memory-related stalls.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", + "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA", "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_branching_overhead", + "MetricThreshold": "tma_bottleneck_branching_overhead > 5", "PublicDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound).", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", + "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead", "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_useful_work", + "MetricThreshold": "tma_bottleneck_useful_work > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "topdown\\-fe\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_frontend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_frontend_bound > 0.15", "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound. Sample with: FRONTEND_RETIRED.LATENCY_GE_4.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( topdown\\-fetch\\-lat / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots )", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", "MetricName": "tma_fetch_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_16, FRONTEND_RETIRED.LATENCY_GE_8.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;Clocks", "MetricName": "tma_icache_misses", @@ -129,7 +148,7 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache", "MetricExpr": "max( 0 , ( FRONTEND_RETIRED.L1I_MISS * cpu_core@FRONTEND_RETIRED.L1I_MISS@R ) / tma_info_thread_clks - tma_code_l2_miss )", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_hit", @@ -137,7 +156,7 @@ "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache", "MetricExpr": "( FRONTEND_RETIRED.L2_MISS * cpu_core@FRONTEND_RETIRED.L2_MISS@R ) / tma_info_thread_clks", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_miss", @@ -145,43 +164,48 @@ "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , ( FRONTEND_RETIRED.ITLB_MISS * cpu_core@FRONTEND_RETIRED.ITLB_MISS@R ) / tma_info_thread_clks - tma_code_stlb_miss )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", "MetricExpr": "( FRONTEND_RETIRED.STLB_MISS * cpu_core@FRONTEND_RETIRED.STLB_MISS@R ) / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", @@ -190,25 +214,25 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "( tma_branch_mispredicts / tma_bad_speculation ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", "ScaleUnit": "100%", "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "( 1 - ( tma_branch_mispredicts / tma_bad_speculation ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", "ScaleUnit": "100%", "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_ms_switches." + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", @@ -217,43 +241,45 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 3 ) * cpu@UOPS_RETIRED.MS\\,cmask\\=0x1\\,edge\\=0x1@ / ( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: FRONTEND_RETIRED.MS_FLOWS. Related metrics: tma_clears_resteers, tma_l1_bound, tma_mixing_vectors." + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: FRONTEND_RETIRED.MS_FLOWS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "DECODE.LCP / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." + "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS_PS." + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "max( 0 , tma_frontend_bound - tma_fetch_latency )", - "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", + "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2.", + "MetricThreshold": "tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.MITE_CYCLES_ANY - IDQ.MITE_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;Slots_Estimated", "MetricName": "tma_mite", @@ -262,7 +288,7 @@ "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder.", + "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder", "MetricExpr": "( cpu@INST_DECODED.DECODERS\\,cmask\\=0x1@ - cpu@INST_DECODED.DECODERS\\,cmask\\=0x2@ ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated;tma_issueD0", "MetricName": "tma_decoder0_alone", @@ -271,7 +297,7 @@ "PublicDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder. Related metrics: tma_few_uops_instructions." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.DSB_CYCLES_ANY - IDQ.DSB_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;Slots_Estimated", "MetricName": "tma_dsb", @@ -280,77 +306,86 @@ "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit", "MetricExpr": "( LSD.CYCLES_ACTIVE - LSD.CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "FetchBW;LSD;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_lsd", "ScaleUnit": "100%", + "MetricThreshold": "tma_lsd > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit. LSD typically does well sustaining Uop supply. However; in some rare cases; optimal uop-delivery could not be reached for small loops whose size (in terms of number of uops) does not suit well the LSD structure." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details", "MetricExpr": "max( IDQ.MS_CYCLES_ANY , cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ / ( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) ) / tma_info_core_core_clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_ms", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "max( 1 - ( tma_frontend_bound + tma_backend_bound + tma_retiring ) , 0 )", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_bad_speculation", "ScaleUnit": "100%", + "MetricThreshold": "tma_bad_speculation > 0.15", "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( topdown\\-br\\-mispredict / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS.", + "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by non-taken conditional branches.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by non-taken conditional branches", "MetricExpr": "( BR_MISP_RETIRED.COND_NTAKEN_COST * cpu_core@BR_MISP_RETIRED.COND_NTAKEN_COST@R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_cond_nt_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_cond_nt_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to misprediction by taken conditional branches.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to misprediction by taken conditional branches", "MetricExpr": "( BR_MISP_RETIRED.COND_TAKEN_COST * cpu_core@BR_MISP_RETIRED.COND_TAKEN_COST@R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_cond_tk_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_cond_tk_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect CALL instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect CALL instructions", "MetricExpr": "( BR_MISP_RETIRED.INDIRECT_CALL_COST * cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_ind_call_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ind_call_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect JMP instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by indirect JMP instructions", "MetricExpr": "max( ( ( BR_MISP_RETIRED.INDIRECT_COST * cpu_core@BR_MISP_RETIRED.INDIRECT_COST@R ) - ( BR_MISP_RETIRED.INDIRECT_CALL_COST * cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@R ) ) / tma_info_thread_clks , 0 )", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_ind_jump_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ind_jump_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by (indirect) RET instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to retired misprediction by (indirect) RET instructions", "MetricExpr": "( BR_MISP_RETIRED.RET_COST * cpu_core@BR_MISP_RETIRED.RET_COST@R ) / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Clocks_Retired", "MetricName": "tma_ret_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ret_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", + "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types)", "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;Slots", "MetricName": "tma_other_mispredicts", @@ -358,17 +393,18 @@ "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "max( 0 , tma_bad_speculation - tma_branch_mispredicts )", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots", + "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT.", + "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering", "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;Slots", "MetricName": "tma_other_nukes", @@ -376,17 +412,18 @@ "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "( topdown\\-be\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_backend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_backend_bound > 0.2", "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( topdown\\-mem\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;Default;Slots", "MetricName": "tma_memory_bound", @@ -397,25 +434,25 @@ "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( EXE_ACTIVITY.BOUND_ON_LOADS - MEMORY_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_ms_switches, tma_ports_utilized_1." + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "( min( ( MEM_INST_RETIRED.STLB_HIT_LOADS * cpu_core@MEM_INST_RETIRED.STLB_HIT_LOADS@R ) , MEM_INST_RETIRED.STLB_HIT_LOADS * ( 7 ) ) if ( cpu_core@MEM_INST_RETIRED.STLB_HIT_LOADS@R >= 0 ) else ( MEM_INST_RETIRED.STLB_HIT_LOADS * ( 7 ) ) ) / tma_info_thread_clks + tma_load_stlb_miss", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: tma_dtlb_store." + "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_dtlb_load - tma_load_stlb_miss )", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", "MetricName": "tma_load_stlb_hit", @@ -423,7 +460,7 @@ "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", "MetricName": "tma_load_stlb_miss", @@ -431,7 +468,7 @@ "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_4k", @@ -439,7 +476,7 @@ "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_2m", @@ -447,7 +484,7 @@ "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_1g", @@ -455,7 +492,7 @@ "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", @@ -464,7 +501,7 @@ "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", + "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache", "MetricExpr": "min( 2 * ( MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETIRED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS ) * 20 / 100 , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - MEMORY_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_l1_latency_dependency", @@ -473,7 +510,7 @@ "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( MEM_INST_RETIRED.LOCK_LOADS * cpu_core@MEM_INST_RETIRED.LOCK_LOADS@R ) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", @@ -482,34 +519,34 @@ "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "( min( ( MEM_INST_RETIRED.SPLIT_LOADS * cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@R ) , MEM_INST_RETIRED.SPLIT_LOADS * tma_info_memory_load_miss_real_latency ) if ( cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@R >= 0 ) else ( MEM_INST_RETIRED.SPLIT_LOADS * tma_info_memory_load_miss_real_latency ) ) / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;Clocks_Calculated", "MetricName": "tma_split_loads", "ScaleUnit": "100%", - "MetricThreshold": "tma_split_loads > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS." + "MetricThreshold": "tma_split_loads > 0.3", + "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "L1D_PEND_MISS.FB_FULL / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", "MetricThreshold": "tma_fb_full > 0.3", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores." + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L1D_MISS - MEMORY_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", "ScaleUnit": "100%", "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT_PS." + "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", + "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited)", "MetricExpr": "( min( ( MEM_LOAD_RETIRED.L2_HIT * cpu_core@MEM_LOAD_RETIRED.L2_HIT@R ) , MEM_LOAD_RETIRED.L2_HIT * ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_RETIRED.L2_HIT@R >= 0 ) else ( MEM_LOAD_RETIRED.L2_HIT * ( 3 * tma_info_system_core_frequency ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", "MetricName": "tma_l2_hit_latency", @@ -518,7 +555,7 @@ "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L2_MISS - MEMORY_ACTIVITY.STALLS_L3_MISS ) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", @@ -527,70 +564,70 @@ "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) + ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 28 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 28 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) * ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;Offcore;Snoop;TopdownL4;tma_L4_group;Clocks_Estimated;tma_issueSyncxn", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_data_sharing, tma_false_sharing." + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) + ( min( ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@R ) , MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@R >= 0 ) else ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 27 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) * ( 1 - ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_contested_accesses, tma_false_sharing." + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( min( ( MEM_LOAD_RETIRED.L3_HIT * cpu_core@MEM_LOAD_RETIRED.L3_HIT@R ) , MEM_LOAD_RETIRED.L3_HIT * ( 12 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) if ( cpu_core@MEM_LOAD_RETIRED.L3_HIT@R >= 0 ) else ( MEM_LOAD_RETIRED.L3_HIT * ( 12 * tma_info_system_core_frequency ) - ( 3 * tma_info_system_core_frequency ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tma_branch_resteers, tma_mem_latency, tma_store_latency." + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_bottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "( XQ.FULL_CYCLES + L1D_PEND_MISS.L2_STALLS ) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", "ScaleUnit": "100%", "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_fb_full, tma_mem_bandwidth." + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clks )", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", "ScaleUnit": "100%", "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS_PS." + "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_fb_full, tma_sq_full." + "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_l3_hit_latency." + "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", @@ -599,34 +636,34 @@ "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( MEM_STORE_RETIRED.L2_HIT * ( 10 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( 28 * tma_info_system_core_frequency ) * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related metrics: tma_contested_accesses, tma_data_sharing." + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "( min( ( MEM_INST_RETIRED.SPLIT_STORES * cpu_core@MEM_INST_RETIRED.SPLIT_STORES@R ) , MEM_INST_RETIRED.SPLIT_STORES * 1 ) if ( cpu_core@MEM_INST_RETIRED.SPLIT_STORES@R >= 0 ) else ( MEM_INST_RETIRED.SPLIT_STORES * 1 ) ) / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", "ScaleUnit": "100%", "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS." + "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores.", + "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores", "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread_clks", "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSmSt", "MetricName": "tma_streaming_stores", @@ -635,16 +672,16 @@ "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma_fb_full." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( min( ( MEM_INST_RETIRED.STLB_HIT_STORES * cpu_core@MEM_INST_RETIRED.STLB_HIT_STORES@R ) , MEM_INST_RETIRED.STLB_HIT_STORES * ( 7 ) ) if ( cpu_core@MEM_INST_RETIRED.STLB_HIT_STORES@R >= 0 ) else ( MEM_INST_RETIRED.STLB_HIT_STORES * ( 7 ) ) ) / tma_info_thread_clks + tma_store_stlb_miss", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_dtlb_load." + "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_load." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_dtlb_store - tma_store_stlb_miss )", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", "MetricName": "tma_store_stlb_hit", @@ -652,7 +689,7 @@ "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", "MetricName": "tma_store_stlb_miss", @@ -660,7 +697,7 @@ "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_4k", @@ -668,7 +705,7 @@ "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_2m", @@ -676,7 +713,7 @@ "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_1g", @@ -684,17 +721,18 @@ "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "max( 0 , tma_backend_bound - tma_memory_bound )", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", "MetricName": "tma_core_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations).", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.DIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;Clocks", "MetricName": "tma_divider", @@ -703,7 +741,7 @@ "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIV_ACTIVE." }, { - "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active", "MetricExpr": "ARITH.FPDIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_fp_divider", @@ -711,7 +749,7 @@ "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_int_divider", @@ -719,15 +757,16 @@ "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations", "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks + tma_c02_wait", - "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", + "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks;tma_issueSO", "MetricName": "tma_serializing_operation", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD." + "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions", "MetricExpr": "CPU_CLK_UNHALTED.PAUSE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;Clocks", "MetricName": "tma_slow_pause", @@ -736,21 +775,31 @@ "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.PAUSE_INST." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings).", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings)", "MetricExpr": "CPU_CLK_UNHALTED.C01 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_c01_wait", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_c01_wait > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" + }, + { + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings)", + "MetricExpr": "CPU_CLK_UNHALTED.C02 / tma_info_thread_clks", + "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;Clocks", + "MetricName": "tma_c02_wait", + "ScaleUnit": "100%", + "MetricThreshold": "tma_c02_wait > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions", "MetricExpr": "13 * MISC2_RETIRED.LFENCE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_memory_fence", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_fence > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( tma_ports_utilized_0 * tma_info_thread_clks + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIV_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - EXE_ACTIVITY.BOUND_ON_LOADS ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;Clocks", "MetricName": "tma_ports_utilization", @@ -759,7 +808,7 @@ "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "max( EXE_ACTIVITY.EXE_BOUND_0_PORTS - RESOURCE_STALLS.SCOREBOARD , 0 ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", @@ -768,7 +817,7 @@ "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", + "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles)", "MetricExpr": "160 * ASSISTS.SSE_AVX_MIX / tma_info_thread_clks", "MetricGroup": "TopdownL5;tma_L5_group;Clocks;tma_issueMV", "MetricName": "tma_mixing_vectors", @@ -777,7 +826,7 @@ "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", @@ -786,7 +835,7 @@ "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", @@ -795,7 +844,7 @@ "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", @@ -804,7 +853,7 @@ "PublicDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Sample with: UOPS_EXECUTED.CYCLES_GE_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + UOPS_DISPATCHED.PORT_5_11 + UOPS_DISPATCHED.PORT_6 ) / ( 5 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;Core_Execution", "MetricName": "tma_alu_op_utilization", @@ -812,7 +861,7 @@ "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED.PORT_0 / tma_info_core_core_clks", "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", @@ -821,7 +870,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_1 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", @@ -830,7 +879,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_6 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", @@ -839,7 +888,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "UOPS_DISPATCHED.PORT_2_3_10 / ( 3 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;Core_Execution", "MetricName": "tma_load_op_utilization", @@ -848,7 +897,7 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations. Sample with: UOPS_DISPATCHED.PORT_2_3_10." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_4_9 + UOPS_DISPATCHED.PORT_7_8 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", @@ -857,27 +906,29 @@ "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations. Sample with: UOPS_DISPATCHED.PORT_7_8." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( topdown\\-retiring / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_retiring", "ScaleUnit": "100%", + "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.1", "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "max( 0 , tma_retiring - tma_heavy_operations )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", "MetricName": "tma_light_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;Uops", "MetricName": "tma_fp_arith", @@ -886,7 +937,7 @@ "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", @@ -895,7 +946,7 @@ "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", @@ -904,7 +955,7 @@ "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", @@ -913,16 +964,16 @@ "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", @@ -931,7 +982,7 @@ "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired)", "MetricExpr": "tma_int_vector_128b + tma_int_vector_256b", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;Uops", "MetricName": "tma_int_operations", @@ -940,7 +991,7 @@ "PublicDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired). Vector/Matrix Int operations and shuffles are counted. Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain." }, { - "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", + "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired", "MetricExpr": "( INT_VEC_RETIRED.ADD_128 + INT_VEC_RETIRED.VNNI_128 ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops;tma_issue2P", "MetricName": "tma_int_vector_128b", @@ -949,7 +1000,7 @@ "PublicDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", + "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired", "MetricExpr": "( INT_VEC_RETIRED.ADD_256 + INT_VEC_RETIRED.MUL_256 + INT_VEC_RETIRED.VNNI_256 ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops;tma_issue2P", "MetricName": "tma_int_vector_256b", @@ -958,14 +1009,15 @@ "PublicDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses", "MetricExpr": "tma_light_operations * MEM_UOP_RETIRED.ANY / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_memory_operations", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_operations > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.MACRO_FUSED / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;Slots", "MetricName": "tma_fused_instructions", @@ -974,15 +1026,16 @@ "PublicDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Other_Light_Ops in MTL!)}." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused", "MetricExpr": "tma_light_operations * ( BR_INST_RETIRED.ALL_BRANCHES - INST_RETIRED.MACRO_FUSED ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_non_fused_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused. Non-conditional branches like direct JMP or CALL would count here. Can be used to examine fusible conditional jumps that were not fused." }, { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", + "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes", "MetricExpr": "max( 0 , tma_light_operations - ( tma_fp_arith + tma_int_operations + tma_memory_operations + tma_fused_instructions + tma_non_fused_branches ) )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;Slots", "MetricName": "tma_other_light_ops", @@ -991,7 +1044,7 @@ "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_light_ops_group;Slots", "MetricName": "tma_nop_instructions", @@ -1000,7 +1053,7 @@ "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer)", "MetricExpr": "tma_light_operations * INT_VEC_RETIRED.SHUFFLES / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_other_light_ops_group;Slots", "MetricName": "tma_shuffles_256b", @@ -1009,34 +1062,36 @@ "PublicDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer). Shuffles may incur slow cross \"vector lane\" data transfers." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "( topdown\\-heavy\\-ops / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", "MetricName": "tma_heavy_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+]). Sample with: UOPS_RETIRED.HEAVY.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or up to ([SNB+] four; [ADL+] five) uops.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops", "MetricExpr": "max( 0 , tma_heavy_operations - tma_microcode_sequencer )", "MetricGroup": "TopdownL3;tma_L3_group;Slots;tma_issueD0", "MetricName": "tma_few_uops_instructions", "ScaleUnit": "100%", "MetricThreshold": "tma_few_uops_instructions > 0.05 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or up to ([SNB+] four; [ADL+] five) uops. This highly-correlates with the number of uops in such instructions. Related metrics: tma_decoder0_alone." + "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions. Related metrics: tma_decoder0_alone." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "UOPS_RETIRED.MS / tma_info_thread_slots", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: UOPS_RETIRED.MS." + "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: UOPS_RETIRED.MS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( ( 99 *3 + 63 + 30 ) / 5 ) * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;Slots_Estimated", "MetricName": "tma_assists", @@ -1045,7 +1100,7 @@ "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults", "MetricExpr": "99 * ASSISTS.PAGE_FAULT / tma_info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_page_faults", @@ -1054,7 +1109,7 @@ "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults. A Page Fault may apply on first application access to a memory page. Note operating system handling of page faults accounts for the majority of its cost." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists", "MetricExpr": "30 * ASSISTS.FP / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_fp_assists", @@ -1063,7 +1118,7 @@ "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists", "MetricExpr": "63 * ASSISTS.SSE_AVX_MIX / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_avx_assists", @@ -1071,7 +1126,7 @@ "MetricThreshold": "tma_avx_assists > 0.1" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;Slots", "MetricName": "tma_cisc", @@ -1080,702 +1135,746 @@ "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources. Sample with: FRONTEND_RETIRED.MS_FLOWS." }, { - "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts.", + "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts", "MetricExpr": "100 * ( 1 - tma_core_bound / tma_ports_utilization if tma_core_bound < tma_ports_utilization else 1 ) if tma_info_system_smt_2t_utilization > 0.5 else 0", "MetricGroup": "Cor;SMT;Metric", - "MetricName": "tma_info_botlnk_l0_core_bound_likely" + "MetricName": "tma_info_botlnk_l0_core_bound_likely", + "MetricThreshold": "tma_info_botlnk_core_bound_likely > 0.5" }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", - "MetricName": "tma_info_thread_uoppi" + "MetricName": "tma_info_thread_uoppi", + "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", - "MetricName": "tma_info_thread_uptb" + "MetricName": "tma_info_thread_uptb", + "MetricThreshold": "tma_info_thread_uptb < 6 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor.", + "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor", "MetricExpr": "tma_info_thread_slots / ( TOPDOWN.SLOTS / 2 ) if #SMT_on else 1", "MetricGroup": "SMT;TmaL1;TopdownL1;tma_L1_group;Metric", "MetricName": "tma_info_thread_slots_utilization" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / tma_info_core_core_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( FP_ARITH_DISPATCHED.PORT_0 + FP_ARITH_DISPATCHED.PORT_1 + FP_ARITH_DISPATCHED.PORT_5 ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "uops Executed per Cycle.", + "BriefDescription": "uops Executed per Cycle", "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", "MetricGroup": "Power;Metric", "MetricName": "tma_info_core_epc" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipload" + "MetricName": "tma_info_inst_mix_ipload", + "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipstore" + "MetricName": "tma_info_inst_mix_ipstore", + "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipbranch" + "MetricName": "tma_info_inst_mix_ipbranch", + "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipcall" + "MetricName": "tma_info_inst_mix_ipcall", + "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", - "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_iptb" + "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", + "MetricName": "tma_info_inst_mix_iptb", + "MetricThreshold": "tma_info_inst_mix_iptb < 6 * 2 + 1", + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipflop" + "MetricName": "tma_info_inst_mix_ipflop", + "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", + "MetricThreshold": "tma_info_inst_mix_iparith < 10", "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", + "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", + "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / CPU_CLK_UNHALTED.PAUSE_INST", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ippause" }, { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / SW_PREFETCH_ACCESS.ANY", "MetricGroup": "Prefetches;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipswpf" + "MetricName": "tma_info_inst_mix_ipswpf", + "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions.", + "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions", "MetricExpr": "INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "MicroSeq;Pipeline;Ret;Metric", - "MetricName": "tma_info_pipeline_strings_cycles" + "MetricName": "tma_info_pipeline_strings_cycles", + "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1" }, { - "BriefDescription": "Instructions per a microcode Assist invocation.", + "BriefDescription": "Instructions per a microcode Assist invocation", "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", "MetricName": "tma_info_pipeline_ipassist", + "MetricThreshold": "tma_info_pipeline_ipassist < 100000", "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core.", + "BriefDescription": "", "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@ )", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Average number of uops fetched from LSD per cycle.", + "BriefDescription": "Average number of uops fetched from LSD per cycle", "MetricExpr": "LSD.UOPS / LSD.CYCLES_ACTIVE", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_lsd" }, { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", + "BriefDescription": "Average number of uops fetched from DSB per cycle", "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_dsb" }, { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", + "BriefDescription": "Average number of uops fetched from MITE per cycle", "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_mite" }, { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", + "BriefDescription": "Average number of Uops issued by front-end when it issued something", "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_frontend_fetch_upc" }, { - "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache).", + "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache)", "MetricExpr": "LSD.UOPS / ( UOPS_ISSUED.ANY )", "MetricGroup": "Fed;LSD;Metric", "MetricName": "tma_info_frontend_lsd_coverage" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( UOPS_ISSUED.ANY )", - "MetricGroup": "DSB;Fed;FetchBW;Metric", - "MetricName": "tma_info_frontend_dsb_coverage" + "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", + "MetricName": "tma_info_frontend_dsb_coverage", + "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 6 > 0.35", + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection.", + "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection", "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / cpu@INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_unknown_branch_cost", "PublicDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection. See Unknown_Branches node." }, { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", + "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "DSBmiss;Metric", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired DSB misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired DSB misses", "MetricExpr": "( FRONTEND_RETIRED.ANY_DSB_MISS * cpu_core@FRONTEND_RETIRED.ANY_DSB_MISS@R ) / tma_info_thread_clks", "MetricGroup": "DSBmiss;Fed;FetchLat;Clocks_Retired", - "MetricName": "tma_info_frontend_dsb_switches_ret" + "MetricName": "tma_info_frontend_dsb_switches_ret", + "MetricThreshold": "tma_info_frontend_dsb_switches_ret > 0.05" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired operations that invoke the Microcode Sequencer.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired operations that invoke the Microcode Sequencer", "MetricExpr": "( FRONTEND_RETIRED.MS_FLOWS * cpu_core@FRONTEND_RETIRED.MS_FLOWS@R ) / tma_info_thread_clks", "MetricGroup": "Fed;FetchLat;MicroSeq;Clocks_Retired", - "MetricName": "tma_info_frontend_ms_latency_ret" + "MetricName": "tma_info_frontend_ms_latency_ret", + "MetricThreshold": "tma_info_frontend_ms_latency_ret > 0.05" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired branches who got branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to retired branches who got branch address clears", "MetricExpr": "( FRONTEND_RETIRED.UNKNOWN_BRANCH * cpu_core@FRONTEND_RETIRED.UNKNOWN_BRANCH@R ) / tma_info_thread_clks", "MetricGroup": "Fed;FetchLat;Clocks_Retired", "MetricName": "tma_info_frontend_unknown_branches_ret" }, { - "BriefDescription": "Average Latency for L1 instruction cache misses.", + "BriefDescription": "Average Latency for L1 instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / ICACHE_DATA.STALL_PERIODS", "MetricGroup": "Fed;FetchLat;IcMiss;Metric", "MetricName": "tma_info_frontend_icache_miss_latency" }, { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", "MetricGroup": "DSBmiss;Fed;Inst_Metric", - "MetricName": "tma_info_frontend_ipdsb_miss_ret" + "MetricName": "tma_info_frontend_ipdsb_miss_ret", + "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache true code cacheline misses per kilo instruction", "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code" }, { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction", "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code_all" }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) )", - "MetricGroup": "DSBmiss;Fed;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_misses" + "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_misses", + "MetricThreshold": "tma_info_botlnk_dsb_misses > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) )", - "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_bandwidth" + "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", + "MetricThreshold": "tma_info_botlnk_dsb_bandwidth > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", + "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_ic_misses" + "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", + "MetricName": "tma_info_botlnk_l2_ic_misses", + "MetricThreshold": "tma_info_botlnk_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmispredict" + "MetricName": "tma_info_bad_spec_ipmispredict", + "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken" + "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", + "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_cond_taken" + "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", + "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_ret" + "MetricName": "tma_info_bad_spec_ipmisp_ret", + "MetricThreshold": "tma_info_bad_spec_ipmisp_ret < 500" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_indirect" + "MetricName": "tma_info_bad_spec_ipmisp_indirect", + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Branch Misprediction Cost: Fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 6 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", - "MetricGroup": "Bad;BrMispredicts;Core_Metric", - "MetricName": "tma_info_bad_spec_branch_misprediction_cost" + "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", + "MetricName": "tma_info_bad_spec_branch_misprediction_cost", + "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_mispredicts_resteers." }, { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", + "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", "MetricGroup": "BrMispredicts;Metric", "MetricName": "tma_info_bad_spec_spec_clears_ratio" }, { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", + "BriefDescription": "Fraction of branches that are non-taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_nt" }, { - "BriefDescription": "Fraction of branches that are taken conditionals.", + "BriefDescription": "Fraction of branches that are taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_tk" }, { - "BriefDescription": "Fraction of branches that are CALL or RET.", + "BriefDescription": "Fraction of branches that are CALL or RET", "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_callret" }, { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", + "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_jump" }, { - "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group).", + "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group)", "MetricExpr": "1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_other_branches" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / MEM_LOAD_COMPLETED.L1_MISS_ANY", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki_load" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * L2_RQSTS.RFO_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_all" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", + "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_fb_hpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Offcore;Metric", "MetricName": "tma_info_memory_l3_cache_access_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", - "MetricName": "tma_info_memory_tlb_page_walks_utilization" + "MetricName": "tma_info_memory_tlb_page_walks_utilization", + "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Fed;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_code_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_load_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand loads.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand loads", "MetricExpr": "( MEM_INST_RETIRED.STLB_MISS_LOADS * cpu_core@MEM_INST_RETIRED.STLB_MISS_LOADS@R ) / tma_info_thread_clks", "MetricGroup": "Mem;MemoryTLB;Clocks_Retired", - "MetricName": "tma_info_memory_tlb_load_stlb_miss_ret" + "MetricName": "tma_info_memory_tlb_load_stlb_miss_ret", + "MetricThreshold": "tma_info_memory_load_stlb_miss_ret > 0.05" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand stores.", + "BriefDescription": "This metric represents fraction of cycles the CPU retirement was stalled likely due to STLB misses by demand stores", "MetricExpr": "( MEM_INST_RETIRED.STLB_MISS_STORES * cpu_core@MEM_INST_RETIRED.STLB_MISS_STORES@R ) / tma_info_thread_clks", "MetricGroup": "Mem;MemoryTLB;Clocks_Retired", - "MetricName": "tma_info_memory_tlb_store_stlb_miss_ret" + "MetricName": "tma_info_memory_tlb_store_stlb_miss_ret", + "MetricThreshold": "tma_info_memory_store_stlb_miss_ret > 0.05" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_access_bw", "MetricGroup": "Mem;MemoryBW;Offcore;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_access_bw_2t" }, { - "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses.", + "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses", "MetricExpr": "L2_LINES_OUT.USELESS_HWPF / ( L2_LINES_OUT.SILENT + L2_LINES_OUT.NON_SILENT )", "MetricGroup": "Prefetches;Metric", - "MetricName": "tma_info_memory_prefetches_useless_hwpf" + "MetricName": "tma_info_memory_prefetches_useless_hwpf", + "MetricThreshold": "tma_info_memory_useless_hwpf > 0.15" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Latency for L3 cache miss demand Loads.", + "BriefDescription": "Average Latency for L3 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.L3_MISS_DEMAND_DATA_RD / OFFCORE_REQUESTS.L3_MISS_DEMAND_DATA_RD", "MetricGroup": "Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l3_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=0x1@", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", + "BriefDescription": "Un-cacheable retired load per kilo instruction", "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_uc_load_pki" }, { - "BriefDescription": "\"Bus lock\" per kilo instruction.", + "BriefDescription": "\"Bus lock\" per kilo instruction", "MetricExpr": "1000 * SQ_MISC.BUS_LOCK / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_bus_lock_pki" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", - "MetricName": "tma_info_system_kernel_utilization" + "MetricName": "tma_info_system_kernel_utilization", + "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states.", + "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states", "MetricExpr": "CPU_CLK_UNHALTED.C0_WAIT / tma_info_thread_clks", "MetricGroup": "C0Wait;Metric", - "MetricName": "tma_info_system_c0_wait" + "MetricName": "tma_info_system_c0_wait", + "MetricThreshold": "tma_info_system_c0_wait > 0.05" }, { - "BriefDescription": "Average number of parallel data read requests to external memory.", + "BriefDescription": "Average number of parallel data read requests to external memory", "MetricExpr": "UNC_ARB_DAT_OCCUPANCY.RD / cpu@UNC_ARB_DAT_OCCUPANCY.RD\\,cmask\\=0x1@", "MetricGroup": "Mem;MemoryBW;SoC;System_Metric", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches." }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "power@energy\\-pkg@ * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", - "MetricGroup": "Power;SoC;SystemMetric", + "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", - "MetricName": "tma_info_system_mux" + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "UNC_CLOCK.SOCKET", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", - "MetricName": "tma_info_system_ipfarbranch" + "MetricName": "tma_info_system_ipfarbranch", + "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" } ] \ No newline at end of file diff --git a/RKL/metrics/perf/rocketlake_metrics_perf.json b/RKL/metrics/perf/rocketlake_metrics_perf.json index 1f34d884..02c95bae 100644 --- a/RKL/metrics/perf/rocketlake_metrics_perf.json +++ b/RKL/metrics/perf/rocketlake_metrics_perf.json @@ -1,1586 +1,1750 @@ [ { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", + "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", + "MetricThreshold": "tma_bottleneck_mispredictions > 20", + "PublicDescription": "Total pipeline cost of Branch Misprediction related bottlenecks. Related metrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", + "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_big_code", + "MetricThreshold": "tma_bottleneck_big_code > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", + "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_mispredicts_resteers ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) ) ) - tma_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_instruction_fetch_bw", + "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) )", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", + "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", + "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks. Related metrics: tma_l3_hit_latency, tma_mem_latency.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", + "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs)", "MetricExpr": "100 * ( tma_memory_bound * ( tma_l1_bound / max( tma_memory_bound , ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) ) * ( tma_dtlb_load / max( tma_l1_bound , ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", + "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", + "PublicDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs). Related metrics: tma_dtlb_load, tma_dtlb_store.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", + "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors)", "MetricExpr": "100 * ( tma_memory_bound * ( ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", - "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", + "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", + "PublicDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", + "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation", "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", + "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", "PublicDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", + "BriefDescription": "Total pipeline cost of irregular execution (e.g", "MetricExpr": "100 * ( ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_mispredicts_resteers ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + tma_core_bound * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", - "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments).", + "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", + "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments). Related metrics: tma_microcode_sequencer, tma_ms_switches.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end.", + "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end", "MetricExpr": "100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", "MetricGroup": "BvOB;Cor;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_other_bottlenecks", + "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", "PublicDescription": "Total pipeline cost of remaining bottlenecks in the back-end. Examples include data-dependencies (Core Bound when Low ILP) and other unlisted memory-related stalls.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", + "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA", "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_branching_overhead", + "MetricThreshold": "tma_bottleneck_branching_overhead > 5", "PublicDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound).", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", + "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead", "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_useful_work", + "MetricThreshold": "tma_bottleneck_useful_work > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "topdown\\-fe\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_frontend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_frontend_bound > 0.15", "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound. Sample with: FRONTEND_RETIRED.LATENCY_GE_4.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( ( 5 ) * IDQ_UOPS_NOT_DELIVERED.CYCLES_0_UOPS_DELIV.CORE - INT_MISC.UOP_DROPPING ) / tma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", "MetricName": "tma_fetch_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_16, FRONTEND_RETIRED.LATENCY_GE_8.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache", "MetricExpr": "max( 0 , tma_icache_misses - tma_code_l2_miss )", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_CODE_RD / tma_info_thread_clks", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_itlb_misses - tma_code_stlb_miss )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + ( 10 ) * BACLEARS.ANY / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "( 1 - ( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "( 10 ) * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: BACLEARS.ANY." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 3 ) * IDQ.MS_SWITCHES / tma_info_thread_clks", - "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated", + "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES." + "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "DECODE.LCP / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." + "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", - "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS." + "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "max( 0 , tma_frontend_bound - tma_fetch_latency )", - "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", + "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2.", + "MetricThreshold": "tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.MITE_CYCLES_ANY - IDQ.MITE_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", "ScaleUnit": "100%", + "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder.", + "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder", "MetricExpr": "( cpu@INST_DECODED.DECODERS\\,cmask\\=0x1@ - cpu@INST_DECODED.DECODERS\\,cmask\\=0x2@ ) / tma_info_core_core_clks / 2", - "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated", + "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated;tma_issueD0", "MetricName": "tma_decoder0_alone", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder. Related metrics: tma_few_uops_instructions." }, { - "BriefDescription": "This metric represents fraction of cycles where (only) 4 uops were delivered by the MITE pipeline.", + "BriefDescription": "This metric represents fraction of cycles where (only) 4 uops were delivered by the MITE pipeline", "MetricExpr": "( cpu@IDQ.MITE_UOPS\\,cmask\\=0x4@ - cpu@IDQ.MITE_UOPS\\,cmask\\=0x5@ ) / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Core_Clocks", "MetricName": "tma_mite_4wide", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_mite_4wide > 0.05 & tma_mite > 0.1 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.DSB_CYCLES_ANY - IDQ.DSB_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", "ScaleUnit": "100%", + "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit", "MetricExpr": "( LSD.CYCLES_ACTIVE - LSD.CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "FetchBW;LSD;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_lsd", "ScaleUnit": "100%", + "MetricThreshold": "tma_lsd > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit. LSD typically does well sustaining Uop supply. However; in some rare cases; optimal uop-delivery could not be reached for small loops whose size (in terms of number of uops) does not suit well the LSD structure." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details", "MetricExpr": "cpu@IDQ.MS_UOPS\\,cmask\\=0x1@ / tma_info_core_core_clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_ms", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "max( 1 - ( tma_frontend_bound + tma_backend_bound + tma_retiring ) , 0 )", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_bad_speculation", "ScaleUnit": "100%", + "MetricThreshold": "tma_bad_speculation > 0.15", "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * tma_bad_speculation", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES.", + "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", + "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types)", "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Slots", "MetricName": "tma_other_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "max( 0 , tma_bad_speculation - tma_branch_mispredicts )", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", + "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT.", + "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering", "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_machine_clears_group;Slots", "MetricName": "tma_other_nukes", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "topdown\\-be\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) + ( ( 5 ) * INT_MISC.CLEARS_COUNT ) / tma_info_thread_slots", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_backend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_backend_bound > 0.2", "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( ( CYCLE_ACTIVITY.STALLS_MEM_ANY + EXE_ACTIVITY.BOUND_ON_STORES ) / ( CYCLE_ACTIVITY.STALLS_TOTAL + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) + EXE_ACTIVITY.BOUND_ON_STORES ) ) * tma_backend_bound", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_memory_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two).", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", + "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT." + "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "min( ( 7 ) * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS." + "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", "MetricName": "tma_load_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", "MetricName": "tma_load_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", + "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache", "MetricExpr": "min( 2 * ( MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETIRED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS ) * 20 / 100 , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_l1_latency_dependency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( 16 * max( 0 , MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS.ALL_RFO ) + ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) * ( ( 10 ) * L2_RQSTS.RFO_HIT + ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) ) / tma_info_thread_clks", - "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks", + "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS." + "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset.", + "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset", "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_4k_aliasing", "ScaleUnit": "100%", + "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset. False match is possible; which incur a few cycles load re-issue. However; the short re-issue duration is often hidden by the out-of-order core and HW optimizations; hence a user may safely ignore a high value of this metric unless it manages to propagate up into parent nodes of the hierarchy (e.g. to L1_Bound)." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "L1D_PEND_MISS.FB_FULL / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", + "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory)." + "MetricThreshold": "tma_fb_full > 0.3", + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( ( MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) ) ) / ( ( MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) ) ) + L1D_PEND_MISS.FB_FULL_PERIODS ) ) * ( ( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks )", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", + "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited)", "MetricExpr": "( 3.5 * tma_info_system_core_frequency ) * MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", "MetricName": "tma_l2_hit_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L2_MISS - CYCLE_ACTIVITY.STALLS_L3_MISS ) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( ( 32.5 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM ) + ( ( 27 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS." + "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( ( 27 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT." + "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( ( 12.5 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_RETIRED.L3_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT." + "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_bottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "L1D_PEND_MISS.L2_STALL / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clks + ( ( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks ) - tma_l2_bound )", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that)." + "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that)." + "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( L2_RQSTS.RFO_HIT * ( 10 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)." + "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( 32.5 * tma_info_system_core_frequency ) * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM." + "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization", + "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores.", + "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores", "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread_clks", - "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSmSt", "MetricName": "tma_streaming_stores", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE." + "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma_fb_full." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( ( 7 ) * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_STORE_MISSES.WALK_ACTIVE ) / tma_info_core_core_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES." + "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_load." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", "MetricName": "tma_store_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", "MetricName": "tma_store_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "max( 0 , tma_backend_bound - tma_memory_bound )", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_core_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations).", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", "ScaleUnit": "100%", + "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIVIDER_ACTIVE." }, { - "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active", "MetricExpr": "ARITH.FP_DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_fp_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_int_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations", "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks", - "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Clocks;tma_issueSO", "MetricName": "tma_serializing_operation", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD." + "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions", "MetricExpr": "140 * MISC_RETIRED.PAUSE_INST / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_slow_pause", "ScaleUnit": "100%", + "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions. Sample with: MISC_RETIRED.PAUSE_INST." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( tma_ports_utilized_0 * tma_info_thread_clks + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIVIDER_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - CYCLE_ACTIVITY.STALLS_MEM_ANY ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "cpu@EXE_ACTIVITY.3_PORTS_UTIL\\,umask\\=0x80@ / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", + "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles)", "MetricExpr": "UOPS_ISSUED.VECTOR_WIDTH_MISMATCH / UOPS_ISSUED.ANY", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks;tma_issueMV", "MetricName": "tma_mixing_vectors", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic." + "MetricThreshold": "tma_mixing_vectors > 0.05", + "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Sample with: UOPS_EXECUTED.CYCLES_GE_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + UOPS_DISPATCHED.PORT_5 + UOPS_DISPATCHED.PORT_6 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED.PORT_0 / tma_info_core_core_clks", - "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0." + "MetricThreshold": "tma_port_0 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_1 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1." + "MetricThreshold": "tma_port_1 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_5 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_5", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED.PORT_5." + "MetricThreshold": "tma_port_5 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED.PORT_5. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_6 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1." + "MetricThreshold": "tma_port_6 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "UOPS_DISPATCHED.PORT_2_3 / ( 2 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_load_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations. Sample with: UOPS_DISPATCHED.PORT_2_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_4_9 + UOPS_DISPATCHED.PORT_7_8 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations. Sample with: UOPS_DISPATCHED.PORT_7_8." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( topdown\\-retiring / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_retiring", "ScaleUnit": "100%", + "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.1", "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "max( 0 , tma_retiring - tma_heavy_operations )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_light_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", "ScaleUnit": "100%", + "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_512b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses", "MetricExpr": "tma_light_operations * MEM_INST_RETIRED.ANY / INST_RETIRED.ANY", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_memory_operations", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_operations > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions", "MetricExpr": "tma_light_operations * BR_INST_RETIRED.ALL_BRANCHES / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_branch_instructions", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_branch_instructions > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", + "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes", "MetricExpr": "max( 0 , tma_light_operations - ( tma_fp_arith + tma_memory_operations + tma_branch_instructions ) )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_other_light_ops", "ScaleUnit": "100%", + "MetricThreshold": "tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_light_operations_group;Slots", "MetricName": "tma_nop_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "tma_microcode_sequencer + tma_retiring * ( UOPS_DECODED.DEC0 - cpu@UOPS_DECODED.DEC0\\,cmask\\=0x1@ ) / IDQ.MITE_UOPS", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_heavy_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+]).", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops", "MetricExpr": "tma_heavy_operations - tma_microcode_sequencer", - "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueD0", "MetricName": "tma_few_uops_instructions", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions." + "MetricThreshold": "tma_few_uops_instructions > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions. Related metrics: tma_decoder0_alone." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) * IDQ.MS_UOPS / tma_info_thread_slots", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS." + "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( 34 ) * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists", "MetricExpr": "34 * ASSISTS.FP / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_fp_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_assists > 0.1", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", "ScaleUnit": "100%", + "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources." }, { - "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts.", + "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts", "MetricExpr": "100 * ( 1 - tma_core_bound / tma_ports_utilization if tma_core_bound < tma_ports_utilization else 1 ) if tma_info_system_smt_2t_utilization > 0.5 else 0", "MetricGroup": "Cor;SMT;Metric", - "MetricName": "tma_info_botlnk_l0_core_bound_likely" + "MetricName": "tma_info_botlnk_l0_core_bound_likely", + "MetricThreshold": "tma_info_botlnk_core_bound_likely > 0.5" }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", - "MetricName": "tma_info_thread_uoppi" + "MetricName": "tma_info_thread_uoppi", + "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", - "MetricName": "tma_info_thread_uptb" + "MetricName": "tma_info_thread_uptb", + "MetricThreshold": "tma_info_thread_uptb < 5 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor.", + "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor", "MetricExpr": "tma_info_thread_slots / ( TOPDOWN.SLOTS / 2 ) if #SMT_on else 1", "MetricGroup": "SMT;TmaL1;TopdownL1;tma_L1_group;Metric", "MetricName": "tma_info_thread_slots_utilization" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / tma_info_core_core_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "uops Executed per Cycle.", + "BriefDescription": "uops Executed per Cycle", "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", "MetricGroup": "Power;Metric", "MetricName": "tma_info_core_epc" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipload" + "MetricName": "tma_info_inst_mix_ipload", + "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipstore" + "MetricName": "tma_info_inst_mix_ipstore", + "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipbranch" + "MetricName": "tma_info_inst_mix_ipbranch", + "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipcall" + "MetricName": "tma_info_inst_mix_ipcall", + "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", - "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_iptb" + "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", + "MetricName": "tma_info_inst_mix_iptb", + "MetricThreshold": "tma_info_inst_mix_iptb < 5 * 2 + 1", + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", "MetricGroup": "Flops;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipflop" + "MetricName": "tma_info_inst_mix_ipflop", + "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", + "MetricThreshold": "tma_info_inst_mix_iparith < 10", "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", + "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", + "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx512", + "MetricThreshold": "tma_info_inst_mix_iparith_avx512 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / MISC_RETIRED.PAUSE_INST", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ippause" }, { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / SW_PREFETCH_ACCESS.ANY", "MetricGroup": "Prefetches;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipswpf" + "MetricName": "tma_info_inst_mix_ipswpf", + "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Instructions per a microcode Assist invocation.", + "BriefDescription": "Instructions per a microcode Assist invocation", "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", "MetricName": "tma_info_pipeline_ipassist", + "MetricThreshold": "tma_info_pipeline_ipassist < 100000", "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core.", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core", "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@ )", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Average number of uops fetched from LSD per cycle.", + "BriefDescription": "Average number of uops fetched from LSD per cycle", "MetricExpr": "LSD.UOPS / LSD.CYCLES_ACTIVE", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_lsd" }, { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", + "BriefDescription": "Average number of uops fetched from DSB per cycle", "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_dsb" }, { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", + "BriefDescription": "Average number of uops fetched from MITE per cycle", "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_mite" }, { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", + "BriefDescription": "Average number of Uops issued by front-end when it issued something", "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_frontend_fetch_upc" }, { - "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache).", + "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache)", "MetricExpr": "LSD.UOPS / ( UOPS_ISSUED.ANY )", "MetricGroup": "Fed;LSD;Metric", "MetricName": "tma_info_frontend_lsd_coverage" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( UOPS_ISSUED.ANY )", - "MetricGroup": "DSB;Fed;FetchBW;Metric", - "MetricName": "tma_info_frontend_dsb_coverage" + "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", + "MetricName": "tma_info_frontend_dsb_coverage", + "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 5 > 0.35", + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", + "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "DSBmiss;Metric", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Average Latency for L1 instruction cache misses.", + "BriefDescription": "Average Latency for L1 instruction cache misses", "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STALL\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "Fed;FetchLat;IcMiss;Metric", "MetricName": "tma_info_frontend_icache_miss_latency" }, { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", "MetricGroup": "DSBmiss;Fed;Inst_Metric", - "MetricName": "tma_info_frontend_ipdsb_miss_ret" + "MetricName": "tma_info_frontend_ipdsb_miss_ret", + "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache true code cacheline misses per kilo instruction", "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code" }, { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction", "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code_all" }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) )", - "MetricGroup": "DSBmiss;Fed;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_misses" + "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_misses", + "MetricThreshold": "tma_info_botlnk_dsb_misses > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) )", - "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_bandwidth" + "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", + "MetricThreshold": "tma_info_botlnk_dsb_bandwidth > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", + "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_ic_misses" + "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", + "MetricName": "tma_info_botlnk_l2_ic_misses", + "MetricThreshold": "tma_info_botlnk_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmispredict" + "MetricName": "tma_info_bad_spec_ipmispredict", + "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken" + "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", + "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_cond_taken" + "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", + "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_ret" + "MetricName": "tma_info_bad_spec_ipmisp_ret", + "MetricThreshold": "tma_info_bad_spec_ipmisp_ret < 500" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_indirect" + "MetricName": "tma_info_bad_spec_ipmisp_indirect", + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 5 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", - "MetricGroup": "Bad;BrMispredicts;Core_Metric", - "MetricName": "tma_info_bad_spec_branch_misprediction_cost" + "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", + "MetricName": "tma_info_bad_spec_branch_misprediction_cost", + "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_mispredicts_resteers." }, { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", + "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", "MetricGroup": "BrMispredicts;Metric", "MetricName": "tma_info_bad_spec_spec_clears_ratio" }, { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", + "BriefDescription": "Fraction of branches that are non-taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_nt" }, { - "BriefDescription": "Fraction of branches that are taken conditionals.", + "BriefDescription": "Fraction of branches that are taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_tk" }, { - "BriefDescription": "Fraction of branches that are CALL or RET.", + "BriefDescription": "Fraction of branches that are CALL or RET", "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_callret" }, { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", + "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_jump" }, { - "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group).", + "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group)", "MetricExpr": "1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_other_branches" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_RETIRED.L1_MISS + MEM_LOAD_RETIRED.FB_HIT )", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki_load" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * L2_RQSTS.RFO_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_all" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", + "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_fb_hpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Offcore;Metric", "MetricName": "tma_info_memory_l3_cache_access_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", - "MetricName": "tma_info_memory_tlb_page_walks_utilization" + "MetricName": "tma_info_memory_tlb_page_walks_utilization", + "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Fed;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_code_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_load_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_access_bw", "MetricGroup": "Mem;MemoryBW;Offcore;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_access_bw_2t" }, { - "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses.", + "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses", "MetricExpr": "L2_LINES_OUT.USELESS_HWPF / ( L2_LINES_OUT.SILENT + L2_LINES_OUT.NON_SILENT )", "MetricGroup": "Prefetches;Metric", - "MetricName": "tma_info_memory_prefetches_useless_hwpf" + "MetricName": "tma_info_memory_prefetches_useless_hwpf", + "MetricThreshold": "tma_info_memory_useless_hwpf > 0.15" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=0x1@", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", + "BriefDescription": "Un-cacheable retired load per kilo instruction", "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_uc_load_pki" }, { - "BriefDescription": "\"Bus lock\" per kilo instruction.", + "BriefDescription": "\"Bus lock\" per kilo instruction", "MetricExpr": "1000 * SQ_MISC.BUS_LOCK / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_bus_lock_pki" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0.", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0", "MetricExpr": "CORE_POWER.LVL0_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license0_utilization", "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0. This includes non-AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes." }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1.", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1", "MetricExpr": "CORE_POWER.LVL1_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license1_utilization", + "MetricThreshold": "tma_info_system_power_license1_utilization > 0.5", "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1. This includes high current AVX 256-bit instructions as well as low current AVX 512-bit instructions." }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX).", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX)", "MetricExpr": "CORE_POWER.LVL2_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license2_utilization", + "MetricThreshold": "tma_info_system_power_license2_utilization > 0.5", "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX). This includes high current AVX 512-bit instructions." }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", - "MetricName": "tma_info_system_kernel_utilization" + "MetricName": "tma_info_system_kernel_utilization", + "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "64 * ( UNC_ARB_TRK_REQUESTS.ALL + UNC_ARB_COH_TRK_REQUESTS.ALL ) / ( 1000000 ) / tma_info_system_time / 1000", - "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec", - "MetricName": "tma_info_system_dram_bw_use" + "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", + "MetricName": "tma_info_system_dram_bw_use", + "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "power@energy\\-pkg@ * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", - "MetricName": "tma_info_system_mux" + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "UNC_CLOCK.SOCKET", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", - "MetricName": "tma_info_system_ipfarbranch" + "MetricName": "tma_info_system_ipfarbranch", + "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" } ] \ No newline at end of file diff --git a/RKL/metrics/rocketlake_metrics.json b/RKL/metrics/rocketlake_metrics.json index d03b21cc..b154f99e 100644 --- a/RKL/metrics/rocketlake_metrics.json +++ b/RKL/metrics/rocketlake_metrics.json @@ -2,7 +2,7 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for 11th Generation Intel(R) Core(TM) Processor0", - "DatePublished": "11/12/2024", + "DatePublished": "11/15/2024", "Version": "1.0", "Legend": "", "TmaVersion": "5.01", @@ -107,7 +107,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Mispredictions > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Mispredictions" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Mispredictions > 20", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP", @@ -171,7 +179,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Big_Code > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Big_Code" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Big_Code > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB", @@ -332,7 +348,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Instruction_Fetch_BW" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", @@ -541,7 +565,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore", @@ -782,7 +814,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Latency" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;Mem;MemoryLat;Offcore", @@ -995,7 +1035,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Data_TLBs" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore", @@ -1188,7 +1236,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Synchronization > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Synchronization" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Synchronization > 10", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;LockCont;Mem;Offcore", @@ -1272,7 +1328,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Compute_Bound_Est" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20", + "ThresholdIssues": "$issueComp" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Cor", @@ -1473,7 +1537,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Irregular_Overhead > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Irregular_Overhead" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Irregular_Overhead > 10", + "ThresholdIssues": "$issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BvIO;Cor;Ret", @@ -1850,7 +1922,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Other_Bottlenecks" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;Cor;Offcore", @@ -1886,7 +1966,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Branching_Overhead > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Branching_Overhead" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Bottleneck_Branching_Overhead > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Ret", @@ -1966,7 +2054,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Useful_Work > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Useful_Work" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Useful_Work > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;Ret", @@ -2010,7 +2106,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;BvIO;TmaL1;PGO", @@ -2043,7 +2147,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Frontend;TmaL2", @@ -2072,7 +2188,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss", @@ -2105,7 +2237,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -2134,7 +2286,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -2163,7 +2335,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB", @@ -2196,7 +2384,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2225,7 +2433,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2262,7 +2490,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2299,7 +2551,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2332,7 +2608,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -2369,7 +2661,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Mispredicts_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP", @@ -2406,7 +2718,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Clears_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;MachineClears", @@ -2435,7 +2767,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Unknown_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat", @@ -2464,7 +2816,23 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC, $issueMS, $issueMV, $issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MicroSeq", @@ -2493,7 +2861,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LCP(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -2522,7 +2906,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchLat", @@ -2571,7 +2971,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;Frontend;TmaL2", @@ -2617,7 +3025,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -2663,7 +3083,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Decoder0_Alone(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -2696,7 +3132,23 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_......MITE_4wide(%) > 5 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MITE_4wide(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_......MITE_4wide(%) > 5 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -2742,7 +3194,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;FetchBW", @@ -2788,7 +3252,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....LSD(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LSD(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....LSD(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;LSD", @@ -2830,7 +3306,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 5 & b > 20", + "BaseFormula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -2878,7 +3366,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", @@ -2935,7 +3431,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2", @@ -2992,7 +3500,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;BrMispredicts", @@ -3049,7 +3573,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueMC, $issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2", @@ -3110,7 +3646,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Nukes(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;Machine_Clears", @@ -3154,7 +3706,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;TmaL1", @@ -3219,7 +3779,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20", + "BaseFormula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2", @@ -3252,7 +3824,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1, $issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -3293,7 +3881,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -3334,7 +3942,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3363,7 +3995,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3404,7 +4060,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3445,7 +4129,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3486,7 +4198,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3515,7 +4255,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Fwd_Blk(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3565,7 +4325,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L1_Latency_Dependency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -3610,7 +4390,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Lock_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Offcore", @@ -3651,7 +4451,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......Split_Loads(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Loads(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......Split_Loads(%) > 30", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3680,7 +4488,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......4K_Aliasing(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......4K_Aliasing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......4K_Aliasing(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3709,7 +4537,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......FB_Full(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FB_Full(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......FB_Full(%) > 30", + "ThresholdIssues": "$issueBW, $issueSL, $issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW", @@ -3758,7 +4594,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem", @@ -3808,7 +4660,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L2_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryLat", @@ -3841,7 +4713,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -3895,7 +4783,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Contested_Accesses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -3945,7 +4853,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Data_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;Offcore;Snoop", @@ -3995,7 +4923,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L3_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -4024,7 +4972,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......SQ_Full(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 30 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -4077,7 +5045,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -4106,7 +5090,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Bandwidth(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -4139,7 +5143,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat;Offcore", @@ -4168,7 +5192,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -4209,7 +5249,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO, $issueSL, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore", @@ -4251,7 +5311,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......False_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -4293,7 +5373,27 @@ "Category": "TMA", "CountDomain": "Core_Utilization", "Threshold": { - "Formula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4322,7 +5422,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Streaming_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBW;Offcore", @@ -4368,7 +5488,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -4414,7 +5554,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4456,7 +5620,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4510,7 +5698,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4564,7 +5780,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4618,7 +5862,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4683,7 +5955,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2;Compute", @@ -4712,7 +5996,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB", @@ -4741,7 +6041,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4774,7 +6094,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......INT_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4803,7 +6143,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;PortsUtil", @@ -4832,7 +6188,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Slow_Pause(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4897,7 +6273,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4926,7 +6318,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_0(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4955,7 +6367,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_........Mixing_Vectors(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Mixing_Vectors(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Mixing_Vectors(%) > 5", + "ThresholdIssues": "$issueMV" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4984,7 +6404,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_1(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -5013,7 +6453,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_2(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -5042,7 +6502,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_3m(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 40 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;PortsUtil", @@ -5096,7 +6576,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........ALU_Op_Utilization(%) > 40" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........ALU_Op_Utilization(%)" + } + ], + "Formula": "a > 40", + "BaseFormula": "metric_TMA_........ALU_Op_Utilization(%) > 40", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5138,7 +6626,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_0(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_0(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_0(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -5180,7 +6676,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_1(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_1(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_1(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5222,7 +6726,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_5(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_5(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_5(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5264,7 +6776,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_6(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_6(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_6(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5306,7 +6826,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Load_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Load_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5352,7 +6880,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Store_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Store_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5388,7 +6924,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Retiring(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 70 | b > 10", + "BaseFormula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;TmaL1", @@ -5453,7 +7001,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -5510,7 +7066,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 20 & b > 60", + "BaseFormula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -5555,7 +7123,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......X87_Use(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -5600,7 +7184,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Scalar(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5645,7 +7245,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5694,7 +7310,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5743,7 +7379,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5792,7 +7448,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_512b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5865,7 +7541,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Memory_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -5934,7 +7622,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Branch_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Branch_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -6027,7 +7727,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 30 & b > 60", + "BaseFormula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -6096,7 +7808,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Nop_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Pipeline", @@ -6161,7 +7889,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -6226,7 +7962,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Few_Uops_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6263,7 +8011,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueMC, $issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -6292,7 +8052,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Assists(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO", @@ -6321,7 +8097,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........FP_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........FP_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -6362,7 +8146,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......CISC(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6455,7 +8255,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_Core_Bound_Likely > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_Core_Bound_Likely" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Botlnk_Core_Bound_Likely > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;SMT", @@ -6482,6 +8290,11 @@ "BaseFormula": " inst_retired.any / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Summary", "LocateWith": "" @@ -6524,7 +8337,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UopPI > 1.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UopPI" + } + ], + "Formula": "a > 1.05", + "BaseFormula": "metric_TMA_Info_Thread_UopPI > 1.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret;Retire", @@ -6568,7 +8389,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UpTB < 5 * 1.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UpTB" + } + ], + "Formula": "a < 5 * 1.5", + "BaseFormula": "metric_TMA_Info_Thread_UpTB < 5 * 1.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW", @@ -6595,6 +8424,11 @@ "BaseFormula": " 1 / tma_info_thread_ipc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Mem", "LocateWith": "" @@ -6616,6 +8450,11 @@ "BaseFormula": " cpu_clk_unhalted.thread", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", "LocateWith": "" @@ -6637,6 +8476,11 @@ "BaseFormula": " topdown.slots:perf_metrics", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", "LocateWith": "" @@ -6671,6 +8515,11 @@ "BaseFormula": " tma_info_thread_slots / ( topdown.slots:percore / 2 ) if smt_on else 1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "SMT;TmaL1", "LocateWith": "" @@ -6696,6 +8545,11 @@ "BaseFormula": " uops_executed.thread / uops_issued.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline", "LocateWith": "" @@ -6734,6 +8588,11 @@ "BaseFormula": " inst_retired.any / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;SMT;TmaL1", "LocateWith": "" @@ -6788,6 +8647,11 @@ "BaseFormula": " ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.8_flops + 16 * fp_arith_inst_retired.512b_packed_single ) / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Flops", "LocateWith": "" @@ -6830,6 +8694,11 @@ "BaseFormula": " ( ( fp_arith_inst_retired.scalar ) + ( fp_arith_inst_retired.vector ) ) / ( 2 * tma_info_core_core_clks )", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -6855,6 +8724,11 @@ "BaseFormula": " uops_executed.thread / uops_executed.thread:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "LocateWith": "" @@ -6880,6 +8754,11 @@ "BaseFormula": " uops_executed.thread / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -6914,6 +8793,11 @@ "BaseFormula": " cpu_clk_unhalted.distributed if smt_on else tma_info_thread_clks", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "SMT", "LocateWith": "" @@ -6940,7 +8824,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpLoad < 3" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpLoad" + } + ], + "Formula": "a < 3", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpLoad < 3", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -6968,7 +8860,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpStore < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpStore" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpStore < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -6996,7 +8896,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpBranch < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpBranch" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpBranch < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;InsType", @@ -7024,7 +8932,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpCall < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpCall" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpCall < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", @@ -7052,7 +8968,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpTB < 5 * 2 + 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpTB" + } + ], + "Formula": "a < 5 * 2 + 1", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpTB < 5 * 2 + 1", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO", @@ -7079,6 +9003,11 @@ "BaseFormula": " br_inst_retired.all_branches / br_inst_retired.near_taken", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", "LocateWith": "" @@ -7121,7 +9050,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpFLOP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -7153,7 +9090,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -7181,7 +9126,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -7209,7 +9162,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -7241,7 +9202,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX128" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -7273,7 +9242,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX256" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -7305,7 +9282,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX512 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX512" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX512 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -7332,6 +9317,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / misc_retired.pause_inst", "Category": "TMA", "CountDomain": "Inst_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", "LocateWith": "" @@ -7358,7 +9348,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpSWPF" + } + ], + "Formula": "a < 100", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -7381,6 +9379,11 @@ "BaseFormula": " inst_retired.any", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;TmaL1", "LocateWith": "INST_RETIRED.PREC_DIST" @@ -7422,6 +9425,11 @@ "BaseFormula": " ( tma_retiring * tma_info_thread_slots ) / uops_retired.slots:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret", "LocateWith": "" @@ -7448,7 +9456,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Pipeline_IpAssist < 100000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Pipeline_IpAssist" + } + ], + "Formula": "a < 100000", + "BaseFormula": "metric_TMA_Info_Pipeline_IpAssist < 100000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", @@ -7488,6 +9504,11 @@ "BaseFormula": " uops_executed.thread / ( ( uops_executed.core_cycles_ge_1 / 2 ) if smt_on else uops_executed.thread:c1 )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "LocateWith": "" @@ -7513,6 +9534,11 @@ "BaseFormula": " lsd.uops / lsd.cycles_active", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7538,6 +9564,11 @@ "BaseFormula": " idq.dsb_uops / idq.dsb_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7563,6 +9594,11 @@ "BaseFormula": " idq.mite_uops / idq.mite_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7588,6 +9624,11 @@ "BaseFormula": " uops_issued.any / uops_issued.any:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7613,6 +9654,11 @@ "BaseFormula": " lsd.uops / ( uops_issued.any )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;LSD", "LocateWith": "" @@ -7639,7 +9685,19 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 5 > 0.35" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_DSB_Coverage" + }, + { + "Alias": "b", + "Value": "IPC" + } + ], + "Formula": "a < 0.7 & b / 5 > 0.35", + "BaseFormula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 5 > 0.35", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -7666,6 +9724,11 @@ "BaseFormula": " dsb2mite_switches.penalty_cycles / dsb2mite_switches.penalty_cycles:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss", "LocateWith": "" @@ -7691,6 +9754,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -7716,6 +9784,11 @@ "BaseFormula": " icache_16b.ifdata_stall / icache_16b.ifdata_stall:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", "LocateWith": "" @@ -7742,7 +9815,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret" + } + ], + "Formula": "a < 50", + "BaseFormula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -7769,6 +9850,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / baclears.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed", "LocateWith": "" @@ -7794,6 +9880,11 @@ "BaseFormula": " 1000 * frontend_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -7819,6 +9910,11 @@ "BaseFormula": " 1000 * l2_rqsts.code_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -7938,7 +10034,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Misses > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Misses" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Misses > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -8031,7 +10135,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Bandwidth" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -8095,7 +10207,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_IC_Misses > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_IC_Misses" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Info_Botlnk_IC_Misses > 5", + "ThresholdIssues": "$issueFL" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", @@ -8123,7 +10243,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMispredict" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts", @@ -8151,7 +10279,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8179,7 +10315,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8207,7 +10351,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Ret" + } + ], + "Formula": "a < 500", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8235,7 +10387,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect" + } + ], + "Formula": "a < 1000", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8338,6 +10498,11 @@ "BaseFormula": " tma_mispredictions * tma_info_thread_slots / ( 5 ) / br_misp_retired.all_branches / 100", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "$issueBM" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", "LocateWith": "" @@ -8367,6 +10532,11 @@ "BaseFormula": " int_misc.clears_count / ( br_misp_retired.all_branches + machine_clears.count )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", "LocateWith": "" @@ -8392,6 +10562,11 @@ "BaseFormula": " br_inst_retired.cond_ntaken / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", "LocateWith": "" @@ -8417,6 +10592,11 @@ "BaseFormula": " br_inst_retired.cond_taken / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", "LocateWith": "" @@ -8446,6 +10626,11 @@ "BaseFormula": " ( br_inst_retired.near_call + br_inst_retired.near_return ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8479,6 +10664,11 @@ "BaseFormula": " ( br_inst_retired.near_taken - br_inst_retired.cond_taken - 2 * br_inst_retired.near_call ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8520,6 +10710,11 @@ "BaseFormula": " 1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8549,6 +10744,11 @@ "BaseFormula": " l1d_pend_miss.pending / ( mem_load_retired.l1_miss + mem_load_retired.fb_hit )", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryLat", "LocateWith": "" @@ -8574,6 +10774,11 @@ "BaseFormula": " l1d_pend_miss.pending / l1d_pend_miss.pending_cycles", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryBW", "LocateWith": "" @@ -8599,6 +10804,11 @@ "BaseFormula": " 1000 * mem_load_retired.l1_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8624,6 +10834,11 @@ "BaseFormula": " 1000 * l2_rqsts.all_demand_data_rd / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8649,6 +10864,11 @@ "BaseFormula": " 1000 * mem_load_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;Backend;CacheHits", "LocateWith": "" @@ -8674,6 +10894,11 @@ "BaseFormula": " 1000 * l2_rqsts.miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem;Offcore", "LocateWith": "" @@ -8699,6 +10924,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8724,6 +10954,11 @@ "BaseFormula": " 1000 * l2_rqsts.rfo_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheMisses;Offcore", "LocateWith": "" @@ -8753,6 +10988,11 @@ "BaseFormula": " 1000 * ( l2_rqsts.references - l2_rqsts.miss ) / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8778,6 +11018,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8803,6 +11048,11 @@ "BaseFormula": " 1000 * mem_load_retired.l3_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -8828,6 +11078,11 @@ "BaseFormula": " 1000 * mem_load_retired.fb_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8854,6 +11109,11 @@ "BaseFormula": " 64 * l1d.replacement / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8880,6 +11140,11 @@ "BaseFormula": " 64 * l2_lines_in.all / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8906,6 +11171,11 @@ "BaseFormula": " 64 * longest_lat_cache.miss / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8932,6 +11202,11 @@ "BaseFormula": " 64 * offcore_requests.all_requests / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW;Offcore", "LocateWith": "" @@ -8979,7 +11254,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Page_Walks_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -9006,6 +11289,11 @@ "BaseFormula": " 1000 * itlb_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;MemoryTLB", "LocateWith": "" @@ -9031,6 +11319,11 @@ "BaseFormula": " 1000 * dtlb_load_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -9056,6 +11349,11 @@ "BaseFormula": " 1000 * dtlb_store_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -9082,6 +11380,11 @@ "BaseFormula": " tma_info_memory_l1d_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9108,6 +11411,11 @@ "BaseFormula": " tma_info_memory_l2_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9134,6 +11442,11 @@ "BaseFormula": " tma_info_memory_l3_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9160,6 +11473,11 @@ "BaseFormula": " tma_info_memory_l3_cache_access_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW;Offcore", "LocateWith": "" @@ -9190,7 +11508,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Useless_HWPF" + } + ], + "Formula": "a > 0.15", + "BaseFormula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -9217,6 +11543,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests.demand_data_rd", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Memory_Lat;Offcore", "LocateWith": "" @@ -9242,6 +11573,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests_outstanding.demand_data_rd:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -9267,6 +11603,11 @@ "BaseFormula": " offcore_requests_outstanding.all_data_rd / offcore_requests_outstanding.cycles_with_data_rd", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -9292,6 +11633,11 @@ "BaseFormula": " 1000 * mem_load_misc_retired.uc / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -9317,6 +11663,11 @@ "BaseFormula": " 1000 * sq_misc.bus_lock / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -9347,6 +11698,11 @@ "BaseFormula": " tma_info_system_cpus_utilized / num_cpus", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC;Summary", "LocateWith": "" @@ -9373,6 +11729,11 @@ "BaseFormula": " cpu_clk_unhalted.ref_tsc / tsc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", "LocateWith": "" @@ -9407,6 +11768,11 @@ "BaseFormula": " tma_info_system_turbo_utilization * tsc / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;Power", "LocateWith": "" @@ -9449,6 +11815,11 @@ "BaseFormula": " ( ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.8_flops + 16 * fp_arith_inst_retired.512b_packed_single ) / ( 1000000000 ) ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -9474,6 +11845,11 @@ "BaseFormula": " tma_info_thread_clks / cpu_clk_unhalted.ref_tsc", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -9512,6 +11888,11 @@ "BaseFormula": " core_power.lvl0_turbo_license / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -9551,7 +11932,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Power_License1_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Power_License1_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_System_Power_License1_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", @@ -9592,7 +11981,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Power_License2_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Power_License2_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_System_Power_License2_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", @@ -9628,6 +12025,11 @@ "BaseFormula": " 1 - cpu_clk_unhalted.one_thread_active / cpu_clk_unhalted.ref_distributed if smt_on else 0", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "SMT", "LocateWith": "" @@ -9654,7 +12056,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Kernel_Utilization > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Kernel_Utilization" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_System_Kernel_Utilization > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "OS", @@ -9681,6 +12091,11 @@ "BaseFormula": " cpu_clk_unhalted.thread_p:sup / inst_retired.any_p:sup", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "OS", "LocateWith": "" @@ -9711,6 +12126,11 @@ "BaseFormula": " 64 * ( unc_arb_trk_requests.all + unc_arb_coh_trk_requests.all ) / ( 1000000 ) / tma_info_system_time / 1000", "Category": "TMA", "CountDomain": "GB/sec", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "$issueBW" + }, "ResolutionLevels": "ARB", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC", "LocateWith": "" @@ -9737,6 +12157,11 @@ "BaseFormula": " unc_pkg_energy_status * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "PKG", "MetricGroup": "Power;SoC", "LocateWith": "" @@ -9759,7 +12184,15 @@ "Category": "TMA", "CountDomain": "Seconds", "Threshold": { - "Formula": "metric_TMA_Info_System_Time < 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Time" + } + ], + "Formula": "a < 1", + "BaseFormula": "metric_TMA_Info_System_Time < 1", + "ThresholdIssues": "" }, "ResolutionLevels": "SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -9787,7 +12220,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_MUX" + } + ], + "Formula": "a > 1.1 | a < 0.9", + "BaseFormula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -9810,6 +12251,11 @@ "BaseFormula": " unc_clock.socket", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CLOCK", "MetricGroup": "SoC", "LocateWith": "" @@ -9836,7 +12282,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_IpFarBranch < 1000000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_IpFarBranch" + } + ], + "Formula": "a < 1000000", + "BaseFormula": "metric_TMA_Info_System_IpFarBranch < 1000000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;OS", diff --git a/SKL/metrics/perf/skylake_metrics_perf.json b/SKL/metrics/perf/skylake_metrics_perf.json index e5b0fd32..6a1ee98c 100644 --- a/SKL/metrics/perf/skylake_metrics_perf.json +++ b/SKL/metrics/perf/skylake_metrics_perf.json @@ -1,1436 +1,1588 @@ [ { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", + "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_mispredictions" + "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueBM", + "MetricName": "tma_bottleneck_mispredictions", + "MetricThreshold": "tma_bottleneck_mispredictions > 20", + "PublicDescription": "Total pipeline cost of Branch Misprediction related bottlenecks. Related metrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers." }, { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", + "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_big_code" + "MetricName": "tma_bottleneck_big_code", + "MetricThreshold": "tma_bottleneck_big_code > 20" }, { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", + "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) ) ) - tma_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_instruction_fetch_bw" + "MetricName": "tma_bottleneck_instruction_fetch_bw", + "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) )", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_cache_memory_bandwidth" + "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueBW", + "MetricName": "tma_bottleneck_cache_memory_bandwidth", + "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_cache_memory_latency" + "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueLat", + "MetricName": "tma_bottleneck_cache_memory_latency", + "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks. Related metrics: tma_l3_hit_latency, tma_mem_latency." }, { - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", + "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs)", "MetricExpr": "100 * ( tma_memory_bound * ( tma_l1_bound / max( tma_memory_bound , ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) ) * ( tma_dtlb_load / max( tma_l1_bound , ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_memory_data_tlbs" + "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueTLB", + "MetricName": "tma_bottleneck_memory_data_tlbs", + "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", + "PublicDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs). Related metrics: tma_dtlb_load, tma_dtlb_store." }, { - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", + "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors)", "MetricExpr": "100 * ( tma_memory_bound * ( ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", - "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_memory_synchronization" + "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueSyncxn", + "MetricName": "tma_bottleneck_memory_synchronization", + "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", + "PublicDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", + "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation", "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Scaled_Slots", + "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", + "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", "PublicDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy." }, { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", + "BriefDescription": "Total pipeline cost of irregular execution (e.g", "MetricExpr": "100 * ( ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + tma_core_bound * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Scaled_Slots", + "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", - "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments)." + "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", + "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments). Related metrics: tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end.", + "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end", "MetricExpr": "100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", "MetricGroup": "BvOB;Cor;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", "MetricName": "tma_bottleneck_other_bottlenecks", + "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", "PublicDescription": "Total pipeline cost of remaining bottlenecks in the back-end. Examples include data-dependencies (Core Bound when Low ILP) and other unlisted memory-related stalls." }, { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", + "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA", "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Scaled_Slots", "MetricName": "tma_bottleneck_branching_overhead", + "MetricThreshold": "tma_bottleneck_branching_overhead > 5", "PublicDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound)." }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", + "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead", "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_useful_work" + "MetricName": "tma_bottleneck_useful_work", + "MetricThreshold": "tma_bottleneck_useful_work > 20" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "IDQ_UOPS_NOT_DELIVERED.CORE / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_frontend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_frontend_bound > 0.15", "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound. Sample with: FRONTEND_RETIRED.LATENCY_GE_4." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( 4 ) * IDQ_UOPS_NOT_DELIVERED.CYCLES_0_UOPS_DELIV.CORE / tma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", "MetricName": "tma_fetch_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_16, FRONTEND_RETIRED.LATENCY_GE_8." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "( ICACHE_16B.IFDATA_STALL + 2 * cpu@ICACHE_16B.IFDATA_STALL\\,cmask\\=0x1\\,edge\\=0x1@ ) / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_itlb_misses - tma_code_stlb_miss )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + ( 9 ) * BACLEARS.ANY / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "( 1 - ( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "( 9 ) * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: BACLEARS.ANY." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 2 ) * IDQ.MS_SWITCHES / tma_info_thread_clks", - "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated", + "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES." + "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "DECODE.LCP / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." + "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", - "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS." + "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "tma_frontend_bound - tma_fetch_latency", - "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", + "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2." + "MetricThreshold": "tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.ALL_MITE_CYCLES_ANY_UOPS - IDQ.ALL_MITE_CYCLES_4_UOPS ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", "ScaleUnit": "100%", + "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder.", + "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder", "MetricExpr": "( cpu@INST_DECODED.DECODERS\\,cmask\\=0x1@ - cpu@INST_DECODED.DECODERS\\,cmask\\=0x2@ ) / tma_info_core_core_clks / 2", - "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated", + "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated;tma_issueD0", "MetricName": "tma_decoder0_alone", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder. Related metrics: tma_few_uops_instructions." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.DSB_CYCLES_ANY - IDQ.DSB_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", "ScaleUnit": "100%", + "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "( UOPS_ISSUED.ANY - ( UOPS_RETIRED.RETIRE_SLOTS ) + ( 4 ) * ( ( INT_MISC.RECOVERY_CYCLES_ANY / 2 ) if #SMT_on else INT_MISC.RECOVERY_CYCLES ) ) / tma_info_thread_slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_bad_speculation", "ScaleUnit": "100%", + "MetricThreshold": "tma_bad_speculation > 0.15", "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example." }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * tma_bad_speculation", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", + "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types)", "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Slots", "MetricName": "tma_other_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "tma_bad_speculation - tma_branch_mispredicts", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", + "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT." + "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering", "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_machine_clears_group;Slots", "MetricName": "tma_other_nukes", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "1 - tma_frontend_bound - ( UOPS_ISSUED.ANY + ( 4 ) * ( ( INT_MISC.RECOVERY_CYCLES_ANY / 2 ) if #SMT_on else INT_MISC.RECOVERY_CYCLES ) ) / tma_info_thread_slots", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_backend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_backend_bound > 0.2", "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound." }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( ( CYCLE_ACTIVITY.STALLS_MEM_ANY + EXE_ACTIVITY.BOUND_ON_STORES ) / ( CYCLE_ACTIVITY.STALLS_TOTAL + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) + EXE_ACTIVITY.BOUND_ON_STORES ) ) * tma_backend_bound", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_memory_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two)." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", + "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT." + "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "min( ( 9 ) * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS." + "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", "MetricName": "tma_load_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", "MetricName": "tma_load_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", + "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache", "MetricExpr": "min( 2 * ( MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETIRED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS ) * 20 / 100 , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_l1_latency_dependency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( 12 * max( 0 , MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS.ALL_RFO ) + ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) * ( ( 9 ) * L2_RQSTS.RFO_HIT + ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) ) / tma_info_thread_clks", - "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks", + "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS." + "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset.", + "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset", "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_4k_aliasing", "ScaleUnit": "100%", + "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset. False match is possible; which incur a few cycles load re-issue. However; the short re-issue duration is often hidden by the out-of-order core and HW optimizations; hence a user may safely ignore a high value of this metric unless it manages to propagate up into parent nodes of the hierarchy (e.g. to L1_Bound)." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PEND_MISS.FB_FULL\\,cmask\\=0x1@ / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", + "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory)." + "MetricThreshold": "tma_fb_full > 0.3", + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( ( MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) ) ) / ( ( MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) ) ) + cpu@L1D_PEND_MISS.FB_FULL\\,cmask\\=0x1@ ) ) * ( ( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks )", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", + "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited)", "MetricExpr": "( 3.5 * tma_info_system_core_frequency ) * MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", "MetricName": "tma_l2_hit_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L2_MISS - CYCLE_ACTIVITY.STALLS_L3_MISS ) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( ( 22 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM ) + ( ( 20 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS." + "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( ( 20 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT." + "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( ( 10 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_RETIRED.L3_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT." + "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_bottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "( ( OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 ) if #SMT_on else OFFCORE_REQUESTS_BUFFER.SQ_FULL ) / tma_info_core_core_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clks + ( ( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks ) - tma_l2_bound )", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that)." + "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that)." + "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( L2_RQSTS.RFO_HIT * ( 9 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)." + "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( 22 * tma_info_system_core_frequency ) * ( OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM." + "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization", + "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", "ScaleUnit": "100%", - "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." + "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES. Related metrics: tma_port_4." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( ( 9 ) * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_STORE_MISSES.WALK_ACTIVE ) / tma_info_core_core_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES." + "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_load." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", "MetricName": "tma_store_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", "MetricName": "tma_store_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "tma_backend_bound - tma_memory_bound", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_core_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations)." }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", "ScaleUnit": "100%", + "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIVIDER_ACTIVE." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations", "MetricExpr": "PARTIAL_RAT_STALLS.SCOREBOARD / tma_info_thread_clks", - "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", + "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks;tma_issueSO", "MetricName": "tma_serializing_operation", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: PARTIAL_RAT_STALLS.SCOREBOARD." + "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: PARTIAL_RAT_STALLS.SCOREBOARD. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( tma_ports_utilized_0 * tma_info_thread_clks + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIVIDER_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - CYCLE_ACTIVITY.STALLS_MEM_ANY ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.EXE_BOUND_0_PORTS / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", + "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles)", "MetricExpr": "UOPS_ISSUED.VECTOR_WIDTH_MISMATCH / UOPS_ISSUED.ANY", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks;tma_issueMV", "MetricName": "tma_mixing_vectors", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic." + "MetricThreshold": "tma_mixing_vectors > 0.05", + "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 - UOPS_EXECUTED.CORE_CYCLES_GE_2 ) / 2 if #SMT_on else EXE_ACTIVITY.1_PORTS_UTIL ) / tma_info_core_core_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful." + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( UOPS_EXECUTED.CORE_CYCLES_GE_2 - UOPS_EXECUTED.CORE_CYCLES_GE_3 ) / 2 if #SMT_on else EXE_ACTIVITY.2_PORTS_UTIL ) / tma_info_core_core_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop." + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( UOPS_EXECUTED.CORE_CYCLES_GE_3 / 2 if #SMT_on else UOPS_EXECUTED.CORE_CYCLES_GE_3 ) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT.PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_0 / tma_info_core_core_clks", - "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0." + "MetricThreshold": "tma_port_0 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_1 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1." + "MetricThreshold": "tma_port_1 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_5 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_5", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5." + "MetricThreshold": "tma_port_5 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_6 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1." + "MetricThreshold": "tma_port_6 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "( UOPS_DISPATCHED_PORT.PORT_2 + UOPS_DISPATCHED_PORT.PORT_3 + UOPS_DISPATCHED_PORT.PORT_7 - UOPS_DISPATCHED_PORT.PORT_4 ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_op_utilization > 0.6" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_2 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_group;Core_Clocks", "MetricName": "tma_port_2", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_2 > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads). Sample with: UOPS_DISPATCHED_PORT.PORT_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_3 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_group;Core_Clocks", "MetricName": "tma_port_3", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_3 > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads). Sample with: UOPS_DISPATCHED_PORT.PORT_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_op_utilization > 0.6" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks;tma_issueSpSt", "MetricName": "tma_port_4", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data). Sample with: UOPS_DISPATCHED_PORT.PORT_4." + "MetricThreshold": "tma_port_4 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data). Sample with: UOPS_DISPATCHED_PORT.PORT_4. Related metrics: tma_split_stores." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_7 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks", "MetricName": "tma_port_7", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_7 > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address). Sample with: UOPS_DISPATCHED_PORT.PORT_7." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / tma_info_thread_slots", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_retiring", "ScaleUnit": "100%", + "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.1", "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.RETIRE_SLOTS." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_light_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", "ScaleUnit": "100%", + "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses", "MetricExpr": "tma_light_operations * MEM_INST_RETIRED.ANY / INST_RETIRED.ANY", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_memory_operations", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_operations > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions", "MetricExpr": "tma_light_operations * UOPS_RETIRED.MACRO_FUSED / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_fused_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Other_Light_Ops in MTL!)}." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused", "MetricExpr": "tma_light_operations * ( BR_INST_RETIRED.ALL_BRANCHES - UOPS_RETIRED.MACRO_FUSED ) / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_non_fused_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused. Non-conditional branches like direct JMP or CALL would count here. Can be used to examine fusible conditional jumps that were not fused." }, { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", + "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes", "MetricExpr": "max( 0 , tma_light_operations - ( tma_fp_arith + tma_memory_operations + tma_fused_instructions + tma_non_fused_branches ) )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_other_light_ops", "ScaleUnit": "100%", + "MetricThreshold": "tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_light_ops_group;Slots", "MetricName": "tma_nop_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "( ( UOPS_RETIRED.RETIRE_SLOTS ) + UOPS_RETIRED.MACRO_FUSED - INST_RETIRED.ANY ) / tma_info_thread_slots", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_heavy_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+])." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops", "MetricExpr": "tma_heavy_operations - tma_microcode_sequencer", - "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueD0", "MetricName": "tma_few_uops_instructions", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions." + "MetricThreshold": "tma_few_uops_instructions > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions. Related metrics: tma_decoder0_alone." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "( ( UOPS_RETIRED.RETIRE_SLOTS ) / UOPS_ISSUED.ANY ) * IDQ.MS_UOPS / tma_info_thread_slots", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS." + "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( 34 ) * ( FP_ASSIST.ANY + OTHER_ASSISTS.ANY ) / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists", "MetricExpr": "34 * FP_ASSIST.ANY / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_fp_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_assists > 0.1", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", "ScaleUnit": "100%", + "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources." }, { - "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts.", + "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts", "MetricExpr": "100 * ( 1 - tma_core_bound / tma_ports_utilization if tma_core_bound < tma_ports_utilization else 1 ) if tma_info_system_smt_2t_utilization > 0.5 else 0", "MetricGroup": "Cor;SMT;Metric", - "MetricName": "tma_info_botlnk_l0_core_bound_likely" + "MetricName": "tma_info_botlnk_l0_core_bound_likely", + "MetricThreshold": "tma_info_botlnk_core_bound_likely > 0.5" }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", - "MetricName": "tma_info_thread_uoppi" + "MetricName": "tma_info_thread_uoppi", + "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", - "MetricName": "tma_info_thread_uptb" + "MetricName": "tma_info_thread_uptb", + "MetricThreshold": "tma_info_thread_uptb < 4 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "( 4 ) * tma_info_core_core_clks", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / tma_info_core_core_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "uops Executed per Cycle.", + "BriefDescription": "uops Executed per Cycle", "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", "MetricGroup": "Power;Metric", "MetricName": "tma_info_core_epc" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "( CPU_CLK_UNHALTED.THREAD_ANY / 2 ) if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipload" + "MetricName": "tma_info_inst_mix_ipload", + "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipstore" + "MetricName": "tma_info_inst_mix_ipstore", + "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipbranch" + "MetricName": "tma_info_inst_mix_ipbranch", + "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipcall" + "MetricName": "tma_info_inst_mix_ipcall", + "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", - "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_iptb" + "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", + "MetricName": "tma_info_inst_mix_iptb", + "MetricThreshold": "tma_info_inst_mix_iptb < 4 * 2 + 1", + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipflop" + "MetricName": "tma_info_inst_mix_ipflop", + "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", + "MetricThreshold": "tma_info_inst_mix_iparith < 10", "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", + "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", + "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / SW_PREFETCH_ACCESS.ANY", "MetricGroup": "Prefetches;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipswpf" + "MetricName": "tma_info_inst_mix_ipswpf", + "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / cpu@UOPS_RETIRED.RETIRE_SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Instructions per a microcode Assist invocation.", + "BriefDescription": "Instructions per a microcode Assist invocation", "MetricExpr": "INST_RETIRED.ANY / ( FP_ASSIST.ANY + OTHER_ASSISTS.ANY )", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", "MetricName": "tma_info_pipeline_ipassist", + "MetricThreshold": "tma_info_pipeline_ipassist < 100000", "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core.", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core", "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@ )", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", + "BriefDescription": "Average number of uops fetched from DSB per cycle", "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_dsb" }, { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", + "BriefDescription": "Average number of uops fetched from MITE per cycle", "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_mite" }, { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", + "BriefDescription": "Average number of Uops issued by front-end when it issued something", "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_frontend_fetch_upc" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS )", - "MetricGroup": "DSB;Fed;FetchBW;Metric", - "MetricName": "tma_info_frontend_dsb_coverage" + "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", + "MetricName": "tma_info_frontend_dsb_coverage", + "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 4 > 0.35", + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", + "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / DSB2MITE_SWITCHES.COUNT", "MetricGroup": "DSBmiss;Metric", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Average Latency for L1 instruction cache misses.", + "BriefDescription": "Average Latency for L1 instruction cache misses", "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STALL\\,cmask\\=0x1\\,edge\\=0x1@ + 2", "MetricGroup": "Fed;FetchLat;IcMiss;Metric", "MetricName": "tma_info_frontend_icache_miss_latency" }, { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", "MetricGroup": "DSBmiss;Fed;Inst_Metric", - "MetricName": "tma_info_frontend_ipdsb_miss_ret" + "MetricName": "tma_info_frontend_ipdsb_miss_ret", + "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache true code cacheline misses per kilo instruction", "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code" }, { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction", "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code_all" }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb ) )", - "MetricGroup": "DSBmiss;Fed;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_misses" + "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_misses", + "MetricThreshold": "tma_info_botlnk_dsb_misses > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb ) ) )", - "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_bandwidth" + "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", + "MetricThreshold": "tma_info_botlnk_dsb_bandwidth > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", + "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_ic_misses" + "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", + "MetricName": "tma_info_botlnk_l2_ic_misses", + "MetricThreshold": "tma_info_botlnk_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmispredict" + "MetricName": "tma_info_bad_spec_ipmispredict", + "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / ( ( ( UOPS_RETIRED.RETIRE_SLOTS ) / UOPS_ISSUED.ANY ) * BR_MISP_EXEC.INDIRECT )", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_indirect" + "MetricName": "tma_info_bad_spec_ipmisp_indirect", + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 4 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", - "MetricGroup": "Bad;BrMispredicts;Core_Metric", - "MetricName": "tma_info_bad_spec_branch_misprediction_cost" + "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", + "MetricName": "tma_info_bad_spec_branch_misprediction_cost", + "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_mispredicts_resteers." }, { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", + "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", "MetricGroup": "BrMispredicts;Metric", "MetricName": "tma_info_bad_spec_spec_clears_ratio" }, { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", + "BriefDescription": "Fraction of branches that are non-taken conditionals", "MetricExpr": "BR_INST_RETIRED.NOT_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_nt" }, { - "BriefDescription": "Fraction of branches that are taken conditionals.", + "BriefDescription": "Fraction of branches that are taken conditionals", "MetricExpr": "( BR_INST_RETIRED.CONDITIONAL - BR_INST_RETIRED.NOT_TAKEN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_tk" }, { - "BriefDescription": "Fraction of branches that are CALL or RET.", + "BriefDescription": "Fraction of branches that are CALL or RET", "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_callret" }, { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", + "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - ( BR_INST_RETIRED.COND - BR_INST_RETIRED.NOT_TAKEN ) - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_jump" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_RETIRED.L1_MISS + MEM_LOAD_RETIRED.FB_HIT )", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki_load" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * OFFCORE_REQUESTS.DEMAND_RFO / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_all" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", + "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_fb_hpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Offcore;Metric", "MetricName": "tma_info_memory_l3_cache_access_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING + EPT.WALK_PENDING ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", - "MetricName": "tma_info_memory_tlb_page_walks_utilization" + "MetricName": "tma_info_memory_tlb_page_walks_utilization", + "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Fed;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_code_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_load_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_access_bw", "MetricGroup": "Mem;MemoryBW;Offcore;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_access_bw_2t" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", + "BriefDescription": "Un-cacheable retired load per kilo instruction", "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_uc_load_pki" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / ( CPU_CLK_UNHALTED.REF_XCLK_ANY / 2 ) if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", - "MetricName": "tma_info_system_kernel_utilization" + "MetricName": "tma_info_system_kernel_utilization", + "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "64 * ( UNC_ARB_TRK_REQUESTS.ALL + UNC_ARB_COH_TRK_REQUESTS.ALL ) / ( 1000000 ) / tma_info_system_time / 1000", - "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec", - "MetricName": "tma_info_system_dram_bw_use" + "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", + "MetricName": "tma_info_system_dram_bw_use", + "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Average latency of data read request to external memory (in nanoseconds).", + "BriefDescription": "Average latency of data read request to external memory (in nanoseconds)", "MetricExpr": "( 1000000000 ) * ( UNC_ARB_TRK_OCCUPANCY.DATA_READ / UNC_ARB_TRK_REQUESTS.DATA_READ ) / ( tma_info_system_socket_clks / tma_info_system_time )", "MetricGroup": "Mem;MemoryLat;SoC;NanoSeconds", "MetricName": "tma_info_system_mem_read_latency", "PublicDescription": "Average latency of data read request to external memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetches. ([RKL+]memory-controller only)." }, { - "BriefDescription": "Average number of parallel data read requests to external memory.", + "BriefDescription": "Average number of parallel data read requests to external memory", "MetricExpr": "UNC_ARB_TRK_OCCUPANCY.DATA_READ / cpu@UNC_ARB_TRK_OCCUPANCY.DATA_READ\\,cmask\\=0x1@", "MetricGroup": "Mem;MemoryBW;SoC;System_Metric", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches." }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "power@energy\\-pkg@ * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", - "MetricName": "tma_info_system_mux" + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "UNC_CLOCK.SOCKET", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", - "MetricName": "tma_info_system_ipfarbranch" + "MetricName": "tma_info_system_ipfarbranch", + "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" } ] \ No newline at end of file diff --git a/SKL/metrics/skylake_metrics.json b/SKL/metrics/skylake_metrics.json index 0a23ac57..161f0a14 100644 --- a/SKL/metrics/skylake_metrics.json +++ b/SKL/metrics/skylake_metrics.json @@ -2,7 +2,7 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for 6th Generation Intel(R) Core(TM) Processor0", - "DatePublished": "11/12/2024", + "DatePublished": "11/15/2024", "Version": "1.0", "Legend": "", "TmaVersion": "5.01", @@ -108,7 +108,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Mispredictions > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Mispredictions" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Mispredictions > 20", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP", @@ -181,7 +189,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Big_Code > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Big_Code" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Big_Code > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB", @@ -306,7 +322,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Instruction_Fetch_BW" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", @@ -519,7 +543,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore", @@ -748,7 +780,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Latency" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;Mem;MemoryLat;Offcore", @@ -949,7 +989,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Data_TLBs" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore", @@ -1130,7 +1178,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Synchronization > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Synchronization" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Synchronization > 10", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;LockCont;Mem;Offcore", @@ -1231,7 +1287,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Compute_Bound_Est" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20", + "ThresholdIssues": "$issueComp" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Cor", @@ -1396,7 +1460,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Irregular_Overhead > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Irregular_Overhead" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Irregular_Overhead > 10", + "ThresholdIssues": "$issueMS" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BvIO;Cor;Ret", @@ -1737,7 +1809,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Other_Bottlenecks" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;Cor;Offcore", @@ -1786,7 +1866,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Branching_Overhead > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Branching_Overhead" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Bottleneck_Branching_Overhead > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Ret", @@ -1863,7 +1951,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Useful_Work > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Useful_Work" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Useful_Work > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;Ret", @@ -1904,7 +2000,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;BvIO;TmaL1;PGO", @@ -1946,7 +2050,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Frontend;TmaL2", @@ -1979,7 +2095,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss", @@ -2008,7 +2140,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB", @@ -2041,7 +2189,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2070,7 +2238,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2107,7 +2295,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2144,7 +2356,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2177,7 +2413,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "~overlap" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -2214,7 +2466,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Mispredicts_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP", @@ -2251,7 +2523,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Clears_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;MachineClears", @@ -2280,7 +2572,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Unknown_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat", @@ -2309,7 +2621,23 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC, $issueMS, $issueMV, $issueSO" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MicroSeq", @@ -2338,7 +2666,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LCP(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -2367,7 +2711,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchLat", @@ -2413,7 +2773,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;Frontend;TmaL2", @@ -2459,7 +2827,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -2505,7 +2885,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Decoder0_Alone(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -2551,7 +2947,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;FetchBW", @@ -2604,7 +3012,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", @@ -2666,7 +3082,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2", @@ -2732,7 +3160,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;BrMispredicts", @@ -2794,7 +3238,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueMC, $issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2", @@ -2860,7 +3316,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Nukes(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;Machine_Clears", @@ -2913,7 +3385,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;TmaL1", @@ -2991,7 +3471,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20", + "BaseFormula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2", @@ -3024,7 +3516,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1, $issueMC" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -3065,7 +3573,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -3106,7 +3634,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3135,7 +3687,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3176,7 +3752,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3217,7 +3821,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3258,7 +3890,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3287,7 +3947,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Fwd_Blk(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3337,7 +4017,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L1_Latency_Dependency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -3382,7 +4082,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Lock_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Offcore", @@ -3423,7 +4143,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......Split_Loads(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Loads(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......Split_Loads(%) > 30", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3452,7 +4180,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......4K_Aliasing(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......4K_Aliasing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......4K_Aliasing(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3493,7 +4241,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......FB_Full(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FB_Full(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......FB_Full(%) > 30", + "ThresholdIssues": "$issueBW, $issueSL, $issueSmSt" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW", @@ -3542,7 +4298,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem", @@ -3592,7 +4364,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L2_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryLat", @@ -3625,7 +4417,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -3679,7 +4487,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Contested_Accesses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -3729,7 +4557,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Data_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;Offcore;Snoop", @@ -3779,7 +4627,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L3_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat, ~overlap" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -3821,7 +4689,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......SQ_Full(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 30 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -3874,7 +4762,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -3903,7 +4807,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Bandwidth(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -3936,7 +4860,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat;Offcore", @@ -3965,7 +4909,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -4006,7 +4966,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO, $issueSL, ~overlap" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore", @@ -4048,7 +5028,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......False_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -4090,7 +5090,27 @@ "Category": "TMA", "CountDomain": "Core_Utilization", "Threshold": { - "Formula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4136,7 +5156,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -4182,7 +5222,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4224,7 +5288,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4278,7 +5366,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4332,7 +5448,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4386,7 +5530,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4464,7 +5636,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2;Compute", @@ -4493,7 +5677,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB", @@ -4522,7 +5722,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSO" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;PortsUtil", @@ -4588,7 +5804,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4617,7 +5849,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_0(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4646,7 +5898,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_........Mixing_Vectors(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Mixing_Vectors(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Mixing_Vectors(%) > 5", + "ThresholdIssues": "$issueMV" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4696,7 +5956,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_1(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4746,7 +6026,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_2(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4788,7 +6088,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_3m(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 40 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;PortsUtil", @@ -4842,7 +6162,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........ALU_Op_Utilization(%) > 40" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........ALU_Op_Utilization(%)" + } + ], + "Formula": "a > 40", + "BaseFormula": "metric_TMA_........ALU_Op_Utilization(%) > 40", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4884,7 +6212,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_0(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_0(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_0(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -4926,7 +6262,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_1(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_1(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_1(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4968,7 +6312,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_5(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_5(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_5(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5010,7 +6362,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_6(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_6(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_6(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5064,7 +6424,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Load_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Load_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5106,7 +6474,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_2(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_2(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_2(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5148,7 +6524,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_3(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_3(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_3(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5190,7 +6574,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Store_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Store_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5232,7 +6624,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_4(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_4(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_4(%) > 60", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5274,7 +6674,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_7(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_7(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_7(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5315,7 +6723,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Retiring(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 70 | b > 10", + "BaseFormula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;TmaL1", @@ -5365,7 +6785,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -5423,7 +6851,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 20 & b > 60", + "BaseFormula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -5473,7 +6913,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......X87_Use(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -5502,7 +6958,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Scalar(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5531,7 +7003,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5564,7 +7052,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5597,7 +7105,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5651,7 +7179,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Memory_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -5701,7 +7241,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Fused_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -5755,7 +7307,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Non_Fused_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -5829,7 +7393,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 30 & b > 60", + "BaseFormula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -5883,7 +7459,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Nop_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Pipeline", @@ -5933,7 +7525,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -5991,7 +7591,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Few_Uops_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6041,7 +7653,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueMC, $issueMS" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -6087,7 +7711,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Assists(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO", @@ -6129,7 +7769,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........FP_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........FP_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -6187,7 +7835,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......CISC(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6280,7 +7944,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_Core_Bound_Likely > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_Core_Bound_Likely" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Botlnk_Core_Bound_Likely > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;SMT", @@ -6307,6 +7979,11 @@ "BaseFormula": " inst_retired.any / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Summary", "LocateWith": "" @@ -6333,7 +8010,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UopPI > 1.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UopPI" + } + ], + "Formula": "a > 1.05", + "BaseFormula": "metric_TMA_Info_Thread_UopPI > 1.05", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret;Retire", @@ -6361,7 +8046,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UpTB < 4 * 1.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UpTB" + } + ], + "Formula": "a < 4 * 1.5", + "BaseFormula": "metric_TMA_Info_Thread_UpTB < 4 * 1.5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW", @@ -6388,6 +8081,11 @@ "BaseFormula": " 1 / tma_info_thread_ipc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Mem", "LocateWith": "" @@ -6409,6 +8107,11 @@ "BaseFormula": " cpu_clk_unhalted.thread", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", "LocateWith": "" @@ -6443,6 +8146,11 @@ "BaseFormula": " ( 4 ) * tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", "LocateWith": "" @@ -6468,6 +8176,11 @@ "BaseFormula": " uops_executed.thread / uops_issued.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline", "LocateWith": "" @@ -6506,6 +8219,11 @@ "BaseFormula": " inst_retired.any / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;SMT;TmaL1", "LocateWith": "" @@ -6556,6 +8274,11 @@ "BaseFormula": " ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.256b_packed_single ) / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Flops", "LocateWith": "" @@ -6598,6 +8321,11 @@ "BaseFormula": " ( ( fp_arith_inst_retired.scalar ) + ( fp_arith_inst_retired.vector ) ) / ( 2 * tma_info_core_core_clks )", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -6623,6 +8351,11 @@ "BaseFormula": " uops_executed.thread / uops_executed.thread:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "LocateWith": "" @@ -6648,6 +8381,11 @@ "BaseFormula": " uops_executed.thread / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -6682,6 +8420,11 @@ "BaseFormula": " ( cpu_clk_unhalted.thread_any / 2 ) if smt_on else tma_info_thread_clks", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "SMT", "LocateWith": "" @@ -6708,7 +8451,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpLoad < 3" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpLoad" + } + ], + "Formula": "a < 3", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpLoad < 3", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -6736,7 +8487,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpStore < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpStore" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpStore < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -6764,7 +8523,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpBranch < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpBranch" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpBranch < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;InsType", @@ -6792,7 +8559,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpCall < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpCall" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpCall < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", @@ -6820,7 +8595,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpTB < 4 * 2 + 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpTB" + } + ], + "Formula": "a < 4 * 2 + 1", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpTB < 4 * 2 + 1", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO", @@ -6847,6 +8630,11 @@ "BaseFormula": " br_inst_retired.all_branches / br_inst_retired.near_taken", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", "LocateWith": "" @@ -6885,7 +8673,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpFLOP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -6917,7 +8713,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -6945,7 +8749,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -6973,7 +8785,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -7005,7 +8825,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX128" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -7037,7 +8865,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX256" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -7065,7 +8901,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpSWPF" + } + ], + "Formula": "a < 100", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -7088,6 +8932,11 @@ "BaseFormula": " inst_retired.any", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;TmaL1", "LocateWith": "INST_RETIRED.PREC_DIST" @@ -7113,6 +8962,11 @@ "BaseFormula": " ( uops_retired.retire_slots ) / uops_retired.retire_slots:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret", "LocateWith": "" @@ -7143,7 +8997,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Pipeline_IpAssist < 100000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Pipeline_IpAssist" + } + ], + "Formula": "a < 100000", + "BaseFormula": "metric_TMA_Info_Pipeline_IpAssist < 100000", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", @@ -7183,6 +9045,11 @@ "BaseFormula": " uops_executed.thread / ( ( uops_executed.core_cycles_ge_1 / 2 ) if smt_on else uops_executed.thread:c1 )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "LocateWith": "" @@ -7208,6 +9075,11 @@ "BaseFormula": " idq.dsb_uops / idq.dsb_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7233,6 +9105,11 @@ "BaseFormula": " idq.mite_uops / idq.mite_cycles", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7258,6 +9135,11 @@ "BaseFormula": " uops_issued.any / uops_issued.any:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7288,7 +9170,19 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 4 > 0.35" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_DSB_Coverage" + }, + { + "Alias": "b", + "Value": "IPC" + } + ], + "Formula": "a < 0.7 & b / 4 > 0.35", + "BaseFormula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 4 > 0.35", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -7315,6 +9209,11 @@ "BaseFormula": " dsb2mite_switches.penalty_cycles / dsb2mite_switches.count", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss", "LocateWith": "" @@ -7340,6 +9239,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -7365,6 +9269,11 @@ "BaseFormula": " icache_16b.ifdata_stall / icache_16b.ifdata_stall:c1:e1 + 2", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", "LocateWith": "" @@ -7391,7 +9300,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret" + } + ], + "Formula": "a < 50", + "BaseFormula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -7418,6 +9335,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / baclears.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Fed", "LocateWith": "" @@ -7443,6 +9365,11 @@ "BaseFormula": " 1000 * frontend_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -7468,6 +9395,11 @@ "BaseFormula": " 1000 * l2_rqsts.code_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -7559,7 +9491,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Misses > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Misses" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Misses > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -7620,7 +9560,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Bandwidth" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -7693,7 +9641,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_IC_Misses > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_IC_Misses" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Info_Botlnk_IC_Misses > 5", + "ThresholdIssues": "$issueFL" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", @@ -7721,7 +9677,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMispredict" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts", @@ -7757,7 +9721,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect" + } + ], + "Formula": "a < 1000", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -7861,6 +9833,11 @@ "BaseFormula": " tma_mispredictions * tma_info_thread_slots / ( 4 ) / br_misp_retired.all_branches / 100", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "$issueBM" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", "LocateWith": "" @@ -7890,6 +9867,11 @@ "BaseFormula": " int_misc.clears_count / ( br_misp_retired.all_branches + machine_clears.count )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", "LocateWith": "" @@ -7915,6 +9897,11 @@ "BaseFormula": " br_inst_retired.not_taken / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", "LocateWith": "" @@ -7944,6 +9931,11 @@ "BaseFormula": " ( br_inst_retired.conditional - br_inst_retired.not_taken ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", "LocateWith": "" @@ -7973,6 +9965,11 @@ "BaseFormula": " ( br_inst_retired.near_call + br_inst_retired.near_return ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8010,6 +10007,11 @@ "BaseFormula": " ( br_inst_retired.near_taken - ( br_inst_retired.cond - br_inst_retired.not_taken ) - 2 * br_inst_retired.near_call ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8039,6 +10041,11 @@ "BaseFormula": " l1d_pend_miss.pending / ( mem_load_retired.l1_miss + mem_load_retired.fb_hit )", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryLat", "LocateWith": "" @@ -8064,6 +10071,11 @@ "BaseFormula": " l1d_pend_miss.pending / l1d_pend_miss.pending_cycles", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryBW", "LocateWith": "" @@ -8089,6 +10101,11 @@ "BaseFormula": " 1000 * mem_load_retired.l1_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8114,6 +10131,11 @@ "BaseFormula": " 1000 * l2_rqsts.all_demand_data_rd / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8139,6 +10161,11 @@ "BaseFormula": " 1000 * mem_load_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;Backend;CacheHits", "LocateWith": "" @@ -8164,6 +10191,11 @@ "BaseFormula": " 1000 * l2_rqsts.miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem;Offcore", "LocateWith": "" @@ -8189,6 +10221,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8214,6 +10251,11 @@ "BaseFormula": " 1000 * offcore_requests.demand_rfo / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheMisses;Offcore", "LocateWith": "" @@ -8243,6 +10285,11 @@ "BaseFormula": " 1000 * ( l2_rqsts.references - l2_rqsts.miss ) / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8268,6 +10315,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8293,6 +10345,11 @@ "BaseFormula": " 1000 * mem_load_retired.l3_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -8318,6 +10375,11 @@ "BaseFormula": " 1000 * mem_load_retired.fb_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8344,6 +10406,11 @@ "BaseFormula": " 64 * l1d.replacement / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8370,6 +10437,11 @@ "BaseFormula": " 64 * l2_lines_in.all / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8396,6 +10468,11 @@ "BaseFormula": " 64 * longest_lat_cache.miss / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8422,6 +10499,11 @@ "BaseFormula": " 64 * offcore_requests.all_requests / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW;Offcore", "LocateWith": "" @@ -8473,7 +10555,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Page_Walks_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -8500,6 +10590,11 @@ "BaseFormula": " 1000 * itlb_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;MemoryTLB", "LocateWith": "" @@ -8525,6 +10620,11 @@ "BaseFormula": " 1000 * dtlb_load_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -8550,6 +10650,11 @@ "BaseFormula": " 1000 * dtlb_store_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -8576,6 +10681,11 @@ "BaseFormula": " tma_info_memory_l1d_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8602,6 +10712,11 @@ "BaseFormula": " tma_info_memory_l2_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8628,6 +10743,11 @@ "BaseFormula": " tma_info_memory_l3_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8654,6 +10774,11 @@ "BaseFormula": " tma_info_memory_l3_cache_access_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW;Offcore", "LocateWith": "" @@ -8679,6 +10804,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests.demand_data_rd", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Memory_Lat;Offcore", "LocateWith": "" @@ -8704,6 +10834,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests_outstanding.cycles_with_demand_data_rd", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -8729,6 +10864,11 @@ "BaseFormula": " offcore_requests_outstanding.all_data_rd / offcore_requests_outstanding.cycles_with_data_rd", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -8754,6 +10894,11 @@ "BaseFormula": " 1000 * mem_load_misc_retired.uc / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -8784,6 +10929,11 @@ "BaseFormula": " tma_info_system_cpus_utilized / num_cpus", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "HPC;Summary", "LocateWith": "" @@ -8810,6 +10960,11 @@ "BaseFormula": " cpu_clk_unhalted.ref_tsc / tsc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", "LocateWith": "" @@ -8844,6 +10999,11 @@ "BaseFormula": " tma_info_system_turbo_utilization * tsc / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "SYSTEM", "MetricGroup": "Summary;Power", "LocateWith": "" @@ -8882,6 +11042,11 @@ "BaseFormula": " ( ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.256b_packed_single ) / ( 1000000000 ) ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -8907,6 +11072,11 @@ "BaseFormula": " tma_info_thread_clks / cpu_clk_unhalted.ref_tsc", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -8941,6 +11111,11 @@ "BaseFormula": " 1 - cpu_clk_unhalted.one_thread_active / ( cpu_clk_unhalted.ref_xclk_any / 2 ) if smt_on else 0", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "SMT", "LocateWith": "" @@ -8967,7 +11142,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Kernel_Utilization > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Kernel_Utilization" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_System_Kernel_Utilization > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "OS", @@ -8994,6 +11177,11 @@ "BaseFormula": " cpu_clk_unhalted.thread_p:sup / inst_retired.any_p:sup", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "OS", "LocateWith": "" @@ -9024,6 +11212,11 @@ "BaseFormula": " 64 * ( unc_arb_trk_requests.all + unc_arb_coh_trk_requests.all ) / ( 1000000 ) / tma_info_system_time / 1000", "Category": "TMA", "CountDomain": "GB/sec", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "$issueBW" + }, "ResolutionLevels": "ARB", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC", "LocateWith": "" @@ -9058,6 +11251,11 @@ "BaseFormula": " ( 1000000000 ) * ( unc_arb_trk_occupancy.data_read / unc_arb_trk_requests.data_read ) / ( tma_info_system_socket_clks / tma_info_system_time )", "Category": "TMA", "CountDomain": "NanoSeconds", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryLat;SoC", "LocateWith": "" @@ -9083,6 +11281,11 @@ "BaseFormula": " unc_arb_trk_occupancy.data_read / unc_arb_trk_occupancy.data_read:c1", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "ARB", "MetricGroup": "Mem;MemoryBW;SoC", "LocateWith": "" @@ -9109,6 +11312,11 @@ "BaseFormula": " unc_pkg_energy_status * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "PKG", "MetricGroup": "Power;SoC", "LocateWith": "" @@ -9131,7 +11339,15 @@ "Category": "TMA", "CountDomain": "Seconds", "Threshold": { - "Formula": "metric_TMA_Info_System_Time < 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Time" + } + ], + "Formula": "a < 1", + "BaseFormula": "metric_TMA_Info_System_Time < 1", + "ThresholdIssues": "" }, "ResolutionLevels": "SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -9159,7 +11375,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_MUX" + } + ], + "Formula": "a > 1.1 | a < 0.9", + "BaseFormula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -9182,6 +11406,11 @@ "BaseFormula": " unc_clock.socket", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CLOCK", "MetricGroup": "SoC", "LocateWith": "" @@ -9208,7 +11437,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_IpFarBranch < 1000000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_IpFarBranch" + } + ], + "Formula": "a < 1000000", + "BaseFormula": "metric_TMA_Info_System_IpFarBranch < 1000000", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;OS", diff --git a/SKX/metrics/perf/skylakex_metrics_perf.json b/SKX/metrics/perf/skylakex_metrics_perf.json index bcc8f73a..a02af914 100644 --- a/SKX/metrics/perf/skylakex_metrics_perf.json +++ b/SKX/metrics/perf/skylakex_metrics_perf.json @@ -1,125 +1,125 @@ [ { - "BriefDescription": "CPU operating frequency (in GHz).", - "MetricExpr": "(( CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000)", + "BriefDescription": "CPU operating frequency (in GHz)", + "MetricExpr": "((CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000)", "MetricGroup": "", "MetricName": "cpu_operating_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Percentage of time spent in the active CPU power state C0.", + "BriefDescription": "Percentage of time spent in the active CPU power state C0", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "", "MetricName": "cpu_utilization", "ScaleUnit": "100%" }, { - "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles.", + "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles", "MetricExpr": "CPU_CLK_UNHALTED.THREAD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions", "MetricExpr": "MEM_INST_RETIRED.ALL_LOADS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "loads_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions", "MetricExpr": "MEM_INST_RETIRED.ALL_STORES / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "stores_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L1D.REPLACEMENT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L1_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions", "MetricExpr": "L2_RQSTS.ALL_CODE_RD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1_i_code_read_misses_with_prefetches_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L2_LINES_IN.ALL / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions", "MetricExpr": "L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_code_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", "MetricExpr": "cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x12d40433@ / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_data_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", "MetricExpr": "cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x12cc0233@ / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_code_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) in nano seconds.", - "MetricExpr": "( 1000000000 * ( cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS\\,config1\\=0x40433@ / cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40433@ ) / ( UNC_CHA_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) in nano seconds", + "MetricExpr": "( 1000000000 * (cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS\\,config1\\=0x40433@ / cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40433@) / (UNC_CHA_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_data_read_demand_plus_prefetch_miss_latency", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to local memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS\\,config1\\=0x40432@ / cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@ ) / ( UNC_CHA_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to local memory in nano seconds", + "MetricExpr": "( 1000000000 * (cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS\\,config1\\=0x40432@ / cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@) / (UNC_CHA_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_data_read_demand_plus_prefetch_miss_latency_for_local_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to remote memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS\\,config1\\=0x40431@ / cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@ ) / ( UNC_CHA_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand and prefetch data read miss (read memory access) addressed to remote memory in nano seconds", + "MetricExpr": "( 1000000000 * (cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS\\,config1\\=0x40431@ / cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@) / (UNC_CHA_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_data_read_demand_plus_prefetch_miss_latency_for_remote_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_mpi", @@ -127,7 +127,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the ITLB (Instruction TLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED_2M_4M / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_large_page_mpi", @@ -135,7 +135,7 @@ "PublicDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the Instruction Translation Lookaside Buffer (ITLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions", "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_load_mpi", @@ -143,7 +143,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions", "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_2mb_large_page_load_mpi", @@ -151,7 +151,7 @@ "PublicDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the Data Translation Lookaside Buffer (DTLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions", "MetricExpr": "DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_store_mpi", @@ -159,1134 +159,1255 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@ / ( cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@ + cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@ )", + "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@ / (cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@ + cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_local_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@ / ( cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@ + cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@ )", + "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@ / (cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40432@ + cha@UNC_CHA_TOR_INSERTS.IA_MISS\\,config1\\=0x40431@)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_remote_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Uncore operating frequency in GHz.", - "MetricExpr": "( UNC_CHA_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) / 1000000000) / duration_time", + "BriefDescription": "Uncore operating frequency in GHz", + "MetricExpr": "(UNC_CHA_CLOCKTICKS / ( #num_cores / #num_packages * #num_packages ) / 1000000000) / duration_time", "MetricGroup": "", "MetricName": "uncore_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data transmit bandwidth (MB/sec).", - "MetricExpr": "( UNC_UPI_TxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", + "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data transmit bandwidth (MB/sec)", + "MetricExpr": "(UNC_UPI_TxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", "MetricGroup": "", "MetricName": "upi_data_transmit_bw", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory read bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory read bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory write bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory write bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory bandwidth (MB/sec)", + "MetricExpr": "((UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_total", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU.", - "MetricExpr": "(( UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART0 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART1 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART2 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART3 ) * 4 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU", + "MetricExpr": "((UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART0 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART1 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART2 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART3) * 4 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU.", - "MetricExpr": "(( UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART0 + UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART1 + UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART2 + UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART3 ) * 4 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU", + "MetricExpr": "((UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART0 + UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART1 + UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART2 + UNC_IIO_PAYLOAD_BYTES_IN.MEM_WRITE.PART3) * 4 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.DSB_UOPS / UOPS_ISSUED.ANY )", + "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.DSB_UOPS / UOPS_ISSUED.ANY)", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_decoded_icache", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MITE_UOPS / UOPS_ISSUED.ANY )", + "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MITE_UOPS / UOPS_ISSUED.ANY)", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_legacy_decode_pipeline", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MS_UOPS / UOPS_ISSUED.ANY )", + "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MS_UOPS / UOPS_ISSUED.ANY)", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_microcode_sequencer", "ScaleUnit": "100%" }, { - "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to local memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.READS_LOCAL * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to local memory", + "MetricExpr": "(UNC_CHA_REQUESTS.READS_LOCAL * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_local_memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to local memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.WRITES_LOCAL * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to local memory", + "MetricExpr": "(UNC_CHA_REQUESTS.WRITES_LOCAL * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_local_memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to remote memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.READS_REMOTE * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to remote memory", + "MetricExpr": "(UNC_CHA_REQUESTS.READS_REMOTE * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_remote_memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to remote memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.WRITES_REMOTE * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to remote memory", + "MetricExpr": "(UNC_CHA_REQUESTS.WRITES_REMOTE * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_remote_memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data receive bandwidth (MB/sec).", - "MetricExpr": "( UNC_UPI_RxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", + "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data receive bandwidth (MB/sec)", + "MetricExpr": "(UNC_UPI_RxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", "MetricGroup": "", "MetricName": "upi_data_receive_bw", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", + "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_mispredictions" + "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueBM", + "MetricName": "tma_bottleneck_mispredictions", + "MetricThreshold": "tma_bottleneck_mispredictions > 20", + "PublicDescription": "Total pipeline cost of Branch Misprediction related bottlenecks. Related metrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers." }, { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", + "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_big_code" + "MetricName": "tma_bottleneck_big_code", + "MetricThreshold": "tma_bottleneck_big_code > 20" }, { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", + "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) ) ) - tma_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_instruction_fetch_bw" + "MetricName": "tma_bottleneck_instruction_fetch_bw", + "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) )", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_cache_memory_bandwidth" + "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueBW", + "MetricName": "tma_bottleneck_cache_memory_bandwidth", + "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_cache_memory_latency" + "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueLat", + "MetricName": "tma_bottleneck_cache_memory_latency", + "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks. Related metrics: tma_l3_hit_latency, tma_mem_latency." }, { - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", + "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs)", "MetricExpr": "100 * ( tma_memory_bound * ( tma_l1_bound / max( tma_memory_bound , ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) ) * ( tma_dtlb_load / max( tma_l1_bound , ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_memory_data_tlbs" + "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueTLB", + "MetricName": "tma_bottleneck_memory_data_tlbs", + "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", + "PublicDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs). Related metrics: tma_dtlb_load, tma_dtlb_store." }, { - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", + "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors)", "MetricExpr": "100 * ( tma_memory_bound * ( ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) * tma_remote_cache / ( tma_local_mem + tma_remote_mem + tma_remote_cache ) + ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", - "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_memory_synchronization" + "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueSyncxn", + "MetricName": "tma_bottleneck_memory_synchronization", + "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", + "PublicDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", + "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation", "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Scaled_Slots", + "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", + "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", "PublicDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy." }, { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", + "BriefDescription": "Total pipeline cost of irregular execution (e.g", "MetricExpr": "100 * ( ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + tma_core_bound * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Scaled_Slots", + "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Scaled_Slots;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", - "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments)." + "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", + "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments). Related metrics: tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end.", + "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end", "MetricExpr": "100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", "MetricGroup": "BvOB;Cor;Offcore;TopdownL1;tma_L1_group;Scaled_Slots", "MetricName": "tma_bottleneck_other_bottlenecks", + "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", "PublicDescription": "Total pipeline cost of remaining bottlenecks in the back-end. Examples include data-dependencies (Core Bound when Low ILP) and other unlisted memory-related stalls." }, { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", + "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA", "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Scaled_Slots", "MetricName": "tma_bottleneck_branching_overhead", + "MetricThreshold": "tma_bottleneck_branching_overhead > 5", "PublicDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound)." }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", + "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead", "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Scaled_Slots", - "MetricName": "tma_bottleneck_useful_work" + "MetricName": "tma_bottleneck_useful_work", + "MetricThreshold": "tma_bottleneck_useful_work > 20" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "IDQ_UOPS_NOT_DELIVERED.CORE / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_frontend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_frontend_bound > 0.15", "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound. Sample with: FRONTEND_RETIRED.LATENCY_GE_4." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( 4 ) * IDQ_UOPS_NOT_DELIVERED.CYCLES_0_UOPS_DELIV.CORE / tma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", "MetricName": "tma_fetch_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_16, FRONTEND_RETIRED.LATENCY_GE_8." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "( ICACHE_16B.IFDATA_STALL + 2 * cpu@ICACHE_16B.IFDATA_STALL\\,cmask\\=0x1\\,edge\\=0x1@ ) / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_itlb_misses - tma_code_stlb_miss )", - "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group", + "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", - "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group", + "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + ( 9 ) * BACLEARS.ANY / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "( 1 - ( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "( 9 ) * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: BACLEARS.ANY." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 2 ) * IDQ.MS_SWITCHES / tma_info_thread_clks", - "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated", + "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES." + "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "DECODE.LCP / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." + "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", - "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS." + "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "tma_frontend_bound - tma_fetch_latency", - "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", + "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2." + "MetricThreshold": "tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.ALL_MITE_CYCLES_ANY_UOPS - IDQ.ALL_MITE_CYCLES_4_UOPS ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", "ScaleUnit": "100%", + "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder.", + "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder", "MetricExpr": "( cpu@INST_DECODED.DECODERS\\,cmask\\=0x1@ - cpu@INST_DECODED.DECODERS\\,cmask\\=0x2@ ) / tma_info_core_core_clks / 2", - "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated", + "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated;tma_issueD0", "MetricName": "tma_decoder0_alone", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder. Related metrics: tma_few_uops_instructions." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.DSB_CYCLES_ANY - IDQ.DSB_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", "ScaleUnit": "100%", + "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "( UOPS_ISSUED.ANY - ( UOPS_RETIRED.RETIRE_SLOTS ) + ( 4 ) * ( ( INT_MISC.RECOVERY_CYCLES_ANY / 2 ) if #SMT_on else INT_MISC.RECOVERY_CYCLES ) ) / tma_info_thread_slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_bad_speculation", "ScaleUnit": "100%", + "MetricThreshold": "tma_bad_speculation > 0.15", "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example." }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * tma_bad_speculation", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", + "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types)", "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Slots", "MetricName": "tma_other_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "tma_bad_speculation - tma_branch_mispredicts", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", + "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT." + "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches, tma_remote_cache." }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering", "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_machine_clears_group;Slots", "MetricName": "tma_other_nukes", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "1 - tma_frontend_bound - ( UOPS_ISSUED.ANY + ( 4 ) * ( ( INT_MISC.RECOVERY_CYCLES_ANY / 2 ) if #SMT_on else INT_MISC.RECOVERY_CYCLES ) ) / tma_info_thread_slots", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_backend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_backend_bound > 0.2", "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound." }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( ( CYCLE_ACTIVITY.STALLS_MEM_ANY + EXE_ACTIVITY.BOUND_ON_STORES ) / ( CYCLE_ACTIVITY.STALLS_TOTAL + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) + EXE_ACTIVITY.BOUND_ON_STORES ) ) * tma_backend_bound", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_memory_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two)." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", + "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT." + "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "min( ( 9 ) * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS." + "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", "MetricName": "tma_load_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", "MetricName": "tma_load_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", + "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache", "MetricExpr": "min( 2 * ( MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETIRED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS ) * 20 / 100 , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_l1_latency_dependency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( 12 * max( 0 , MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS.ALL_RFO ) + ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) * ( ( 11 ) * L2_RQSTS.RFO_HIT + ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) ) / tma_info_thread_clks", - "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks", + "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS." + "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset.", + "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset", "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_4k_aliasing", "ScaleUnit": "100%", + "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset. False match is possible; which incur a few cycles load re-issue. However; the short re-issue duration is often hidden by the out-of-order core and HW optimizations; hence a user may safely ignore a high value of this metric unless it manages to propagate up into parent nodes of the hierarchy (e.g. to L1_Bound)." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PEND_MISS.FB_FULL\\,cmask\\=0x1@ / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", + "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory)." + "MetricThreshold": "tma_fb_full > 0.3", + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( ( MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) ) ) / ( ( MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) ) ) + cpu@L1D_PEND_MISS.FB_FULL\\,cmask\\=0x1@ ) ) * ( ( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks )", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", + "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited)", "MetricExpr": "( 3.5 * tma_info_system_core_frequency ) * MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", "MetricName": "tma_l2_hit_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L2_MISS - CYCLE_ACTIVITY.STALLS_L3_MISS ) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( ( 47.5 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * ( OFFCORE_RESPONSE.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE / ( OFFCORE_RESPONSE.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE + OFFCORE_RESPONSE.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) + ( ( 47.5 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS." + "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( ( 47.5 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * ( 1 - ( OFFCORE_RESPONSE.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE / ( OFFCORE_RESPONSE.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE + OFFCORE_RESPONSE.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT." + "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( ( 20.5 * tma_info_system_core_frequency ) - ( 3.5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_RETIRED.L3_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT." + "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_bottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "( ( OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 ) if #SMT_on else OFFCORE_REQUESTS_BUFFER.SQ_FULL ) / tma_info_core_core_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clks + ( ( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks ) - tma_l2_bound )", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that)." + "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that)." + "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory", "MetricExpr": "( ( 80 * tma_info_system_core_frequency ) - ( 20.5 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_local_mem", "ScaleUnit": "100%", + "MetricThreshold": "tma_local_mem > 0.1 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory. Caching will improve the latency and increase performance. Sample with: MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory", "MetricExpr": "( ( 147.5 * tma_info_system_core_frequency ) - ( 20.5 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_remote_mem", "ScaleUnit": "100%", + "MetricThreshold": "tma_remote_mem > 0.1 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues", "MetricExpr": "( ( ( 110 * tma_info_system_core_frequency ) - ( 20.5 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM + ( ( 110 * tma_info_system_core_frequency ) - ( 20.5 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", + "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_remote_cache", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD." + "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( L2_RQSTS.RFO_HIT * ( 11 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)." + "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( ( 110 * tma_info_system_core_frequency ) * ( OFFCORE_RESPONSE.DEMAND_RFO.L3_MISS.REMOTE_HITM + OFFCORE_RESPONSE.PF_L2_RFO.L3_MISS.REMOTE_HITM ) + ( 47.5 * tma_info_system_core_frequency ) * ( OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.HITM_OTHER_CORE + OFFCORE_RESPONSE.PF_L2_RFO.L3_HIT.HITM_OTHER_CORE ) ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.HITM_OTHER_CORE." + "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM, OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.HITM_OTHER_CORE. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization", + "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", "ScaleUnit": "100%", - "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." + "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES. Related metrics: tma_port_4." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( ( 9 ) * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_STORE_MISSES.WALK_ACTIVE ) / tma_info_core_core_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES." + "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_load." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", "MetricName": "tma_store_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", "MetricName": "tma_store_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "tma_backend_bound - tma_memory_bound", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_core_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations)." }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", "ScaleUnit": "100%", + "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIVIDER_ACTIVE." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations", "MetricExpr": "PARTIAL_RAT_STALLS.SCOREBOARD / tma_info_thread_clks", - "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Clocks;tma_issueSO", "MetricName": "tma_serializing_operation", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: PARTIAL_RAT_STALLS.SCOREBOARD." + "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: PARTIAL_RAT_STALLS.SCOREBOARD. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( tma_ports_utilized_0 * tma_info_thread_clks + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIVIDER_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - CYCLE_ACTIVITY.STALLS_MEM_ANY ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.EXE_BOUND_0_PORTS / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", + "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles)", "MetricExpr": "UOPS_ISSUED.VECTOR_WIDTH_MISMATCH / UOPS_ISSUED.ANY", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks;tma_issueMV", "MetricName": "tma_mixing_vectors", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic." + "MetricThreshold": "tma_mixing_vectors > 0.05", + "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 - UOPS_EXECUTED.CORE_CYCLES_GE_2 ) / 2 if #SMT_on else EXE_ACTIVITY.1_PORTS_UTIL ) / tma_info_core_core_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful." + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( ( UOPS_EXECUTED.CORE_CYCLES_GE_2 - UOPS_EXECUTED.CORE_CYCLES_GE_3 ) / 2 if #SMT_on else EXE_ACTIVITY.2_PORTS_UTIL ) / tma_info_core_core_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop." + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( UOPS_EXECUTED.CORE_CYCLES_GE_3 / 2 if #SMT_on else UOPS_EXECUTED.CORE_CYCLES_GE_3 ) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT.PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_0 / tma_info_core_core_clks", - "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0." + "MetricThreshold": "tma_port_0 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_1 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1." + "MetricThreshold": "tma_port_1 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_5 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_5", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5." + "MetricThreshold": "tma_port_5 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_6 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1." + "MetricThreshold": "tma_port_6 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "( UOPS_DISPATCHED_PORT.PORT_2 + UOPS_DISPATCHED_PORT.PORT_3 + UOPS_DISPATCHED_PORT.PORT_7 - UOPS_DISPATCHED_PORT.PORT_4 ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_op_utilization > 0.6" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_2 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_group;Core_Clocks", "MetricName": "tma_port_2", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_2 > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 2 ([SNB+]Loads and Store-address; [ICL+] Loads). Sample with: UOPS_DISPATCHED_PORT.PORT_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_3 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_group;Core_Clocks", "MetricName": "tma_port_3", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_3 > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 3 ([SNB+]Loads and Store-address; [ICL+] Loads). Sample with: UOPS_DISPATCHED_PORT.PORT_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_op_utilization > 0.6" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks;tma_issueSpSt", "MetricName": "tma_port_4", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data). Sample with: UOPS_DISPATCHED_PORT.PORT_4." + "MetricThreshold": "tma_port_4 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 4 (Store-data). Sample with: UOPS_DISPATCHED_PORT.PORT_4. Related metrics: tma_split_stores." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address)", "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_7 / tma_info_core_core_clks", "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_group;Core_Clocks", "MetricName": "tma_port_7", "ScaleUnit": "100%", + "MetricThreshold": "tma_port_7 > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 7 ([HSW+]simple Store-address). Sample with: UOPS_DISPATCHED_PORT.PORT_7." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / tma_info_thread_slots", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Slots", "MetricName": "tma_retiring", "ScaleUnit": "100%", + "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.1", "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.RETIRE_SLOTS." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_light_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", "ScaleUnit": "100%", + "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( cpu@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE\\,umask\\=0xFC@ ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / ( UOPS_RETIRED.RETIRE_SLOTS )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_512b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses", "MetricExpr": "tma_light_operations * MEM_INST_RETIRED.ANY / INST_RETIRED.ANY", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_memory_operations", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_operations > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions", "MetricExpr": "tma_light_operations * UOPS_RETIRED.MACRO_FUSED / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_fused_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Other_Light_Ops in MTL!)}." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused", "MetricExpr": "tma_light_operations * ( BR_INST_RETIRED.ALL_BRANCHES - UOPS_RETIRED.MACRO_FUSED ) / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_non_fused_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused. Non-conditional branches like direct JMP or CALL would count here. Can be used to examine fusible conditional jumps that were not fused." }, { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", + "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes", "MetricExpr": "max( 0 , tma_light_operations - ( tma_fp_arith + tma_memory_operations + tma_fused_instructions + tma_non_fused_branches ) )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_other_light_ops", "ScaleUnit": "100%", + "MetricThreshold": "tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( UOPS_RETIRED.RETIRE_SLOTS )", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_light_operations_group;Slots", "MetricName": "tma_nop_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "( ( UOPS_RETIRED.RETIRE_SLOTS ) + UOPS_RETIRED.MACRO_FUSED - INST_RETIRED.ANY ) / tma_info_thread_slots", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_heavy_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+])." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops", "MetricExpr": "tma_heavy_operations - tma_microcode_sequencer", - "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueD0", "MetricName": "tma_few_uops_instructions", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions." + "MetricThreshold": "tma_few_uops_instructions > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions. Related metrics: tma_decoder0_alone." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "( ( UOPS_RETIRED.RETIRE_SLOTS ) / UOPS_ISSUED.ANY ) * IDQ.MS_UOPS / tma_info_thread_slots", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS." + "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( 34 ) * ( FP_ASSIST.ANY + OTHER_ASSISTS.ANY ) / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists", "MetricExpr": "34 * FP_ASSIST.ANY / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_fp_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_assists > 0.1", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", "ScaleUnit": "100%", + "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources." }, { - "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts.", + "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts", "MetricExpr": "100 * ( 1 - tma_core_bound / tma_ports_utilization if tma_core_bound < tma_ports_utilization else 1 ) if tma_info_system_smt_2t_utilization > 0.5 else 0", "MetricGroup": "Cor;SMT;Metric", "MetricName": "tma_info_botlnk_l0_core_bound_likely", "MetricThreshold": "tma_info_botlnk_core_bound_likely > 0.5" }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", "MetricName": "tma_info_thread_uoppi", "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", "MetricName": "tma_info_thread_uptb", "MetricThreshold": "tma_info_thread_uptb < 4 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "( 4 ) * tma_info_core_core_clks", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / tma_info_core_core_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( cpu@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE\\,umask\\=0xFC@ ) ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "uops Executed per Cycle.", + "BriefDescription": "uops Executed per Cycle", "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", "MetricGroup": "Power;Metric", "MetricName": "tma_info_core_epc" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "( CPU_CLK_UNHALTED.THREAD_ANY / 2 ) if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipload", "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipstore", "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipbranch", "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", "MetricName": "tma_info_inst_mix_ipcall", "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", "MetricThreshold": "tma_info_inst_mix_iptb < 4 * 2 + 1", - "PublicDescription": "Instructions per taken branch. Related metrics: tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage." + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipflop", "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( cpu@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE\\,umask\\=0xFC@ ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", @@ -1294,7 +1415,7 @@ "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", @@ -1302,7 +1423,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", @@ -1310,7 +1431,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", @@ -1318,7 +1439,7 @@ "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", @@ -1326,7 +1447,7 @@ "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx512", @@ -1334,27 +1455,27 @@ "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / SW_PREFETCH_ACCESS.ANY", "MetricGroup": "Prefetches;Inst_Metric", "MetricName": "tma_info_inst_mix_ipswpf", "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( UOPS_RETIRED.RETIRE_SLOTS ) / cpu@UOPS_RETIRED.RETIRE_SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Instructions per a microcode Assist invocation.", + "BriefDescription": "Instructions per a microcode Assist invocation", "MetricExpr": "INST_RETIRED.ANY / ( FP_ASSIST.ANY + OTHER_ASSISTS.ANY )", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", "MetricName": "tma_info_pipeline_ipassist", @@ -1362,96 +1483,98 @@ "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core.", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core", "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@ )", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", + "BriefDescription": "Average number of uops fetched from DSB per cycle", "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_dsb" }, { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", + "BriefDescription": "Average number of uops fetched from MITE per cycle", "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_mite" }, { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", + "BriefDescription": "Average number of Uops issued by front-end when it issued something", "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_frontend_fetch_upc" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS )", "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", "MetricName": "tma_info_frontend_dsb_coverage", "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 4 > 0.35", - "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb." + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", + "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / DSB2MITE_SWITCHES.COUNT", "MetricGroup": "DSBmiss;Metric", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Average Latency for L1 instruction cache misses.", + "BriefDescription": "Average Latency for L1 instruction cache misses", "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STALL\\,cmask\\=0x1\\,edge\\=0x1@ + 2", "MetricGroup": "Fed;FetchLat;IcMiss;Metric", "MetricName": "tma_info_frontend_icache_miss_latency" }, { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", "MetricGroup": "DSBmiss;Fed;Inst_Metric", "MetricName": "tma_info_frontend_ipdsb_miss_ret", "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache true code cacheline misses per kilo instruction", "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code" }, { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction", "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code_all" }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb ) )", "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_dsb_misses > 10", - "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_info_inst_mix_iptb, tma_info_frontend_dsb_coverage." + "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb ) ) )", - "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_bandwidth" + "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", + "MetricThreshold": "tma_info_botlnk_dsb_bandwidth > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", + "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", @@ -1459,284 +1582,284 @@ "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmispredict", "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / ( ( ( UOPS_RETIRED.RETIRE_SLOTS ) / UOPS_ISSUED.ANY ) * BR_MISP_EXEC.INDIRECT )", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_indirect", "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 4 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", "MetricName": "tma_info_bad_spec_branch_misprediction_cost", - "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)." + "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_mispredicts_resteers." }, { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", + "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", "MetricGroup": "BrMispredicts;Metric", "MetricName": "tma_info_bad_spec_spec_clears_ratio" }, { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", + "BriefDescription": "Fraction of branches that are non-taken conditionals", "MetricExpr": "BR_INST_RETIRED.NOT_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_nt" }, { - "BriefDescription": "Fraction of branches that are taken conditionals.", + "BriefDescription": "Fraction of branches that are taken conditionals", "MetricExpr": "( BR_INST_RETIRED.CONDITIONAL - BR_INST_RETIRED.NOT_TAKEN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_tk" }, { - "BriefDescription": "Fraction of branches that are CALL or RET.", + "BriefDescription": "Fraction of branches that are CALL or RET", "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_callret" }, { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", + "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - ( BR_INST_RETIRED.COND - BR_INST_RETIRED.NOT_TAKEN ) - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_jump" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_RETIRED.L1_MISS + MEM_LOAD_RETIRED.FB_HIT )", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki_load" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * OFFCORE_REQUESTS.DEMAND_RFO / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_all" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", + "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_fb_hpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Offcore;Metric", "MetricName": "tma_info_memory_l3_cache_access_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING + EPT.WALK_PENDING ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", "MetricName": "tma_info_memory_tlb_page_walks_utilization", "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Fed;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_code_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_load_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_access_bw", "MetricGroup": "Mem;MemoryBW;Offcore;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_access_bw_2t" }, { - "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory).", + "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory)", "MetricExpr": "1000 * L2_LINES_OUT.SILENT / tma_info_inst_mix_instructions", "MetricGroup": "L2Evicts;Mem;Server;Core_Metric", "MetricName": "tma_info_memory_core_l2_evictions_silent_pki" }, { - "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction.", + "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction", "MetricExpr": "1000 * L2_LINES_OUT.NON_SILENT / tma_info_inst_mix_instructions", "MetricGroup": "L2Evicts;Mem;Server;Core_Metric", "MetricName": "tma_info_memory_core_l2_evictions_nonsilent_pki" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", + "BriefDescription": "Un-cacheable retired load per kilo instruction", "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_uc_load_pki" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz].", + "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz]", "MetricExpr": "tma_info_system_socket_clks / 1e9 / tma_info_system_time", "MetricGroup": "SoC;System_Metric", "MetricName": "tma_info_system_uncore_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0.", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0", "MetricExpr": "CORE_POWER.LVL0_TURBO_LICENSE / 2 / tma_info_core_core_clks if #SMT_on else CORE_POWER.LVL0_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license0_utilization", "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0. This includes non-AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes." }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1.", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1", "MetricExpr": "CORE_POWER.LVL1_TURBO_LICENSE / 2 / tma_info_core_core_clks if #SMT_on else CORE_POWER.LVL1_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license1_utilization", @@ -1744,7 +1867,7 @@ "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1. This includes high current AVX 256-bit instructions as well as low current AVX 512-bit instructions." }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX).", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX)", "MetricExpr": "CORE_POWER.LVL2_TURBO_LICENSE / 2 / tma_info_core_core_clks if #SMT_on else CORE_POWER.LVL2_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license2_utilization", @@ -1752,94 +1875,95 @@ "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX). This includes high current AVX 512-bit instructions." }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / ( CPU_CLK_UNHALTED.REF_XCLK_ANY / 2 ) if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_utilization", "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "( 64 * ( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", "MetricName": "tma_info_system_dram_bw_use", - "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]." + "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Average latency of data read request to external memory (in nanoseconds).", + "BriefDescription": "Average latency of data read request to external memory (in nanoseconds)", "MetricExpr": "( 1000000000 ) * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( tma_info_system_socket_clks / tma_info_system_time )", "MetricGroup": "Mem;MemoryLat;SoC;NanoSeconds", "MetricName": "tma_info_system_mem_read_latency", "PublicDescription": "Average latency of data read request to external memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetches. ([RKL+]memory-controller only)." }, { - "BriefDescription": "Average number of parallel data read requests to external memory.", + "BriefDescription": "Average number of parallel data read requests to external memory", "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD\\,thresh\\=0x1@", "MetricGroup": "Mem;MemoryBW;SoC;System_Metric", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches." }, { - "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds].", + "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]", "MetricExpr": "( 1000000000 ) * ( UNC_M_RPQ_OCCUPANCY / UNC_M_RPQ_INSERTS ) / imc_0@event\\=0x0@", "MetricGroup": "MemOffcore;MemoryLat;SoC;Server;NanoSeconds", "MetricName": "tma_info_system_mem_dram_read_latency", "PublicDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]. Accounts for demand loads and L1/L2 data-read prefetches." }, { - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec].", + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]", "MetricExpr": "( UNC_IIO_DATA_REQ_OF_CPU.MEM_WRITE.PART0 + UNC_IIO_DATA_REQ_OF_CPU.MEM_WRITE.PART1 + UNC_IIO_DATA_REQ_OF_CPU.MEM_WRITE.PART2 + UNC_IIO_DATA_REQ_OF_CPU.MEM_WRITE.PART3 ) * 4 / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "IoBW;MemOffcore;SoC;Server;GB/sec", "MetricName": "tma_info_system_io_read_bw", "PublicDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]. Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU." }, { - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec].", + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]", "MetricExpr": "( UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART0 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART1 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART2 + UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.PART3 ) * 4 / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "IoBW;MemOffcore;SoC;Server;GB/sec", "MetricName": "tma_info_system_io_write_bw", "PublicDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]. Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU." }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "( power@energy\\-pkg@ * ( 61 ) + 15.6 * power@energy\\-ram@ ) / ( ( duration_time ) * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", "MetricName": "tma_info_system_mux", "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "cha_0@event\\=0x0@", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", "MetricName": "tma_info_system_ipfarbranch", diff --git a/SKX/metrics/skylakex_metrics.json b/SKX/metrics/skylakex_metrics.json index 86a6f2e8..48cbc4ef 100644 --- a/SKX/metrics/skylakex_metrics.json +++ b/SKX/metrics/skylakex_metrics.json @@ -2,7 +2,7 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for Intel(R) Xeon(R) Processor Scalable Family based on Skylake microarchitecture0", - "DatePublished": "11/05/2024", + "DatePublished": "11/15/2024", "Version": "1.0", "Legend": "", "TmaVersion": "5.01", @@ -960,7 +960,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Mispredictions > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Mispredictions" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Mispredictions > 20", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP", @@ -1033,7 +1041,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Big_Code > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Big_Code" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Big_Code > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB", @@ -1158,7 +1174,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Instruction_Fetch_BW" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", @@ -1379,7 +1403,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore", @@ -1628,7 +1660,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Latency" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;Mem;MemoryLat;Offcore", @@ -1841,7 +1881,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Data_TLBs" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore", @@ -2066,7 +2114,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Synchronization > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Synchronization" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Synchronization > 10", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;LockCont;Mem;Offcore", @@ -2167,7 +2223,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Compute_Bound_Est" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20", + "ThresholdIssues": "$issueComp" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Cor", @@ -2332,7 +2396,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Irregular_Overhead > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Irregular_Overhead" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Irregular_Overhead > 10", + "ThresholdIssues": "$issueMS" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BvIO;Cor;Ret", @@ -2709,7 +2781,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Other_Bottlenecks" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;Cor;Offcore", @@ -2758,7 +2838,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Branching_Overhead > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Branching_Overhead" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Bottleneck_Branching_Overhead > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Ret", @@ -2835,7 +2923,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Useful_Work > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Useful_Work" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Useful_Work > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;Ret", @@ -2876,7 +2972,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;BvIO;TmaL1;PGO", @@ -2918,7 +3022,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Frontend;TmaL2", @@ -2951,7 +3067,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss", @@ -2980,7 +3112,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB", @@ -3011,9 +3159,29 @@ "Formula": "100 * ( max( 0 , ( a / ( b ) ) - ( c / ( b ) ) ) )", "BaseFormula": " max( 0 , tma_itlb_misses - tma_code_stlb_miss )", "Category": "TMA", - "CountDomain": "", + "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "> 5 & P" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3040,9 +3208,29 @@ "Formula": "100 * ( a / ( b ) )", "BaseFormula": " itlb_misses.walk_active / tma_info_thread_clks", "Category": "TMA", - "CountDomain": "", + "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "> 5 & P" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3079,7 +3267,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3116,7 +3328,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3149,7 +3385,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "~overlap" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -3186,7 +3438,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Mispredicts_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP", @@ -3223,7 +3495,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Clears_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;MachineClears", @@ -3252,7 +3544,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Unknown_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat", @@ -3281,7 +3593,23 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC, $issueMS, $issueMV, $issueSO" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MicroSeq", @@ -3310,7 +3638,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LCP(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -3339,7 +3683,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchLat", @@ -3385,7 +3745,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;Frontend;TmaL2", @@ -3431,12 +3799,24 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" - }, - "ResolutionLevels": "CORE, SOCKET, SYSTEM", - "MetricGroup": "DSBmiss;FetchBW", - "LocateWith": " FRONTEND_RETIRED.ANY_DSB_MISS" - }, + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" + }, + "ResolutionLevels": "CORE, SOCKET, SYSTEM", + "MetricGroup": "DSBmiss;FetchBW", + "LocateWith": " FRONTEND_RETIRED.ANY_DSB_MISS" + }, { "MetricName": "Decoder0_Alone", "LegacyName": "metric_TMA_......Decoder0_Alone(%)", @@ -3477,7 +3857,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Decoder0_Alone(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -3523,7 +3919,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;FetchBW", @@ -3576,7 +3984,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", @@ -3638,7 +4054,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2", @@ -3704,7 +4132,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;BrMispredicts", @@ -3766,7 +4210,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueMC, $issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2", @@ -3832,7 +4288,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Nukes(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;Machine_Clears", @@ -3885,7 +4357,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;TmaL1", @@ -3963,7 +4443,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20", + "BaseFormula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2", @@ -3996,7 +4488,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1, $issueMC" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -4037,7 +4545,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -4078,7 +4606,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4107,7 +4659,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4148,7 +4724,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4189,7 +4793,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4230,7 +4862,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4259,7 +4919,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Fwd_Blk(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4309,7 +4989,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L1_Latency_Dependency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -4354,7 +5054,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Lock_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Offcore", @@ -4395,7 +5115,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......Split_Loads(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Loads(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......Split_Loads(%) > 30", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4424,7 +5152,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......4K_Aliasing(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......4K_Aliasing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......4K_Aliasing(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4465,7 +5213,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......FB_Full(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FB_Full(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......FB_Full(%) > 30", + "ThresholdIssues": "$issueBW, $issueSL, $issueSmSt" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW", @@ -4514,7 +5270,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem", @@ -4564,7 +5336,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L2_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryLat", @@ -4597,7 +5389,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -4659,7 +5467,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Contested_Accesses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -4721,7 +5549,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Data_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;Offcore;Snoop", @@ -4771,7 +5619,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L3_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat, ~overlap" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -4813,7 +5681,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......SQ_Full(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 30 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -4866,7 +5754,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -4895,7 +5799,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Bandwidth(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -4928,7 +5852,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat;Offcore", @@ -4978,7 +5922,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Local_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Local_MEM(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Local_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Server", @@ -5028,7 +5996,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Remote_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Remote_MEM(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Remote_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Server;Snoop", @@ -5082,7 +6074,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Remote_Cache(%) > 5 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Remote_Cache(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Remote_Cache(%) > 5 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Offcore;Server;Snoop", @@ -5111,7 +6127,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -5152,7 +6184,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO, $issueSL, ~overlap" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore", @@ -5206,7 +6258,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......False_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -5248,7 +6320,27 @@ "Category": "TMA", "CountDomain": "Core_Utilization", "Threshold": { - "Formula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5294,7 +6386,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -5340,7 +6452,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5382,7 +6518,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5436,7 +6596,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5490,7 +6678,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5544,7 +6760,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5622,7 +6866,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2;Compute", @@ -5651,7 +6907,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB", @@ -5680,7 +6952,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSO" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;PortsUtil", @@ -5746,7 +7034,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -5775,7 +7079,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_0(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -5804,7 +7128,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_........Mixing_Vectors(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Mixing_Vectors(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Mixing_Vectors(%) > 5", + "ThresholdIssues": "$issueMV" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5854,7 +7186,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_1(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -5904,7 +7256,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_2(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -5946,7 +7318,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_3m(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 40 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;PortsUtil", @@ -6000,7 +7392,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........ALU_Op_Utilization(%) > 40" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........ALU_Op_Utilization(%)" + } + ], + "Formula": "a > 40", + "BaseFormula": "metric_TMA_........ALU_Op_Utilization(%) > 40", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6042,7 +7442,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_0(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_0(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_0(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -6084,7 +7492,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_1(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_1(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_1(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6126,7 +7542,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_5(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_5(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_5(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6168,7 +7592,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_6(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_6(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_6(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6222,7 +7654,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Load_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Load_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6264,7 +7704,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_2(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_2(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_2(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6306,7 +7754,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_3(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_3(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_3(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6348,7 +7804,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Store_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Store_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6390,7 +7854,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_4(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_4(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_4(%) > 60", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6432,7 +7904,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_7(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_7(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_7(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6473,7 +7953,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Retiring(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 70 | b > 10", + "BaseFormula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;TmaL1", @@ -6523,7 +8015,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -6581,7 +8081,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 20 & b > 60", + "BaseFormula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -6631,7 +8143,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......X87_Use(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -6660,7 +8188,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Scalar(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -6689,7 +8233,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -6722,7 +8282,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -6755,7 +8335,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -6788,7 +8388,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_512b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -6842,7 +8462,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Memory_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -6892,7 +8524,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Fused_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -6946,7 +8590,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Non_Fused_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -7020,7 +8676,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 30 & b > 60", + "BaseFormula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -7074,7 +8742,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Nop_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Pipeline", @@ -7124,7 +8808,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -7182,7 +8874,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Few_Uops_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -7232,7 +8936,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueMC, $issueMS" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -7278,7 +8994,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Assists(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO", @@ -7320,7 +9052,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........FP_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........FP_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -7378,7 +9118,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......CISC(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -8650,6 +10406,11 @@ "BaseFormula": " idq.dsb_uops / idq.dsb_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -8675,6 +10436,11 @@ "BaseFormula": " idq.mite_uops / idq.mite_cycles", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -8804,6 +10570,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -9120,7 +10891,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Bandwidth" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -9803,6 +11582,11 @@ "BaseFormula": " 1000 * offcore_requests.demand_rfo / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "CacheMisses;Offcore", "LocateWith": "" @@ -11246,7 +13030,15 @@ "Category": "TMA", "CountDomain": "Seconds", "Threshold": { - "Formula": "metric_TMA_Info_System_Time < 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Time" + } + ], + "Formula": "a < 1", + "BaseFormula": "metric_TMA_Info_System_Time < 1", + "ThresholdIssues": "" }, "ResolutionLevels": "SOCKET, SYSTEM", "MetricGroup": "Summary", diff --git a/SPR/metrics/perf/sapphirerapids_metrics_perf.json b/SPR/metrics/perf/sapphirerapids_metrics_perf.json index 14be9f9f..dca91990 100644 --- a/SPR/metrics/perf/sapphirerapids_metrics_perf.json +++ b/SPR/metrics/perf/sapphirerapids_metrics_perf.json @@ -1,139 +1,139 @@ [ { - "BriefDescription": "CPU operating frequency (in GHz).", - "MetricExpr": "( CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000", + "BriefDescription": "CPU operating frequency (in GHz)", + "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000", "MetricGroup": "", "MetricName": "cpu_operating_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Percentage of time spent in the active CPU power state C0.", + "BriefDescription": "Percentage of time spent in the active CPU power state C0", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "", "MetricName": "cpu_utilization", "ScaleUnit": "100%" }, { - "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles.", + "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles", "MetricExpr": "CPU_CLK_UNHALTED.THREAD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions", "MetricExpr": "MEM_INST_RETIRED.ALL_LOADS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "loads_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions.", + "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions", "MetricExpr": "MEM_INST_RETIRED.ALL_STORES / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "stores_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L1D.REPLACEMENT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L1_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1d_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions", "MetricExpr": "L2_RQSTS.ALL_CODE_RD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l1_i_code_read_misses_with_prefetches_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_hits_per_instr", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions", "MetricExpr": "L2_LINES_IN.ALL / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions", "MetricExpr": "MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_data_read_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions", "MetricExpr": "L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "l2_demand_code_mpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_LLCPREFDATA + UNC_CHA_TOR_INSERTS.IA_MISS_DRD + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF ) / INST_RETIRED.ANY", + "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IA_MISS_LLCPREFDATA + UNC_CHA_TOR_INSERTS.IA_MISS_DRD + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF) / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_data_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", + "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", "MetricExpr": "UNC_CHA_TOR_INSERTS.IA_MISS_CRD / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "llc_code_read_mpi_demand_plus_prefetch", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_latency", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to local memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to local memory in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_latency_for_local_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to remote memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to remote memory in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_latency_for_remote_requests", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to Intel(R) Optane(TM) Persistent Memory(PMEM) in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PMM ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to Intel(R) Optane(TM) Persistent Memory(PMEM) in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PMM) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_to_pmem_latency", "ScaleUnit": "1ns" }, { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to DRAM in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR) * #num_packages ) ) ) * duration_time", + "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to DRAM in nano seconds", + "MetricExpr": "( 1000000000 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR) / (UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR) * #num_packages ) ) ) * duration_time", "MetricGroup": "", "MetricName": "llc_demand_data_read_miss_to_dram_latency", "ScaleUnit": "1ns" }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_2nd_level_mpi", @@ -141,7 +141,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the ITLB (Instruction TLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions", "MetricExpr": "ITLB_MISSES.WALK_COMPLETED_2M_4M / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "itlb_2nd_level_large_page_mpi", @@ -149,7 +149,7 @@ "PublicDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the Instruction Translation Lookaside Buffer (ITLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions", "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_2nd_level_load_mpi", @@ -157,7 +157,7 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions", "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_2nd_level_2mb_large_page_load_mpi", @@ -165,7 +165,7 @@ "PublicDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the Data Translation Lookaside Buffer (DTLB) and further levels of TLB." }, { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions.", + "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions", "MetricExpr": "DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "", "MetricName": "dtlb_2nd_level_store_mpi", @@ -173,1365 +173,1500 @@ "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." }, { - "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL ) / ( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE )", + "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL) / (UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_local_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE ) / ( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE )", + "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE) / (UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE)", "MetricGroup": "", "MetricName": "numa_reads_addressed_to_remote_dram", "ScaleUnit": "100%" }, { - "BriefDescription": "Uncore operating frequency in GHz.", - "MetricExpr": "( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) / 1000000000) / duration_time", + "BriefDescription": "Uncore operating frequency in GHz", + "MetricExpr": "(UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) / 1000000000) / duration_time", "MetricGroup": "", "MetricName": "uncore_frequency", "ScaleUnit": "1GHz" }, { - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data transmit bandwidth (MB/sec).", - "MetricExpr": "( UNC_UPI_TxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", + "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data transmit bandwidth (MB/sec)", + "MetricExpr": "(UNC_UPI_TxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", "MetricGroup": "", "MetricName": "upi_data_transmit_bw", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory read bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory read bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory write bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory write bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "DDR memory bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) * 64 / 1000000) / duration_time", + "BriefDescription": "DDR memory bandwidth (MB/sec)", + "MetricExpr": "((UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_bandwidth_total", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Memory write bandwidth (MB/sec) caused by directory updates; includes DDR and Intel(R) Optane(TM) Persistent Memory(PMEM).", - "MetricExpr": "(( UNC_CHA_DIR_UPDATE.HA + UNC_CHA_DIR_UPDATE.TOR + UNC_M2M_DIRECTORY_UPDATE.ANY ) * 64 / 1000000) / duration_time", + "BriefDescription": "Memory write bandwidth (MB/sec) caused by directory updates; includes DDR and Intel(R) Optane(TM) Persistent Memory(PMEM)", + "MetricExpr": "((UNC_CHA_DIR_UPDATE.HA + UNC_CHA_DIR_UPDATE.TOR + UNC_M2M_DIRECTORY_UPDATE.ANY) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "memory_extra_write_bw_due_to_directory_updates", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory read bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_PMM_RPQ_INSERTS * 64 / 1000000) / duration_time", + "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory read bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_PMM_RPQ_INSERTS * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "pmem_memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory write bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_PMM_WPQ_INSERTS * 64 / 1000000) / duration_time", + "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory write bandwidth (MB/sec)", + "MetricExpr": "(UNC_M_PMM_WPQ_INSERTS * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "pmem_memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_PMM_RPQ_INSERTS + UNC_M_PMM_WPQ_INSERTS ) * 64 / 1000000) / duration_time", + "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory bandwidth (MB/sec)", + "MetricExpr": "((UNC_M_PMM_RPQ_INSERTS + UNC_M_PMM_WPQ_INSERTS) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "pmem_memory_bandwidth_total", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_ITOM + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR ) * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_ITOM + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Percentage of inbound reads initiated by end device controllers that miss the L3 cache.", + "BriefDescription": "Percentage of inbound reads initiated by end device controllers that miss the L3 cache", "MetricExpr": "UNC_CHA_TOR_INSERTS.IO_MISS_PCIRDCUR / UNC_CHA_TOR_INSERTS.IO_PCIRDCUR", "MetricGroup": "", "MetricName": "io_percent_of_inbound_reads_that_miss_l3", "ScaleUnit": "100%" }, { - "BriefDescription": "Percentage of inbound partial cacheline writes initiated by end device controllers that miss the L3 cache.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_MISS_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_MISS_RFO ) / ( UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_RFO ) )", + "BriefDescription": "Percentage of inbound partial cacheline writes initiated by end device controllers that miss the L3 cache", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_MISS_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_MISS_RFO) / (UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_RFO) )", "MetricGroup": "", "MetricName": "io_percent_of_inbound_partial_writes_that_miss_l3", "ScaleUnit": "100%" }, { - "BriefDescription": "Percentage of inbound full cacheline writes initiated by end device controllers that miss the L3 cache.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_MISS_ITOM / UNC_CHA_TOR_INSERTS.IO_ITOM )", + "BriefDescription": "Percentage of inbound full cacheline writes initiated by end device controllers that miss the L3 cache", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IO_MISS_ITOM / UNC_CHA_TOR_INSERTS.IO_ITOM)", "MetricGroup": "", "MetricName": "io_percent_of_inbound_full_writes_that_miss_l3", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.DSB_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.DSB_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_decoded_icache", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MITE_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MITE_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_legacy_decode_pipeline", "ScaleUnit": "100%" }, { - "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MS_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", + "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue", + "MetricExpr": "(IDQ.MS_UOPS / (IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS) )", "MetricGroup": "", "MetricName": "percent_uops_delivered_from_microcode_sequencer", "ScaleUnit": "100%" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the local CPU socket.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_LOCAL * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the local CPU socket", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_LOCAL * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read_local", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from a remote CPU socket.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_REMOTE * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from a remote CPU socket", + "MetricExpr": "(UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_REMOTE * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_read_remote", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the local CPU socket.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_ITOM_LOCAL + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_LOCAL ) * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the local CPU socket", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_ITOM_LOCAL + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_LOCAL) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write_local", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to a remote CPU socket.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_ITOM_REMOTE + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_REMOTE ) * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to a remote CPU socket", + "MetricExpr": "((UNC_CHA_TOR_INSERTS.IO_ITOM_REMOTE + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_REMOTE) * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "io_bandwidth_write_remote", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to local memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.READS_LOCAL * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to local memory", + "MetricExpr": "(UNC_CHA_REQUESTS.READS_LOCAL * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_local_memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to local memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.WRITES_LOCAL * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to local memory", + "MetricExpr": "(UNC_CHA_REQUESTS.WRITES_LOCAL * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_local_memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to remote memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.READS_REMOTE * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to remote memory", + "MetricExpr": "(UNC_CHA_REQUESTS.READS_REMOTE * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_remote_memory_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to remote memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.WRITES_REMOTE * 64 / 1000000) / duration_time", + "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to remote memory", + "MetricExpr": "(UNC_CHA_REQUESTS.WRITES_REMOTE * 64 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "llc_miss_remote_memory_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data receive bandwidth (MB/sec).", - "MetricExpr": "( UNC_UPI_RxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", + "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data receive bandwidth (MB/sec)", + "MetricExpr": "(UNC_UPI_RxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", "MetricGroup": "", "MetricName": "upi_data_receive_bw", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth observed by the integrated I/O traffic controller (IIO) of IO reads that are initiated by end device controllers that are requesting memory from the CPU.", - "MetricExpr": "( UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.ALL_PARTS * 4 / 1000000) / duration_time", + "BriefDescription": "Bandwidth observed by the integrated I/O traffic controller (IIO) of IO reads that are initiated by end device controllers that are requesting memory from the CPU", + "MetricExpr": "(UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.ALL_PARTS * 4 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "iio_bandwidth_read", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Bandwidth observed by the integrated I/O traffic controller (IIO) of IO writes that are initiated by end device controllers that are writing memory to the CPU.", - "MetricExpr": "( UNC_IIO_DATA_REQ_OF_CPU.MEM_WRITE.ALL_PARTS * 4 / 1000000) / duration_time", + "BriefDescription": "Bandwidth observed by the integrated I/O traffic controller (IIO) of IO writes that are initiated by end device controllers that are writing memory to the CPU", + "MetricExpr": "(UNC_IIO_DATA_REQ_OF_CPU.MEM_WRITE.ALL_PARTS * 4 / 1000000) / duration_time", "MetricGroup": "", "MetricName": "iio_bandwidth_write", "ScaleUnit": "1MB/s" }, { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", + "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", + "MetricThreshold": "tma_bottleneck_mispredictions > 20", + "PublicDescription": "Total pipeline cost of Branch Misprediction related bottlenecks. Related metrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", + "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_big_code", + "MetricThreshold": "tma_bottleneck_big_code > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", + "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_ms ) ) ) ) - tma_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_instruction_fetch_bw", + "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) )", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", + "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", + "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks. Related metrics: tma_l3_hit_latency, tma_mem_latency.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", + "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs)", "MetricExpr": "100 * ( tma_memory_bound * ( tma_l1_bound / max( tma_memory_bound , ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) ) * ( tma_dtlb_load / max( tma_l1_bound , ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", + "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", + "PublicDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs). Related metrics: tma_dtlb_load, tma_dtlb_store.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", + "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors)", "MetricExpr": "100 * ( tma_memory_bound * ( ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) * tma_remote_cache / ( tma_local_mem + tma_remote_mem + tma_remote_cache ) + ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", - "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", + "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", + "PublicDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears, tma_remote_cache.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", + "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation", "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( tma_core_bound * tma_amx_busy / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", + "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", "PublicDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", + "BriefDescription": "Total pipeline cost of irregular execution (e.g", "MetricExpr": "100 * ( ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_ms ) ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + RS.EMPTY_RESOURCE / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", - "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments).", + "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", + "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments). Related metrics: tma_microcode_sequencer, tma_ms_switches.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end.", + "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end", "MetricExpr": "100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", "MetricGroup": "BvOB;Cor;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_other_bottlenecks", + "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", "PublicDescription": "Total pipeline cost of remaining bottlenecks in the back-end. Examples include data-dependencies (Core Bound when Low ILP) and other unlisted memory-related stalls.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", + "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA", "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_branching_overhead", + "MetricThreshold": "tma_bottleneck_branching_overhead > 5", "PublicDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound).", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", + "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead", "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_useful_work", + "MetricThreshold": "tma_bottleneck_useful_work > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "topdown\\-fe\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_frontend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_frontend_bound > 0.15", "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound. Sample with: FRONTEND_RETIRED.LATENCY_GE_4.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( topdown\\-fetch\\-lat / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots )", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", "MetricName": "tma_fetch_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_16, FRONTEND_RETIRED.LATENCY_GE_8.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache", "MetricExpr": "max( 0 , tma_icache_misses - tma_code_l2_miss )", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_CODE_RD / tma_info_thread_clks", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_itlb_misses - tma_code_stlb_miss )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + tma_unknown_branches", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "( tma_branch_mispredicts / tma_bad_speculation ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "( 1 - ( tma_branch_mispredicts / tma_bad_speculation ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 3 ) * cpu@UOPS_RETIRED.MS\\,cmask\\=0x1\\,edge\\=0x1@ / ( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) / tma_info_thread_clks", - "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated", + "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: FRONTEND_RETIRED.MS_FLOWS." + "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: FRONTEND_RETIRED.MS_FLOWS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "DECODE.LCP / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." + "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", - "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS." + "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "max( 0 , tma_frontend_bound - tma_fetch_latency )", - "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", + "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2.", + "MetricThreshold": "tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.MITE_CYCLES_ANY - IDQ.MITE_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", "ScaleUnit": "100%", + "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder.", + "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder", "MetricExpr": "( cpu@INST_DECODED.DECODERS\\,cmask\\=0x1@ - cpu@INST_DECODED.DECODERS\\,cmask\\=0x2@ ) / tma_info_core_core_clks / 2", - "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated", + "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated;tma_issueD0", "MetricName": "tma_decoder0_alone", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder. Related metrics: tma_few_uops_instructions." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.DSB_CYCLES_ANY - IDQ.DSB_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", "ScaleUnit": "100%", + "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details", "MetricExpr": "max( IDQ.MS_CYCLES_ANY , cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ / ( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) ) / tma_info_core_core_clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_ms", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "max( 1 - ( tma_frontend_bound + tma_backend_bound + tma_retiring ) , 0 )", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_bad_speculation", "ScaleUnit": "100%", + "MetricThreshold": "tma_bad_speculation > 0.15", "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( topdown\\-br\\-mispredict / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS.", + "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", + "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types)", "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Slots", "MetricName": "tma_other_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "max( 0 , tma_bad_speculation - tma_branch_mispredicts )", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots", + "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT.", + "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches, tma_remote_cache.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering", "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_machine_clears_group;Slots", "MetricName": "tma_other_nukes", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "( topdown\\-be\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_backend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_backend_bound > 0.2", "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( topdown\\-mem\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", "MetricName": "tma_memory_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two).", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( EXE_ACTIVITY.BOUND_ON_LOADS - MEMORY_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", + "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT." + "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "min( ( 7 ) * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - MEMORY_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS." + "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", "MetricName": "tma_load_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", "MetricName": "tma_load_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", + "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache", "MetricExpr": "min( 2 * ( MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETIRED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS ) * 20 / 100 , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - MEMORY_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_l1_latency_dependency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( 16 * max( 0 , MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS.ALL_RFO ) + ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) * ( ( 10 ) * L2_RQSTS.RFO_HIT + ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) ) / tma_info_thread_clks", - "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks", + "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS." + "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "L1D_PEND_MISS.FB_FULL / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", + "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory)." + "MetricThreshold": "tma_fb_full > 0.3", + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L1D_MISS - MEMORY_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", + "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited)", "MetricExpr": "( 4.4 * tma_info_system_core_frequency ) * MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", "MetricName": "tma_l2_hit_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L2_MISS - MEMORY_ACTIVITY.STALLS_L3_MISS ) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( ( 81 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) + ( ( 79 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS." + "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( ( 79 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD + MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 1 - ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD." + "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( ( 37 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_RETIRED.L3_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT." + "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_bottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "( XQ.FULL_CYCLES + L1D_PEND_MISS.L2_STALLS ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clks )", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that)." + "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to memory bandwidth Allocation feature (RDT's memory bandwidth throttling).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to memory bandwidth Allocation feature (RDT's memory bandwidth throttling)", "MetricExpr": "INT_MISC.MBA_STALLS / tma_info_thread_clks", "MetricGroup": "MemoryBW;Offcore;Server;TopdownL5;tma_L5_group;tma_mem_bandwidth_group;Clocks", "MetricName": "tma_mba_stalls", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_mba_stalls > 0.1 & tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that)." + "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory", "MetricExpr": "( ( 109 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_local_mem", "ScaleUnit": "100%", + "MetricThreshold": "tma_local_mem > 0.1 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory. Caching will improve the latency and increase performance. Sample with: MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory", "MetricExpr": "( ( 190 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", "MetricName": "tma_remote_mem", "ScaleUnit": "100%", + "MetricThreshold": "tma_remote_mem > 0.1 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues", "MetricExpr": "( ( ( 170 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM + ( ( 170 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", + "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_remote_cache", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD." + "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( MEM_STORE_RETIRED.L2_HIT * ( 10 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)." + "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( ( 170 * tma_info_system_core_frequency ) * cpu@OCR.DEMAND_RFO.L3_MISS\\,offcore_rsp\\=0x103b800002@ + ( 81 * tma_info_system_core_frequency ) * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM." + "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cache." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization", + "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores.", + "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores", "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread_clks", - "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSmSt", "MetricName": "tma_streaming_stores", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE." + "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma_fb_full." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( ( 7 ) * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_STORE_MISSES.WALK_ACTIVE ) / tma_info_core_core_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES." + "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_load." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", "MetricName": "tma_store_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", "MetricName": "tma_store_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "max( 0 , tma_backend_bound - tma_memory_bound )", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", "MetricName": "tma_core_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations).", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.DIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", "ScaleUnit": "100%", + "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIVIDER_ACTIVE." }, { - "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active", "MetricExpr": "ARITH.FPDIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_fp_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_int_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations", "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks + tma_c02_wait", - "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Clocks;tma_issueSO", "MetricName": "tma_serializing_operation", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD." + "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions", "MetricExpr": "CPU_CLK_UNHALTED.PAUSE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_slow_pause", "ScaleUnit": "100%", + "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.PAUSE_INST." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings).", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings)", "MetricExpr": "CPU_CLK_UNHALTED.C01 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_c01_wait", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_c01_wait > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings).", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings)", "MetricExpr": "CPU_CLK_UNHALTED.C02 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_c02_wait", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_c02_wait > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions", "MetricExpr": "13 * MISC2_RETIRED.LFENCE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_memory_fence", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_fence > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates fraction of cycles where the Advanced Matrix eXtensions (AMX) execution engine was busy with tile (arithmetic) operations.", + "BriefDescription": "This metric estimates fraction of cycles where the Advanced Matrix eXtensions (AMX) execution engine was busy with tile (arithmetic) operations", "MetricExpr": "EXE.AMX_BUSY / tma_info_core_core_clks", "MetricGroup": "BvCB;Compute;HPC;Server;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Core_Clocks", "MetricName": "tma_amx_busy", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_amx_busy > 0.5 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( tma_ports_utilized_0 * tma_info_thread_clks + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIV_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - EXE_ACTIVITY.BOUND_ON_LOADS ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "( EXE_ACTIVITY.EXE_BOUND_0_PORTS + max( RS.EMPTY_RESOURCE - RESOURCE_STALLS.SCOREBOARD , 0 ) ) / tma_info_thread_clks * ( CYCLE_ACTIVITY.STALLS_TOTAL - EXE_ACTIVITY.BOUND_ON_LOADS ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", + "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles)", "MetricExpr": "160 * ASSISTS.SSE_AVX_MIX / tma_info_thread_clks", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks;tma_issueMV", "MetricName": "tma_mixing_vectors", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic." + "MetricThreshold": "tma_mixing_vectors > 0.05", + "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Sample with: UOPS_EXECUTED.CYCLES_GE_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + UOPS_DISPATCHED.PORT_5_11 + UOPS_DISPATCHED.PORT_6 ) / ( 5 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED.PORT_0 / tma_info_core_core_clks", - "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0." + "MetricThreshold": "tma_port_0 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_1 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1." + "MetricThreshold": "tma_port_1 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_6 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1." + "MetricThreshold": "tma_port_6 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "UOPS_DISPATCHED.PORT_2_3_10 / ( 3 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_load_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations. Sample with: UOPS_DISPATCHED.PORT_2_3_10." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_4_9 + UOPS_DISPATCHED.PORT_7_8 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations. Sample with: UOPS_DISPATCHED.PORT_7_8." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( topdown\\-retiring / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_retiring", "ScaleUnit": "100%", + "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.1", "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "max( 0 , tma_retiring - tma_heavy_operations )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", "MetricName": "tma_light_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", "ScaleUnit": "100%", + "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR + FP_ARITH_INST_RETIRED2.VECTOR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.128B_PACKED_HALF ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.256B_PACKED_HALF ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.512B_PACKED_HALF ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_512b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired)", "MetricExpr": "tma_int_vector_128b + tma_int_vector_256b", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_int_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_int_operations > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired). Vector/Matrix Int operations and shuffles are counted. Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain." }, { - "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", + "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired", "MetricExpr": "( INT_VEC_RETIRED.ADD_128 + INT_VEC_RETIRED.VNNI_128 ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops", + "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops;tma_issue2P", "MetricName": "tma_int_vector_128b", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_vector_128b > 0.1 & tma_int_operations > 0.1 & tma_light_operations > 0.6", + "PublicDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", + "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired", "MetricExpr": "( INT_VEC_RETIRED.ADD_256 + INT_VEC_RETIRED.MUL_256 + INT_VEC_RETIRED.VNNI_256 ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops", + "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops;tma_issue2P", "MetricName": "tma_int_vector_256b", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_vector_256b > 0.1 & tma_int_operations > 0.1 & tma_light_operations > 0.6", + "PublicDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses", "MetricExpr": "tma_light_operations * MEM_UOP_RETIRED.ANY / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_memory_operations", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_operations > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.MACRO_FUSED / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_fused_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Other_Light_Ops in MTL!)}." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused", "MetricExpr": "tma_light_operations * ( BR_INST_RETIRED.ALL_BRANCHES - INST_RETIRED.MACRO_FUSED ) / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_non_fused_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused. Non-conditional branches like direct JMP or CALL would count here. Can be used to examine fusible conditional jumps that were not fused." }, { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", + "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes", "MetricExpr": "max( 0 , tma_light_operations - ( tma_fp_arith + tma_int_operations + tma_memory_operations + tma_fused_instructions + tma_non_fused_branches ) )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_other_light_ops", "ScaleUnit": "100%", + "MetricThreshold": "tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_light_operations_group;Slots", "MetricName": "tma_nop_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer)", "MetricExpr": "tma_light_operations * INT_VEC_RETIRED.SHUFFLES / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Slots", "MetricName": "tma_shuffles_256b", "ScaleUnit": "100%", + "MetricThreshold": "tma_shuffles_256b > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer). Shuffles may incur slow cross \"vector lane\" data transfers." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "( topdown\\-heavy\\-ops / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", "MetricName": "tma_heavy_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+]). Sample with: UOPS_RETIRED.HEAVY.", "DefaultMetricgroupName": "TopdownL2", "MetricGroupnoGroup": "TopdownL2;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops", "MetricExpr": "max( 0 , tma_heavy_operations - tma_microcode_sequencer )", - "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueD0", "MetricName": "tma_few_uops_instructions", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions." + "MetricThreshold": "tma_few_uops_instructions > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions. Related metrics: tma_decoder0_alone." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "UOPS_RETIRED.MS / tma_info_thread_slots", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: UOPS_RETIRED.MS." + "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: UOPS_RETIRED.MS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( ( 99 *3 + 63 + 30 ) / 5 ) * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults", "MetricExpr": "99 * ASSISTS.PAGE_FAULT / tma_info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_page_faults", "ScaleUnit": "100%", + "MetricThreshold": "tma_page_faults > 0.05", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults. A Page Fault may apply on first application access to a memory page. Note operating system handling of page faults accounts for the majority of its cost." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists", "MetricExpr": "30 * ASSISTS.FP / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_fp_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_assists > 0.1", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists", "MetricExpr": "63 * ASSISTS.SSE_AVX_MIX / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_avx_assists", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_avx_assists > 0.1" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", "ScaleUnit": "100%", + "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources. Sample with: FRONTEND_RETIRED.MS_FLOWS." }, { - "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts.", + "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts", "MetricExpr": "100 * ( 1 - tma_core_bound / tma_ports_utilization if tma_core_bound < tma_ports_utilization else 1 ) if tma_info_system_smt_2t_utilization > 0.5 else 0", "MetricGroup": "Cor;SMT;Metric", "MetricName": "tma_info_botlnk_l0_core_bound_likely", "MetricThreshold": "tma_info_botlnk_core_bound_likely > 0.5" }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", "MetricName": "tma_info_thread_uoppi", "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", "MetricName": "tma_info_thread_uptb", "MetricThreshold": "tma_info_thread_uptb < 6 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor.", + "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor", "MetricExpr": "tma_info_thread_slots / ( TOPDOWN.SLOTS / 2 ) if #SMT_on else 1", "MetricGroup": "SMT;TmaL1;TopdownL1;tma_L1_group;Metric", "MetricName": "tma_info_thread_slots_utilization" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR_HALF ) + 2 * ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED2.COMPLEX_SCALAR_HALF ) + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * ( FP_ARITH_INST_RETIRED2.128B_PACKED_HALF + FP_ARITH_INST_RETIRED.8_FLOPS ) + 16 * ( FP_ARITH_INST_RETIRED2.256B_PACKED_HALF + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) + 32 * FP_ARITH_INST_RETIRED2.512B_PACKED_HALF ) ) / tma_info_core_core_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( FP_ARITH_DISPATCHED.PORT_0 + FP_ARITH_DISPATCHED.PORT_1 + FP_ARITH_DISPATCHED.PORT_5 ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "uops Executed per Cycle.", + "BriefDescription": "uops Executed per Cycle", "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", "MetricGroup": "Power;Metric", "MetricName": "tma_info_core_epc" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipload", "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipstore", "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipbranch", "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", "MetricName": "tma_info_inst_mix_ipcall", "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", "MetricThreshold": "tma_info_inst_mix_iptb < 6 * 2 + 1", - "PublicDescription": "Instructions per taken branch. Related metrics: tma_info_frontend_dsb_coverage, tma_info_botlnk_l2_dsb_misses, tma_info_botlnk_l2_dsb_bandwidth." + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR_HALF ) + 2 * ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED2.COMPLEX_SCALAR_HALF ) + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * ( FP_ARITH_INST_RETIRED2.128B_PACKED_HALF + FP_ARITH_INST_RETIRED.8_FLOPS ) + 16 * ( FP_ARITH_INST_RETIRED2.256B_PACKED_HALF + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) + 32 * FP_ARITH_INST_RETIRED2.512B_PACKED_HALF ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ipflop", "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR + FP_ARITH_INST_RETIRED2.VECTOR ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", @@ -1539,7 +1674,7 @@ "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Half-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Half-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED2.SCALAR", "MetricGroup": "Flops;FpScalar;InsType;Server;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_hp", @@ -1547,7 +1682,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Half-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", @@ -1555,7 +1690,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", @@ -1563,7 +1698,7 @@ "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.128B_PACKED_HALF )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", @@ -1571,7 +1706,7 @@ "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.256B_PACKED_HALF )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", @@ -1579,7 +1714,7 @@ "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.512B_PACKED_HALF )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx512", @@ -1587,40 +1722,40 @@ "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / CPU_CLK_UNHALTED.PAUSE_INST", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ippause" }, { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / SW_PREFETCH_ACCESS.ANY", "MetricGroup": "Prefetches;Inst_Metric", "MetricName": "tma_info_inst_mix_ipswpf", "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions.", + "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions", "MetricExpr": "INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "MicroSeq;Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_strings_cycles", "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1" }, { - "BriefDescription": "Instructions per a microcode Assist invocation.", + "BriefDescription": "Instructions per a microcode Assist invocation", "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", "MetricName": "tma_info_pipeline_ipassist", @@ -1628,105 +1763,105 @@ "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core.", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core", "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@ )", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", + "BriefDescription": "Average number of uops fetched from DSB per cycle", "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_dsb" }, { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", + "BriefDescription": "Average number of uops fetched from MITE per cycle", "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_mite" }, { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", + "BriefDescription": "Average number of Uops issued by front-end when it issued something", "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_frontend_fetch_upc" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( UOPS_ISSUED.ANY )", "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", "MetricName": "tma_info_frontend_dsb_coverage", "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 6 > 0.35", - "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_info_inst_mix_iptb, tma_info_botlnk_l2_dsb_misses, tma_info_botlnk_l2_dsb_bandwidth." + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection.", + "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection", "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / cpu@INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_unknown_branch_cost", "PublicDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection. See Unknown_Branches node." }, { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", + "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "DSBmiss;Metric", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Average Latency for L1 instruction cache misses.", + "BriefDescription": "Average Latency for L1 instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / cpu@ICACHE_DATA.STALLS\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "Fed;FetchLat;IcMiss;Metric", "MetricName": "tma_info_frontend_icache_miss_latency" }, { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", "MetricGroup": "DSBmiss;Fed;Inst_Metric", "MetricName": "tma_info_frontend_ipdsb_miss_ret", "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache true code cacheline misses per kilo instruction", "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code" }, { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction", "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code_all" }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb + tma_ms ) )", "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_dsb_misses > 10", - "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_info_inst_mix_iptb, tma_info_frontend_dsb_coverage, tma_info_botlnk_l2_dsb_bandwidth." + "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb + tma_ms ) ) )", "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_dsb_bandwidth > 10", - "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_info_inst_mix_iptb, tma_info_frontend_dsb_coverage, tma_info_botlnk_l2_dsb_misses." + "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", + "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", @@ -1734,462 +1869,463 @@ "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmispredict", "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_ret", "MetricThreshold": "tma_info_bad_spec_ipmisp_ret < 500" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", "MetricName": "tma_info_bad_spec_ipmisp_indirect", "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 6 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", "MetricName": "tma_info_bad_spec_branch_misprediction_cost", - "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)." + "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_mispredicts_resteers." }, { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", + "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", "MetricGroup": "BrMispredicts;Metric", "MetricName": "tma_info_bad_spec_spec_clears_ratio" }, { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", + "BriefDescription": "Fraction of branches that are non-taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_nt" }, { - "BriefDescription": "Fraction of branches that are taken conditionals.", + "BriefDescription": "Fraction of branches that are taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_tk" }, { - "BriefDescription": "Fraction of branches that are CALL or RET.", + "BriefDescription": "Fraction of branches that are CALL or RET", "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_callret" }, { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", + "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_jump" }, { - "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group).", + "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group)", "MetricExpr": "1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_other_branches" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / MEM_LOAD_COMPLETED.L1_MISS_ANY", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki_load" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * L2_RQSTS.RFO_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_all" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", + "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_fb_hpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Offcore;Metric", "MetricName": "tma_info_memory_l3_cache_access_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", "MetricName": "tma_info_memory_tlb_page_walks_utilization", "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Fed;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_code_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_load_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_access_bw", "MetricGroup": "Mem;MemoryBW;Offcore;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_access_bw_2t" }, { - "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory).", + "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory)", "MetricExpr": "1000 * L2_LINES_OUT.SILENT / tma_info_inst_mix_instructions", "MetricGroup": "L2Evicts;Mem;Server;Core_Metric", "MetricName": "tma_info_memory_core_l2_evictions_silent_pki" }, { - "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction.", + "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction", "MetricExpr": "1000 * L2_LINES_OUT.NON_SILENT / tma_info_inst_mix_instructions", "MetricGroup": "L2Evicts;Mem;Server;Core_Metric", "MetricName": "tma_info_memory_core_l2_evictions_nonsilent_pki" }, { - "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses.", + "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses", "MetricExpr": "L2_LINES_OUT.USELESS_HWPF / ( L2_LINES_OUT.SILENT + L2_LINES_OUT.NON_SILENT )", "MetricGroup": "Prefetches;Metric", - "MetricName": "tma_info_memory_prefetches_useless_hwpf" + "MetricName": "tma_info_memory_prefetches_useless_hwpf", + "MetricThreshold": "tma_info_memory_useless_hwpf > 0.15" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Latency for L3 cache miss demand Loads.", + "BriefDescription": "Average Latency for L3 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.L3_MISS_DEMAND_DATA_RD / OFFCORE_REQUESTS.L3_MISS_DEMAND_DATA_RD", "MetricGroup": "Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l3_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=0x1@", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Off-core accesses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches).", + "BriefDescription": "Off-core accesses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches)", "MetricExpr": "1000 * OCR.READS_TO_CORE.ANY_RESPONSE / tma_info_inst_mix_instructions", "MetricGroup": "CacheHits;Offcore;Metric", "MetricName": "tma_info_memory_mix_offcore_read_any_pki" }, { - "BriefDescription": "L3 cache misses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches).", + "BriefDescription": "L3 cache misses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches)", "MetricExpr": "1000 * OCR.READS_TO_CORE.L3_MISS / tma_info_inst_mix_instructions", "MetricGroup": "Offcore;Metric", "MetricName": "tma_info_memory_mix_offcore_read_l3m_pki" }, { - "BriefDescription": "Off-core accesses per kilo instruction for modified write requests.", + "BriefDescription": "Off-core accesses per kilo instruction for modified write requests", "MetricExpr": "1000 * OCR.MODIFIED_WRITE.ANY_RESPONSE / tma_info_inst_mix_instructions", "MetricGroup": "Offcore;Metric", "MetricName": "tma_info_memory_mix_offcore_mwrite_any_pki" }, { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", + "BriefDescription": "Un-cacheable retired load per kilo instruction", "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_uc_load_pki" }, { - "BriefDescription": "\"Bus lock\" per kilo instruction.", + "BriefDescription": "\"Bus lock\" per kilo instruction", "MetricExpr": "1000 * SQ_MISC.BUS_LOCK / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_bus_lock_pki" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz].", + "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz]", "MetricExpr": "tma_info_system_socket_clks / 1e9 / tma_info_system_time", "MetricGroup": "SoC;System_Metric", "MetricName": "tma_info_system_uncore_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR_HALF ) + 2 * ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED2.COMPLEX_SCALAR_HALF ) + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * ( FP_ARITH_INST_RETIRED2.128B_PACKED_HALF + FP_ARITH_INST_RETIRED.8_FLOPS ) + 16 * ( FP_ARITH_INST_RETIRED2.256B_PACKED_HALF + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) + 32 * FP_ARITH_INST_RETIRED2.512B_PACKED_HALF ) ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_utilization", "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states.", + "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states", "MetricExpr": "CPU_CLK_UNHALTED.C0_WAIT / tma_info_thread_clks", "MetricGroup": "C0Wait;Metric", "MetricName": "tma_info_system_c0_wait", "MetricThreshold": "tma_info_system_c0_wait > 0.05" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "( 64 * ( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", "MetricName": "tma_info_system_dram_bw_use", - "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]." + "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Average latency of data read request to external memory (in nanoseconds).", + "BriefDescription": "Average latency of data read request to external memory (in nanoseconds)", "MetricExpr": "( 1000000000 ) * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( tma_info_system_socket_clks / tma_info_system_time )", "MetricGroup": "Mem;MemoryLat;SoC;NanoSeconds", "MetricName": "tma_info_system_mem_read_latency", "PublicDescription": "Average latency of data read request to external memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetches. ([RKL+]memory-controller only)." }, { - "BriefDescription": "Average number of parallel data read requests to external memory.", + "BriefDescription": "Average number of parallel data read requests to external memory", "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD\\,thresh\\=0x1@", "MetricGroup": "Mem;MemoryBW;SoC;System_Metric", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches." }, { - "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds].", + "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]", "MetricExpr": "( 1000000000 ) * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR ) / cha_0@event\\=0x0@", "MetricGroup": "MemOffcore;MemoryLat;SoC;Server;NanoSeconds", "MetricName": "tma_info_system_mem_dram_read_latency", "PublicDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]. Accounts for demand loads and L1/L2 data-read prefetches." }, { - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec].", + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]", "MetricExpr": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "IoBW;MemOffcore;SoC;Server;GB/sec", "MetricName": "tma_info_system_io_read_bw", "PublicDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]. Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU." }, { - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec].", + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]", "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_ITOM + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR ) * 64 / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "IoBW;MemOffcore;SoC;Server;GB/sec", "MetricName": "tma_info_system_io_write_bw", "PublicDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]. Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU." }, { - "BriefDescription": "Cross-socket Ultra Path Interconnect (UPI) data transmit bandwidth for data only [MB / sec].", - "MetricExpr": "UNC_UPI_TxL_FLITS.ALL_DATA * 64 / 9 / 1000000", + "BriefDescription": "Cross-socket Ultra Path Interconnect (UPI) data transmit bandwidth for data only [MB / sec]", + "MetricExpr": "UNC_UPI_TXL_FLITS.ALL_DATA * 64 / 9 / 1000000", "MetricGroup": "SoC;Server;MB/sec", "MetricName": "tma_info_system_upi_data_transmit_bw" }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "( power@energy\\-pkg@ * ( 61 ) + 15.6 * power@energy\\-ram@ ) / ( ( duration_time ) * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", "MetricName": "tma_info_system_time", "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", "MetricName": "tma_info_system_mux", "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "cha_0@event\\=0x0@", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", "MetricName": "tma_info_system_ipfarbranch", "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" }, { - "BriefDescription": "Average Off-core access BW for Reads-to-Core (R2C).", + "BriefDescription": "Average Off-core access BW for Reads-to-Core (R2C)", "MetricExpr": "64 * OCR.READS_TO_CORE.ANY_RESPONSE / 1e9 / tma_info_system_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC;GB/sec", "MetricName": "tma_info_memory_soc_r2c_offcore_bw", "PublicDescription": "Average Off-core access BW for Reads-to-Core (R2C). R2C account for demand or prefetch load/RFO/code access that fill data into the Core caches." }, { - "BriefDescription": "Average L3-cache miss BW for Reads-to-Core (R2C).", + "BriefDescription": "Average L3-cache miss BW for Reads-to-Core (R2C)", "MetricExpr": "64 * OCR.READS_TO_CORE.L3_MISS / 1e9 / tma_info_system_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC;GB/sec", "MetricName": "tma_info_memory_soc_r2c_l3m_bw", "PublicDescription": "Average L3-cache miss BW for Reads-to-Core (R2C). This covering going to DRAM or other memory off-chip memory tears. See R2C_Offcore_BW." }, { - "BriefDescription": "Average DRAM BW for Reads-to-Core (R2C) covering for memory attached to local- and remote-socket.", + "BriefDescription": "Average DRAM BW for Reads-to-Core (R2C) covering for memory attached to local- and remote-socket", "MetricExpr": "64 * OCR.READS_TO_CORE.DRAM / 1e9 / tma_info_system_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC;GB/sec", "MetricName": "tma_info_memory_soc_r2c_dram_bw", diff --git a/SPR/metrics/perf/sapphirerapidshbm_metrics_perf.json b/SPR/metrics/perf/sapphirerapidshbm_metrics_perf.json deleted file mode 100644 index 980f995d..00000000 --- a/SPR/metrics/perf/sapphirerapidshbm_metrics_perf.json +++ /dev/null @@ -1,2229 +0,0 @@ -[ - { - "BriefDescription": "CPU operating frequency (in GHz).", - "MetricExpr": "( CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC * #SYSTEM_TSC_FREQ ) / 1000000000", - "MetricGroup": "", - "MetricName": "cpu_operating_frequency", - "ScaleUnit": "1GHz" - }, - { - "BriefDescription": "Percentage of time spent in the active CPU power state C0.", - "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", - "MetricGroup": "", - "MetricName": "cpu_utilization", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles.", - "MetricExpr": "CPU_CLK_UNHALTED.THREAD / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "cpi", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions.", - "MetricExpr": "MEM_INST_RETIRED.ALL_LOADS / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "loads_per_instr", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions.", - "MetricExpr": "MEM_INST_RETIRED.ALL_STORES / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "stores_per_instr", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions.", - "MetricExpr": "L1D.REPLACEMENT / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "l1d_mpi", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions.", - "MetricExpr": "MEM_LOAD_RETIRED.L1_HIT / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "l1d_demand_data_read_hits_per_instr", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions.", - "MetricExpr": "L2_RQSTS.ALL_CODE_RD / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "l1_i_code_read_misses_with_prefetches_per_instr", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions.", - "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "l2_demand_data_read_hits_per_instr", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions.", - "MetricExpr": "L2_LINES_IN.ALL / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "l2_mpi", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions.", - "MetricExpr": "MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "l2_demand_data_read_mpi", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions.", - "MetricExpr": "L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "l2_demand_code_mpi", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_LLCPREFDATA + UNC_CHA_TOR_INSERTS.IA_MISS_DRD + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF ) / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "llc_data_read_mpi_demand_plus_prefetch", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions.", - "MetricExpr": "UNC_CHA_TOR_INSERTS.IA_MISS_CRD / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "llc_code_read_mpi_demand_plus_prefetch", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD) * #num_packages ) ) ) * duration_time", - "MetricGroup": "", - "MetricName": "llc_demand_data_read_miss_latency", - "ScaleUnit": "1ns" - }, - { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to local memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL) * #num_packages ) ) ) * duration_time", - "MetricGroup": "", - "MetricName": "llc_demand_data_read_miss_latency_for_local_requests", - "ScaleUnit": "1ns" - }, - { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to remote memory in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE) * #num_packages ) ) ) * duration_time", - "MetricGroup": "", - "MetricName": "llc_demand_data_read_miss_latency_for_remote_requests", - "ScaleUnit": "1ns" - }, - { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to Intel(R) Optane(TM) Persistent Memory(PMEM) in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PMM ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM) * #num_packages ) ) ) * duration_time", - "MetricGroup": "", - "MetricName": "llc_demand_data_read_miss_to_pmem_latency", - "ScaleUnit": "1ns" - }, - { - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to DRAM in nano seconds.", - "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR ) / ( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR) * #num_packages ) ) ) * duration_time", - "MetricGroup": "", - "MetricName": "llc_demand_data_read_miss_to_dram_latency", - "ScaleUnit": "1ns" - }, - { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions.", - "MetricExpr": "ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "itlb_2nd_level_mpi", - "ScaleUnit": "1per_instr", - "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the ITLB (Instruction TLB) and further levels of TLB." - }, - { - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions.", - "MetricExpr": "ITLB_MISSES.WALK_COMPLETED_2M_4M / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "itlb_2nd_level_large_page_mpi", - "ScaleUnit": "1per_instr", - "PublicDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the Instruction Translation Lookaside Buffer (ITLB) and further levels of TLB." - }, - { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions.", - "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "dtlb_2nd_level_load_mpi", - "ScaleUnit": "1per_instr", - "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." - }, - { - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions.", - "MetricExpr": "DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "dtlb_2nd_level_2mb_large_page_load_mpi", - "ScaleUnit": "1per_instr", - "PublicDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the Data Translation Lookaside Buffer (DTLB) and further levels of TLB." - }, - { - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions.", - "MetricExpr": "DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", - "MetricGroup": "", - "MetricName": "dtlb_2nd_level_store_mpi", - "ScaleUnit": "1per_instr", - "PublicDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB." - }, - { - "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL ) / ( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE )", - "MetricGroup": "", - "MetricName": "numa_reads_addressed_to_local_dram", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE ) / ( UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE + UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE )", - "MetricGroup": "", - "MetricName": "numa_reads_addressed_to_remote_dram", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Uncore operating frequency in GHz.", - "MetricExpr": "( UNC_CHA_CLOCKTICKS / ( source_count(UNC_CHA_CLOCKTICKS) * #num_packages ) / 1000000000) / duration_time", - "MetricGroup": "", - "MetricName": "uncore_frequency", - "ScaleUnit": "1GHz" - }, - { - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data transmit bandwidth (MB/sec).", - "MetricExpr": "( UNC_UPI_TxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "upi_data_transmit_bw", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "DDR memory read bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.RD * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "memory_bandwidth_read", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "DDR memory write bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_CAS_COUNT.WR * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "memory_bandwidth_write", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "DDR memory bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "memory_bandwidth_total", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Memory write bandwidth (MB/sec) caused by directory updates; includes DDR and Intel(R) Optane(TM) Persistent Memory(PMEM).", - "MetricExpr": "(( UNC_CHA_DIR_UPDATE.HA + UNC_CHA_DIR_UPDATE.TOR + UNC_M2M_DIRECTORY_UPDATE.ANY ) * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "memory_extra_write_bw_due_to_directory_updates", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory read bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_PMM_RPQ_INSERTS * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "pmem_memory_bandwidth_read", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory write bandwidth (MB/sec).", - "MetricExpr": "( UNC_M_PMM_WPQ_INSERTS * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "pmem_memory_bandwidth_write", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory bandwidth (MB/sec).", - "MetricExpr": "(( UNC_M_PMM_RPQ_INSERTS + UNC_M_PMM_WPQ_INSERTS ) * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "pmem_memory_bandwidth_total", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "io_bandwidth_read", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_ITOM + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR ) * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "io_bandwidth_write", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Percentage of inbound reads initiated by end device controllers that miss the L3 cache.", - "MetricExpr": "UNC_CHA_TOR_INSERTS.IO_MISS_PCIRDCUR / UNC_CHA_TOR_INSERTS.IO_PCIRDCUR", - "MetricGroup": "", - "MetricName": "io_percent_of_inbound_reads_that_miss_l3", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of inbound partial cacheline writes initiated by end device controllers that miss the L3 cache.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_MISS_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_MISS_RFO ) / ( UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_RFO ) )", - "MetricGroup": "", - "MetricName": "io_percent_of_inbound_partial_writes_that_miss_l3", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of inbound full cacheline writes initiated by end device controllers that miss the L3 cache.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_MISS_ITOM / UNC_CHA_TOR_INSERTS.IO_ITOM )", - "MetricGroup": "", - "MetricName": "io_percent_of_inbound_full_writes_that_miss_l3", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.DSB_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", - "MetricGroup": "", - "MetricName": "percent_uops_delivered_from_decoded_icache", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MITE_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", - "MetricGroup": "", - "MetricName": "percent_uops_delivered_from_legacy_decode_pipeline", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue.", - "MetricExpr": "( IDQ.MS_UOPS / ( IDQ.DSB_UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS + LSD.UOPS ) )", - "MetricGroup": "", - "MetricName": "percent_uops_delivered_from_microcode_sequencer", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the local CPU socket.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_LOCAL * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "io_bandwidth_read_local", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from a remote CPU socket.", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_REMOTE * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "io_bandwidth_read_remote", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the local CPU socket.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_ITOM_LOCAL + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_LOCAL ) * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "io_bandwidth_write_local", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to a remote CPU socket.", - "MetricExpr": "(( UNC_CHA_TOR_INSERTS.IO_ITOM_REMOTE + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_REMOTE ) * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "io_bandwidth_write_remote", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to local memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.READS_LOCAL * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "llc_miss_local_memory_bandwidth_read", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to local memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.WRITES_LOCAL * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "llc_miss_local_memory_bandwidth_write", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to remote memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.READS_REMOTE * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "llc_miss_remote_memory_bandwidth_read", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to remote memory.", - "MetricExpr": "( UNC_CHA_REQUESTS.WRITES_REMOTE * 64 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "llc_miss_remote_memory_bandwidth_write", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data receive bandwidth (MB/sec).", - "MetricExpr": "( UNC_UPI_RxL_FLITS.ALL_DATA * (64 / 9.0) / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "upi_data_receive_bw", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Bandwidth observed by the integrated I/O traffic controller (IIO) of IO reads that are initiated by end device controllers that are requesting memory from the CPU.", - "MetricExpr": "( UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.ALL_PARTS * 4 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "iio_bandwidth_read", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Bandwidth observed by the integrated I/O traffic controller (IIO) of IO writes that are initiated by end device controllers that are writing memory to the CPU.", - "MetricExpr": "( UNC_IIO_DATA_REQ_OF_CPU.MEM_WRITE.ALL_PARTS * 4 / 1000000) / duration_time", - "MetricGroup": "", - "MetricName": "iio_bandwidth_write", - "ScaleUnit": "1MB/s" - }, - { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", - "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots", - "MetricName": "tma_bottleneck_mispredictions", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", - "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", - "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Default;Scaled_Slots", - "MetricName": "tma_bottleneck_big_code", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", - "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_ms ) ) ) ) - tma_big_code", - "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Default;Scaled_Slots", - "MetricName": "tma_bottleneck_instruction_fetch_bw", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", - "MetricExpr": "( 100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * tma_hbm_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) ) )", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBW", - "MetricName": "tma_bottleneck_cache_memory_bandwidth", - "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", - "PublicDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks. Related metrics: tma_info_system_dram_bw_use.", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", - "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * tma_hbm_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueLat", - "MetricName": "tma_bottleneck_cache_memory_latency", - "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", - "PublicDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", - "MetricExpr": "100 * ( tma_memory_bound * ( tma_l1_bound / max( tma_memory_bound , ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) ) * ( tma_dtlb_load / max( tma_l1_bound , ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueTLB", - "MetricName": "tma_bottleneck_memory_data_tlbs", - "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", - "PublicDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", - "MetricExpr": "100 * ( tma_memory_bound * ( ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) * tma_remote_cache / ( tma_local_mem + tma_remote_mem + tma_remote_cache ) + ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", - "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueSyncxn", - "MetricName": "tma_bottleneck_memory_synchronization", - "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", - "PublicDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", - "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( tma_core_bound * tma_amx_busy / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots", - "MetricName": "tma_bottleneck_compute_bound_est", - "PublicDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy.", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", - "MetricExpr": "100 * ( ( ( 1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_ms ) ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + RS.EMPTY_RESOURCE / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", - "MetricName": "tma_bottleneck_irregular_overhead", - "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments).", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end.", - "MetricExpr": "100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", - "MetricGroup": "BvOB;Cor;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", - "MetricName": "tma_bottleneck_other_bottlenecks", - "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", - "PublicDescription": "Total pipeline cost of remaining bottlenecks in the back-end. Examples include data-dependencies (Core Bound when Low ILP) and other unlisted memory-related stalls.", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", - "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", - "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", - "MetricName": "tma_bottleneck_branching_overhead", - "PublicDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound).", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", - "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", - "MetricName": "tma_bottleneck_useful_work", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", - "MetricExpr": "topdown\\-fe\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots", - "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Default;Slots", - "MetricName": "tma_frontend_bound", - "ScaleUnit": "100%", - "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound. Sample with: FRONTEND_RETIRED.LATENCY_GE_4.", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", - "MetricExpr": "( topdown\\-fetch\\-lat / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots )", - "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", - "MetricName": "tma_fetch_latency", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_16, FRONTEND_RETIRED.LATENCY_GE_8.", - "DefaultMetricgroupName": "TopdownL2", - "MetricGroupnoGroup": "TopdownL2;Default" - }, - { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", - "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", - "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", - "MetricName": "tma_icache_misses", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." - }, - { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache.", - "MetricExpr": "max( 0 , tma_icache_misses - tma_code_l2_miss )", - "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", - "MetricName": "tma_code_l2_hit", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache.", - "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_CODE_RD / tma_info_thread_clks", - "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", - "MetricName": "tma_code_l2_miss", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", - "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", - "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", - "MetricName": "tma_itlb_misses", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." - }, - { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", - "MetricExpr": "max( 0 , tma_itlb_misses - tma_code_stlb_miss )", - "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", - "MetricName": "tma_code_stlb_hit", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", - "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", - "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", - "MetricName": "tma_code_stlb_miss", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", - "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", - "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", - "MetricName": "tma_code_stlb_miss_4k", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", - "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", - "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", - "MetricName": "tma_code_stlb_miss_2m", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", - "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + tma_unknown_branches", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", - "MetricName": "tma_branch_resteers", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." - }, - { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", - "MetricExpr": "( tma_branch_mispredicts / tma_bad_speculation ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", - "MetricName": "tma_mispredicts_resteers", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." - }, - { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", - "MetricExpr": "( 1 - ( tma_branch_mispredicts / tma_bad_speculation ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", - "MetricName": "tma_clears_resteers", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." - }, - { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", - "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / tma_info_thread_clks", - "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", - "MetricName": "tma_unknown_branches", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH." - }, - { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", - "MetricExpr": "( 3 ) * cpu@UOPS_RETIRED.MS\\,cmask\\=0x1\\,edge\\=0x1@ / ( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) / tma_info_thread_clks", - "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated", - "MetricName": "tma_ms_switches", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: FRONTEND_RETIRED.MS_FLOWS." - }, - { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", - "MetricExpr": "DECODE.LCP / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", - "MetricName": "tma_lcp", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." - }, - { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", - "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", - "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", - "MetricName": "tma_dsb_switches", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS." - }, - { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", - "MetricExpr": "max( 0 , tma_frontend_bound - tma_fetch_latency )", - "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Default;Slots", - "MetricName": "tma_fetch_bandwidth", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2.", - "DefaultMetricgroupName": "TopdownL2", - "MetricGroupnoGroup": "TopdownL2;Default" - }, - { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", - "MetricExpr": "( IDQ.MITE_CYCLES_ANY - IDQ.MITE_CYCLES_OK ) / tma_info_core_core_clks / 2", - "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", - "MetricName": "tma_mite", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." - }, - { - "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder.", - "MetricExpr": "( cpu@INST_DECODED.DECODERS\\,cmask\\=0x1@ - cpu@INST_DECODED.DECODERS\\,cmask\\=0x2@ ) / tma_info_core_core_clks / 2", - "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated", - "MetricName": "tma_decoder0_alone", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", - "MetricExpr": "( IDQ.DSB_CYCLES_ANY - IDQ.DSB_CYCLES_OK ) / tma_info_core_core_clks / 2", - "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", - "MetricName": "tma_dsb", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." - }, - { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details.", - "MetricExpr": "max( IDQ.MS_CYCLES_ANY , cpu@UOPS_RETIRED.MS\\,cmask\\=0x1@ / ( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) ) / tma_info_core_core_clks / 2", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", - "MetricName": "tma_ms", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", - "MetricExpr": "max( 1 - ( tma_frontend_bound + tma_backend_bound + tma_retiring ) , 0 )", - "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Default;Slots", - "MetricName": "tma_bad_speculation", - "ScaleUnit": "100%", - "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", - "MetricExpr": "( topdown\\-br\\-mispredict / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots", - "MetricName": "tma_branch_mispredicts", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS.", - "DefaultMetricgroupName": "TopdownL2", - "MetricGroupnoGroup": "TopdownL2;Default" - }, - { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", - "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", - "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Slots", - "MetricName": "tma_other_mispredicts", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", - "MetricExpr": "max( 0 , tma_bad_speculation - tma_branch_mispredicts )", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Default;Slots", - "MetricName": "tma_machine_clears", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT.", - "DefaultMetricgroupName": "TopdownL2", - "MetricGroupnoGroup": "TopdownL2;Default" - }, - { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", - "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", - "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_machine_clears_group;Slots", - "MetricName": "tma_other_nukes", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", - "MetricExpr": "( topdown\\-be\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", - "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Default;Slots", - "MetricName": "tma_backend_bound", - "ScaleUnit": "100%", - "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS.", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", - "MetricExpr": "( topdown\\-mem\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", - "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", - "MetricName": "tma_memory_bound", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two).", - "DefaultMetricgroupName": "TopdownL2", - "MetricGroupnoGroup": "TopdownL2;Default" - }, - { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", - "MetricExpr": "max( ( EXE_ACTIVITY.BOUND_ON_LOADS - MEMORY_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", - "MetricName": "tma_l1_bound", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT." - }, - { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", - "MetricExpr": "min( ( 7 ) * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - MEMORY_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", - "MetricName": "tma_dtlb_load", - "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS." - }, - { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", - "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", - "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", - "MetricName": "tma_load_stlb_hit", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", - "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", - "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", - "MetricName": "tma_load_stlb_miss", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", - "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", - "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", - "MetricName": "tma_load_stlb_miss_4k", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", - "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", - "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", - "MetricName": "tma_load_stlb_miss_2m", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", - "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", - "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", - "MetricName": "tma_load_stlb_miss_1g", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", - "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", - "MetricName": "tma_store_fwd_blk", - "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." - }, - { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", - "MetricExpr": "min( 2 * ( MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETIRED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS ) * 20 / 100 , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - MEMORY_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", - "MetricName": "tma_l1_latency_dependency", - "ScaleUnit": "100%", - "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT." - }, - { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", - "MetricExpr": "( 16 * max( 0 , MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS.ALL_RFO ) + ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) * ( ( 10 ) * L2_RQSTS.RFO_HIT + ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) ) / tma_info_thread_clks", - "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks", - "MetricName": "tma_lock_latency", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS." - }, - { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", - "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", - "MetricName": "tma_split_loads", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." - }, - { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", - "MetricExpr": "L1D_PEND_MISS.FB_FULL / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", - "MetricName": "tma_fb_full", - "ScaleUnit": "100%", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory)." - }, - { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", - "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L1D_MISS - MEMORY_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks", - "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", - "MetricName": "tma_l2_bound", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." - }, - { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", - "MetricExpr": "( 4.4 * tma_info_system_core_frequency ) * MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", - "MetricName": "tma_l2_hit_latency", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." - }, - { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", - "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L2_MISS - MEMORY_ACTIVITY.STALLS_L3_MISS ) / tma_info_thread_clks", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", - "MetricName": "tma_l3_bound", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." - }, - { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", - "MetricExpr": "( ( ( 81 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) + ( ( 79 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", - "MetricName": "tma_contested_accesses", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS." - }, - { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", - "MetricExpr": "( ( 79 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD + MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 1 - ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", - "MetricName": "tma_data_sharing", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD." - }, - { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", - "MetricExpr": "( ( 37 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_RETIRED.L3_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", - "MetricName": "tma_l3_hit_latency", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT." - }, - { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", - "MetricExpr": "( XQ.FULL_CYCLES + L1D_PEND_MISS.L2_STALLS ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks", - "MetricName": "tma_sq_full", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates how often the CPU was stalled due to High Bandwidth Memory (HBM) accesses by loads.", - "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clks ) * OCR.DEMAND_DATA_RD.PMM / OCR.READS_TO_CORE.L3_MISS", - "MetricGroup": "BvMB;BvML;MemoryBound;Offcore;Server;TmaL3mem;TopdownL3;tma_L3_group;Stalls", - "MetricName": "tma_hbm_bound", - "ScaleUnit": "100%", - "MetricThreshold": "tma_hbm_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" - }, - { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", - "MetricExpr": "( MEMORY_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clks ) - tma_hbm_bound", - "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", - "MetricName": "tma_dram_bound", - "ScaleUnit": "100%", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." - }, - { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", - "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;Clocks", - "MetricName": "tma_mem_bandwidth", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that)." - }, - { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to memory bandwidth Allocation feature (RDT's memory bandwidth throttling).", - "MetricExpr": "INT_MISC.MBA_STALLS / tma_info_thread_clks", - "MetricGroup": "MemoryBW;Offcore;Server;TopdownL5;tma_L5_group;tma_mem_bandwidth_group;Clocks", - "MetricName": "tma_mba_stalls", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", - "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;Clocks", - "MetricName": "tma_mem_latency", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that)." - }, - { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory.", - "MetricExpr": "( ( 109 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", - "MetricName": "tma_local_mem", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory. Caching will improve the latency and increase performance. Sample with: MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM." - }, - { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory.", - "MetricExpr": "( ( 190 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", - "MetricName": "tma_remote_mem", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM." - }, - { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues.", - "MetricExpr": "( ( ( 170 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM + ( ( 170 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latency_group;Clocks_Estimated", - "MetricName": "tma_remote_cache", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD." - }, - { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", - "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", - "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", - "MetricName": "tma_store_bound", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." - }, - { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", - "MetricExpr": "( ( MEM_STORE_RETIRED.L2_HIT * ( 10 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", - "MetricName": "tma_store_latency", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)." - }, - { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", - "MetricExpr": "( ( 170 * tma_info_system_core_frequency ) * cpu@OCR.DEMAND_RFO.L3_MISS\\,offcore_rsp\\=0x103b800002@ + ( 81 * tma_info_system_core_frequency ) * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", - "MetricName": "tma_false_sharing", - "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM." - }, - { - "BriefDescription": "This metric represents rate of split store accesses.", - "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization", - "MetricName": "tma_split_stores", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." - }, - { - "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores.", - "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread_clks", - "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", - "MetricName": "tma_streaming_stores", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE." - }, - { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", - "MetricExpr": "( ( 7 ) * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_STORE_MISSES.WALK_ACTIVE ) / tma_info_core_core_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", - "MetricName": "tma_dtlb_store", - "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES." - }, - { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", - "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", - "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", - "MetricName": "tma_store_stlb_hit", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", - "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_clks", - "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", - "MetricName": "tma_store_stlb_miss", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", - "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", - "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", - "MetricName": "tma_store_stlb_miss_4k", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", - "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", - "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", - "MetricName": "tma_store_stlb_miss_2m", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", - "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", - "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", - "MetricName": "tma_store_stlb_miss_1g", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", - "MetricExpr": "max( 0 , tma_backend_bound - tma_memory_bound )", - "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Default;Slots", - "MetricName": "tma_core_bound", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations).", - "DefaultMetricgroupName": "TopdownL2", - "MetricGroupnoGroup": "TopdownL2;Default" - }, - { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", - "MetricExpr": "ARITH.DIV_ACTIVE / tma_info_thread_clks", - "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", - "MetricName": "tma_divider", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIVIDER_ACTIVE." - }, - { - "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active.", - "MetricExpr": "ARITH.FPDIV_ACTIVE / tma_info_thread_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", - "MetricName": "tma_fp_divider", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active.", - "MetricExpr": "tma_divider - tma_fp_divider", - "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", - "MetricName": "tma_int_divider", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", - "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks + tma_c02_wait", - "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Clocks", - "MetricName": "tma_serializing_operation", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD." - }, - { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", - "MetricExpr": "CPU_CLK_UNHALTED.PAUSE / tma_info_thread_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", - "MetricName": "tma_slow_pause", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.PAUSE_INST." - }, - { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings).", - "MetricExpr": "CPU_CLK_UNHALTED.C01 / tma_info_thread_clks", - "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", - "MetricName": "tma_c01_wait", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings).", - "MetricExpr": "CPU_CLK_UNHALTED.C02 / tma_info_thread_clks", - "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", - "MetricName": "tma_c02_wait", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions.", - "MetricExpr": "13 * MISC2_RETIRED.LFENCE / tma_info_thread_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", - "MetricName": "tma_memory_fence", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates fraction of cycles where the Advanced Matrix eXtensions (AMX) execution engine was busy with tile (arithmetic) operations.", - "MetricExpr": "EXE.AMX_BUSY / tma_info_core_core_clks", - "MetricGroup": "BvCB;Compute;HPC;Server;TopdownL3;tma_L3_group;tma_ports_utilized_0_group;Core_Clocks", - "MetricName": "tma_amx_busy", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", - "MetricExpr": "( tma_ports_utilized_0 * tma_info_thread_clks + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIV_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - EXE_ACTIVITY.BOUND_ON_LOADS ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL ) / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", - "MetricName": "tma_ports_utilization", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." - }, - { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", - "MetricExpr": "( EXE_ACTIVITY.EXE_BOUND_0_PORTS + max( RS.EMPTY_RESOURCE - RESOURCE_STALLS.SCOREBOARD , 0 ) ) / tma_info_thread_clks * ( CYCLE_ACTIVITY.STALLS_TOTAL - EXE_ACTIVITY.BOUND_ON_LOADS ) / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", - "MetricName": "tma_ports_utilized_0", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." - }, - { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", - "MetricExpr": "160 * ASSISTS.SSE_AVX_MIX / tma_info_thread_clks", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks", - "MetricName": "tma_mixing_vectors", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic." - }, - { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", - "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", - "MetricName": "tma_ports_utilized_1", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL." - }, - { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", - "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", - "MetricName": "tma_ports_utilized_2", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL." - }, - { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", - "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", - "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", - "MetricName": "tma_ports_utilized_3m", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Sample with: UOPS_EXECUTED.CYCLES_GE_3." - }, - { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", - "MetricExpr": "( UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + UOPS_DISPATCHED.PORT_5_11 + UOPS_DISPATCHED.PORT_6 ) / ( 5 * tma_info_core_core_clks )", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", - "MetricName": "tma_alu_op_utilization", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", - "MetricExpr": "UOPS_DISPATCHED.PORT_0 / tma_info_core_core_clks", - "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", - "MetricName": "tma_port_0", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0." - }, - { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", - "MetricExpr": "UOPS_DISPATCHED.PORT_1 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", - "MetricName": "tma_port_1", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1." - }, - { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", - "MetricExpr": "UOPS_DISPATCHED.PORT_6 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", - "MetricName": "tma_port_6", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1." - }, - { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", - "MetricExpr": "UOPS_DISPATCHED.PORT_2_3_10 / ( 3 * tma_info_core_core_clks )", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", - "MetricName": "tma_load_op_utilization", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations. Sample with: UOPS_DISPATCHED.PORT_2_3_10." - }, - { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", - "MetricExpr": "( UOPS_DISPATCHED.PORT_4_9 + UOPS_DISPATCHED.PORT_7_8 ) / ( 4 * tma_info_core_core_clks )", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", - "MetricName": "tma_store_op_utilization", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations. Sample with: UOPS_DISPATCHED.PORT_7_8." - }, - { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", - "MetricExpr": "( topdown\\-retiring / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", - "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Default;Slots", - "MetricName": "tma_retiring", - "ScaleUnit": "100%", - "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.SLOTS.", - "DefaultMetricgroupName": "TopdownL1", - "MetricGroupnoGroup": "TopdownL1;Default" - }, - { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", - "MetricExpr": "max( 0 , tma_retiring - tma_heavy_operations )", - "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", - "MetricName": "tma_light_operations", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST.", - "DefaultMetricgroupName": "TopdownL2", - "MetricGroupnoGroup": "TopdownL2;Default" - }, - { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", - "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", - "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", - "MetricName": "tma_fp_arith", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." - }, - { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", - "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", - "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", - "MetricName": "tma_x87_use", - "ScaleUnit": "100%", - "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." - }, - { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", - "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", - "MetricName": "tma_fp_scalar", - "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting." - }, - { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", - "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR + FP_ARITH_INST_RETIRED2.VECTOR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", - "MetricName": "tma_fp_vector", - "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting." - }, - { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", - "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.128B_PACKED_HALF ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", - "MetricName": "tma_fp_vector_128b", - "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL." - }, - { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", - "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.256B_PACKED_HALF ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", - "MetricName": "tma_fp_vector_256b", - "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL." - }, - { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors.", - "MetricExpr": "( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.512B_PACKED_HALF ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", - "MetricName": "tma_fp_vector_512b", - "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting." - }, - { - "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired).", - "MetricExpr": "tma_int_vector_128b + tma_int_vector_256b", - "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", - "MetricName": "tma_int_operations", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired). Vector/Matrix Int operations and shuffles are counted. Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain." - }, - { - "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", - "MetricExpr": "( INT_VEC_RETIRED.ADD_128 + INT_VEC_RETIRED.VNNI_128 ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops", - "MetricName": "tma_int_vector_128b", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", - "MetricExpr": "( INT_VEC_RETIRED.ADD_256 + INT_VEC_RETIRED.MUL_256 + INT_VEC_RETIRED.VNNI_256 ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Uops", - "MetricName": "tma_int_vector_256b", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", - "MetricExpr": "tma_light_operations * MEM_UOP_RETIRED.ANY / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", - "MetricName": "tma_memory_operations", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions.", - "MetricExpr": "tma_light_operations * INST_RETIRED.MACRO_FUSED / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", - "MetricName": "tma_fused_instructions", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Other_Light_Ops in MTL!)}." - }, - { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused.", - "MetricExpr": "tma_light_operations * ( BR_INST_RETIRED.ALL_BRANCHES - INST_RETIRED.MACRO_FUSED ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", - "MetricName": "tma_non_fused_branches", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused. Non-conditional branches like direct JMP or CALL would count here. Can be used to examine fusible conditional jumps that were not fused." - }, - { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", - "MetricExpr": "max( 0 , tma_light_operations - ( tma_fp_arith + tma_int_operations + tma_memory_operations + tma_fused_instructions + tma_non_fused_branches ) )", - "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", - "MetricName": "tma_other_light_ops", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." - }, - { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", - "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_light_operations_group;Slots", - "MetricName": "tma_nop_instructions", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." - }, - { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer).", - "MetricExpr": "tma_light_operations * INT_VEC_RETIRED.SHUFFLES / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_int_operations_group;Slots", - "MetricName": "tma_shuffles_256b", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer). Shuffles may incur slow cross \"vector lane\" data transfers." - }, - { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", - "MetricExpr": "( topdown\\-heavy\\-ops / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", - "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Default;Slots", - "MetricName": "tma_heavy_operations", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+]). Sample with: UOPS_RETIRED.HEAVY.", - "DefaultMetricgroupName": "TopdownL2", - "MetricGroupnoGroup": "TopdownL2;Default" - }, - { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops.", - "MetricExpr": "max( 0 , tma_heavy_operations - tma_microcode_sequencer )", - "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", - "MetricName": "tma_few_uops_instructions", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions." - }, - { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", - "MetricExpr": "UOPS_RETIRED.MS / tma_info_thread_slots", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", - "MetricName": "tma_microcode_sequencer", - "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: UOPS_RETIRED.MS." - }, - { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", - "MetricExpr": "( ( 99 *3 + 63 + 30 ) / 5 ) * ASSISTS.ANY / tma_info_thread_slots", - "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", - "MetricName": "tma_assists", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY." - }, - { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults.", - "MetricExpr": "99 * ASSISTS.PAGE_FAULT / tma_info_thread_slots", - "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", - "MetricName": "tma_page_faults", - "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults. A Page Fault may apply on first application access to a memory page. Note operating system handling of page faults accounts for the majority of its cost." - }, - { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", - "MetricExpr": "30 * ASSISTS.FP / tma_info_thread_slots", - "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", - "MetricName": "tma_fp_assists", - "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." - }, - { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.", - "MetricExpr": "63 * ASSISTS.SSE_AVX_MIX / tma_info_thread_slots", - "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", - "MetricName": "tma_avx_assists", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", - "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", - "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", - "MetricName": "tma_cisc", - "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources. Sample with: FRONTEND_RETIRED.MS_FLOWS." - }, - { - "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts.", - "MetricExpr": "100 * ( 1 - tma_core_bound / tma_ports_utilization if tma_core_bound < tma_ports_utilization else 1 ) if tma_info_system_smt_2t_utilization > 0.5 else 0", - "MetricGroup": "Cor;SMT;Metric", - "MetricName": "tma_info_botlnk_l0_core_bound_likely", - "MetricThreshold": "tma_info_botlnk_core_bound_likely > 0.5" - }, - { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", - "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", - "MetricGroup": "Ret;Summary;Metric", - "MetricName": "tma_info_thread_ipc" - }, - { - "BriefDescription": "Uops Per Instruction.", - "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / INST_RETIRED.ANY", - "MetricGroup": "Pipeline;Ret;Retire;Metric", - "MetricName": "tma_info_thread_uoppi", - "MetricThreshold": "tma_info_thread_uoppi > 1.05" - }, - { - "BriefDescription": "Uops per taken branch.", - "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / BR_INST_RETIRED.NEAR_TAKEN", - "MetricGroup": "Branches;Fed;FetchBW;Metric", - "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 6 * 1.5" - }, - { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", - "MetricExpr": "1 / tma_info_thread_ipc", - "MetricGroup": "Pipeline;Mem;Metric", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", - "MetricExpr": "CPU_CLK_UNHALTED.THREAD", - "MetricGroup": "Pipeline;Count", - "MetricName": "tma_info_thread_clks" - }, - { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", - "MetricExpr": "slots", - "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", - "MetricName": "tma_info_thread_slots" - }, - { - "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor.", - "MetricExpr": "tma_info_thread_slots / ( TOPDOWN.SLOTS / 2 ) if #SMT_on else 1", - "MetricGroup": "SMT;TmaL1;TopdownL1;tma_L1_group;Metric", - "MetricName": "tma_info_thread_slots_utilization" - }, - { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", - "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", - "MetricGroup": "Cor;Pipeline;Metric", - "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." - }, - { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", - "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", - "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", - "MetricName": "tma_info_core_coreipc" - }, - { - "BriefDescription": "Floating Point Operations Per Cycle.", - "MetricExpr": "( ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR_HALF ) + 2 * ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED2.COMPLEX_SCALAR_HALF ) + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * ( FP_ARITH_INST_RETIRED2.128B_PACKED_HALF + FP_ARITH_INST_RETIRED.8_FLOPS ) + 16 * ( FP_ARITH_INST_RETIRED2.256B_PACKED_HALF + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) + 32 * FP_ARITH_INST_RETIRED2.512B_PACKED_HALF ) ) / tma_info_core_core_clks", - "MetricGroup": "Ret;Flops;Core_Metric", - "MetricName": "tma_info_core_flopc" - }, - { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", - "MetricExpr": "( FP_ARITH_DISPATCHED.PORT_0 + FP_ARITH_DISPATCHED.PORT_1 + FP_ARITH_DISPATCHED.PORT_5 ) / ( 2 * tma_info_core_core_clks )", - "MetricGroup": "Cor;Flops;HPC;Core_Metric", - "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." - }, - { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", - "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", - "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", - "MetricName": "tma_info_core_ilp" - }, - { - "BriefDescription": "uops Executed per Cycle.", - "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", - "MetricGroup": "Power;Metric", - "MetricName": "tma_info_core_epc" - }, - { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", - "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED if #SMT_on else tma_info_thread_clks", - "MetricGroup": "SMT;Count", - "MetricName": "tma_info_core_core_clks" - }, - { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", - "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipload", - "MetricThreshold": "tma_info_inst_mix_ipload < 3" - }, - { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", - "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipstore", - "MetricThreshold": "tma_info_inst_mix_ipstore < 8" - }, - { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", - "MetricGroup": "Branches;Fed;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipbranch", - "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" - }, - { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", - "MetricGroup": "Branches;Fed;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipcall", - "MetricThreshold": "tma_info_inst_mix_ipcall < 200" - }, - { - "BriefDescription": "Instructions per taken branch.", - "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", - "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", - "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 6 * 2 + 1", - "PublicDescription": "Instructions per taken branch. Related metrics: tma_info_frontend_dsb_coverage, tma_info_botlnk_l2_dsb_misses, tma_info_botlnk_l2_dsb_bandwidth." - }, - { - "BriefDescription": "Branch instructions per taken branch.", - "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", - "MetricGroup": "Branches;Fed;PGO;Metric", - "MetricName": "tma_info_inst_mix_bptkbranch" - }, - { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / ( ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR_HALF ) + 2 * ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED2.COMPLEX_SCALAR_HALF ) + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * ( FP_ARITH_INST_RETIRED2.128B_PACKED_HALF + FP_ARITH_INST_RETIRED.8_FLOPS ) + 16 * ( FP_ARITH_INST_RETIRED2.256B_PACKED_HALF + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) + 32 * FP_ARITH_INST_RETIRED2.512B_PACKED_HALF ) )", - "MetricGroup": "Flops;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipflop", - "MetricThreshold": "tma_info_inst_mix_ipflop < 10" - }, - { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR + FP_ARITH_INST_RETIRED2.VECTOR ) )", - "MetricGroup": "Flops;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_iparith", - "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." - }, - { - "BriefDescription": "Instructions per FP Arithmetic Scalar Half-Precision instruction (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED2.SCALAR", - "MetricGroup": "Flops;FpScalar;InsType;Server;Inst_Metric", - "MetricName": "tma_info_inst_mix_iparith_scalar_hp", - "MetricThreshold": "tma_info_inst_mix_iparith_scalar_hp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Half-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." - }, - { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", - "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_iparith_scalar_sp", - "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." - }, - { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", - "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_iparith_scalar_dp", - "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." - }, - { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.128B_PACKED_HALF )", - "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_iparith_avx128", - "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." - }, - { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.256B_PACKED_HALF )", - "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_iparith_avx256", - "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." - }, - { - "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.512B_PACKED_HALF )", - "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_iparith_avx512", - "MetricThreshold": "tma_info_inst_mix_iparith_avx512 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." - }, - { - "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate).", - "MetricExpr": "tma_info_inst_mix_instructions / CPU_CLK_UNHALTED.PAUSE_INST", - "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ippause" - }, - { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / SW_PREFETCH_ACCESS.ANY", - "MetricGroup": "Prefetches;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipswpf", - "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" - }, - { - "BriefDescription": "Total number of retired Instructions.", - "MetricExpr": "INST_RETIRED.ANY", - "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", - "MetricName": "tma_info_inst_mix_instructions", - "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." - }, - { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", - "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", - "MetricGroup": "Pipeline;Ret;Metric", - "MetricName": "tma_info_pipeline_retire" - }, - { - "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions.", - "MetricExpr": "INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", - "MetricGroup": "MicroSeq;Pipeline;Ret;Metric", - "MetricName": "tma_info_pipeline_strings_cycles", - "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1" - }, - { - "BriefDescription": "Instructions per a microcode Assist invocation.", - "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", - "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", - "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", - "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." - }, - { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core.", - "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@ )", - "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", - "MetricName": "tma_info_pipeline_execute" - }, - { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", - "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", - "MetricGroup": "Fed;FetchBW;Metric", - "MetricName": "tma_info_pipeline_fetch_dsb" - }, - { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", - "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES_ANY", - "MetricGroup": "Fed;FetchBW;Metric", - "MetricName": "tma_info_pipeline_fetch_mite" - }, - { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", - "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", - "MetricGroup": "Fed;FetchBW;Metric", - "MetricName": "tma_info_frontend_fetch_upc" - }, - { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", - "MetricExpr": "IDQ.DSB_UOPS / ( UOPS_ISSUED.ANY )", - "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", - "MetricName": "tma_info_frontend_dsb_coverage", - "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 6 > 0.35", - "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_info_inst_mix_iptb, tma_info_botlnk_l2_dsb_misses, tma_info_botlnk_l2_dsb_bandwidth." - }, - { - "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection.", - "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / cpu@INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", - "MetricGroup": "Fed;Metric", - "MetricName": "tma_info_frontend_unknown_branch_cost", - "PublicDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection. See Unknown_Branches node." - }, - { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", - "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", - "MetricGroup": "DSBmiss;Metric", - "MetricName": "tma_info_frontend_dsb_switch_cost" - }, - { - "BriefDescription": "Taken Branches retired Per Cycle.", - "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", - "MetricGroup": "Branches;FetchBW;Metric", - "MetricName": "tma_info_frontend_tbpc" - }, - { - "BriefDescription": "Average Latency for L1 instruction cache misses.", - "MetricExpr": "ICACHE_DATA.STALLS / cpu@ICACHE_DATA.STALLS\\,cmask\\=0x1\\,edge\\=0x1@", - "MetricGroup": "Fed;FetchLat;IcMiss;Metric", - "MetricName": "tma_info_frontend_icache_miss_latency" - }, - { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", - "MetricGroup": "DSBmiss;Fed;Inst_Metric", - "MetricName": "tma_info_frontend_ipdsb_miss_ret", - "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" - }, - { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", - "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", - "MetricGroup": "Fed;Metric", - "MetricName": "tma_info_frontend_ipunknown_branch" - }, - { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", - "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", - "MetricGroup": "IcMiss;Metric", - "MetricName": "tma_info_frontend_l2mpki_code" - }, - { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", - "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", - "MetricGroup": "IcMiss;Metric", - "MetricName": "tma_info_frontend_l2mpki_code_all" - }, - { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", - "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb + tma_ms ) )", - "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", - "MetricName": "tma_info_botlnk_l2_dsb_misses", - "MetricThreshold": "tma_info_botlnk_dsb_misses > 10", - "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_info_inst_mix_iptb, tma_info_frontend_dsb_coverage, tma_info_botlnk_l2_dsb_bandwidth." - }, - { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", - "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb + tma_ms ) ) )", - "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", - "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", - "MetricThreshold": "tma_info_botlnk_dsb_bandwidth > 10", - "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_info_inst_mix_iptb, tma_info_frontend_dsb_coverage, tma_info_botlnk_l2_dsb_misses." - }, - { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", - "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", - "MetricName": "tma_info_botlnk_l2_ic_misses", - "MetricThreshold": "tma_info_botlnk_ic_misses > 5", - "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." - }, - { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", - "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmispredict", - "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" - }, - { - "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", - "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", - "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" - }, - { - "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", - "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", - "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" - }, - { - "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", - "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_ret", - "MetricThreshold": "tma_info_bad_spec_ipmisp_ret < 500" - }, - { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", - "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", - "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" - }, - { - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", - "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 6 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", - "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", - "MetricName": "tma_info_bad_spec_branch_misprediction_cost", - "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)." - }, - { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", - "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", - "MetricGroup": "BrMispredicts;Metric", - "MetricName": "tma_info_bad_spec_spec_clears_ratio" - }, - { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", - "MetricExpr": "BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES", - "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", - "MetricName": "tma_info_branches_cond_nt" - }, - { - "BriefDescription": "Fraction of branches that are taken conditionals.", - "MetricExpr": "BR_INST_RETIRED.COND_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", - "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", - "MetricName": "tma_info_branches_cond_tk" - }, - { - "BriefDescription": "Fraction of branches that are CALL or RET.", - "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", - "MetricGroup": "Bad;Branches;Fraction", - "MetricName": "tma_info_branches_callret" - }, - { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", - "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", - "MetricGroup": "Bad;Branches;Fraction", - "MetricName": "tma_info_branches_jump" - }, - { - "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group).", - "MetricExpr": "1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", - "MetricGroup": "Bad;Branches;Fraction", - "MetricName": "tma_info_branches_other_branches" - }, - { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", - "MetricExpr": "L1D_PEND_MISS.PENDING / MEM_LOAD_COMPLETED.L1_MISS_ANY", - "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", - "MetricName": "tma_info_memory_load_miss_real_latency" - }, - { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", - "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", - "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", - "MetricName": "tma_info_memory_mlp", - "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." - }, - { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", - "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", - "MetricGroup": "CacheHits;Mem;Metric", - "MetricName": "tma_info_memory_l1mpki" - }, - { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", - "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", - "MetricGroup": "CacheHits;Mem;Metric", - "MetricName": "tma_info_memory_l1mpki_load" - }, - { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", - "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", - "MetricGroup": "Mem;Backend;CacheHits;Metric", - "MetricName": "tma_info_memory_l2mpki" - }, - { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", - "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", - "MetricGroup": "CacheHits;Mem;Offcore;Metric", - "MetricName": "tma_info_memory_l2mpki_all" - }, - { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", - "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", - "MetricGroup": "CacheHits;Mem;Metric", - "MetricName": "tma_info_memory_l2mpki_load" - }, - { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", - "MetricExpr": "1000 * L2_RQSTS.RFO_MISS / INST_RETIRED.ANY", - "MetricGroup": "CacheMisses;Offcore;Metric", - "MetricName": "tma_info_memory_l2mpki_rfo" - }, - { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", - "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", - "MetricGroup": "CacheHits;Mem;Metric", - "MetricName": "tma_info_memory_l2hpki_all" - }, - { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", - "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", - "MetricGroup": "CacheHits;Mem;Metric", - "MetricName": "tma_info_memory_l2hpki_load" - }, - { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", - "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", - "MetricGroup": "Mem;Metric", - "MetricName": "tma_info_memory_l3mpki" - }, - { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", - "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", - "MetricGroup": "CacheHits;Mem;Metric", - "MetricName": "tma_info_memory_fb_hpki" - }, - { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", - "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", - "MetricGroup": "Mem;MemoryBW;Metric", - "MetricName": "tma_info_memory_l1d_cache_fill_bw" - }, - { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", - "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", - "MetricGroup": "Mem;MemoryBW;Metric", - "MetricName": "tma_info_memory_l2_cache_fill_bw" - }, - { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", - "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", - "MetricGroup": "Mem;MemoryBW;Metric", - "MetricName": "tma_info_memory_l3_cache_fill_bw" - }, - { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", - "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", - "MetricGroup": "Mem;MemoryBW;Offcore;Metric", - "MetricName": "tma_info_memory_l3_cache_access_bw" - }, - { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", - "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 4 * tma_info_core_core_clks )", - "MetricGroup": "Mem;MemoryTLB;Core_Metric", - "MetricName": "tma_info_memory_tlb_page_walks_utilization", - "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" - }, - { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", - "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", - "MetricGroup": "Fed;MemoryTLB;Metric", - "MetricName": "tma_info_memory_tlb_code_stlb_mpki" - }, - { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", - "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", - "MetricGroup": "Mem;MemoryTLB;Metric", - "MetricName": "tma_info_memory_tlb_load_stlb_mpki" - }, - { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", - "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", - "MetricGroup": "Mem;MemoryTLB;Metric", - "MetricName": "tma_info_memory_tlb_store_stlb_mpki" - }, - { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", - "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", - "MetricGroup": "Mem;MemoryBW;Core_Metric", - "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" - }, - { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", - "MetricExpr": "tma_info_memory_l2_cache_fill_bw", - "MetricGroup": "Mem;MemoryBW;Core_Metric", - "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" - }, - { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", - "MetricExpr": "tma_info_memory_l3_cache_fill_bw", - "MetricGroup": "Mem;MemoryBW;Core_Metric", - "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" - }, - { - "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec].", - "MetricExpr": "tma_info_memory_l3_cache_access_bw", - "MetricGroup": "Mem;MemoryBW;Offcore;Core_Metric", - "MetricName": "tma_info_memory_core_l3_cache_access_bw_2t" - }, - { - "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory).", - "MetricExpr": "1000 * L2_LINES_OUT.SILENT / tma_info_inst_mix_instructions", - "MetricGroup": "L2Evicts;Mem;Server;Core_Metric", - "MetricName": "tma_info_memory_core_l2_evictions_silent_pki" - }, - { - "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction.", - "MetricExpr": "1000 * L2_LINES_OUT.NON_SILENT / tma_info_inst_mix_instructions", - "MetricGroup": "L2Evicts;Mem;Server;Core_Metric", - "MetricName": "tma_info_memory_core_l2_evictions_nonsilent_pki" - }, - { - "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses.", - "MetricExpr": "L2_LINES_OUT.USELESS_HWPF / ( L2_LINES_OUT.SILENT + L2_LINES_OUT.NON_SILENT )", - "MetricGroup": "Prefetches;Metric", - "MetricName": "tma_info_memory_prefetches_useless_hwpf" - }, - { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", - "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", - "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", - "MetricName": "tma_info_memory_latency_load_l2_miss_latency" - }, - { - "BriefDescription": "Average Latency for L3 cache miss demand Loads.", - "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.L3_MISS_DEMAND_DATA_RD / OFFCORE_REQUESTS.L3_MISS_DEMAND_DATA_RD", - "MetricGroup": "Memory_Lat;Offcore;Clocks_Latency", - "MetricName": "tma_info_memory_latency_load_l3_miss_latency" - }, - { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", - "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=0x1@", - "MetricGroup": "Memory_BW;Offcore;Metric", - "MetricName": "tma_info_memory_latency_load_l2_mlp" - }, - { - "BriefDescription": "Average Parallel L2 cache miss data reads.", - "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", - "MetricGroup": "Memory_BW;Offcore;Metric", - "MetricName": "tma_info_memory_latency_data_l2_mlp" - }, - { - "BriefDescription": "Off-core accesses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches).", - "MetricExpr": "1000 * OCR.READS_TO_CORE.ANY_RESPONSE / tma_info_inst_mix_instructions", - "MetricGroup": "CacheHits;Offcore;Metric", - "MetricName": "tma_info_memory_mix_offcore_read_any_pki" - }, - { - "BriefDescription": "L3 cache misses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches).", - "MetricExpr": "1000 * OCR.READS_TO_CORE.L3_MISS / tma_info_inst_mix_instructions", - "MetricGroup": "Offcore;Metric", - "MetricName": "tma_info_memory_mix_offcore_read_l3m_pki" - }, - { - "BriefDescription": "High-Bandwidth Memory (HBM) accesses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches).", - "MetricExpr": "if 0 > 2 else 1000 * OCR.DEMAND_DATA_RD.PMM / tma_info_inst_mix_instructions", - "MetricGroup": "Offcore;Server;Metric", - "MetricName": "tma_info_memory_mix_offcore_read_hbm_pki" - }, - { - "BriefDescription": "Off-core accesses per kilo instruction for modified write requests.", - "MetricExpr": "1000 * OCR.MODIFIED_WRITE.ANY_RESPONSE / tma_info_inst_mix_instructions", - "MetricGroup": "Offcore;Metric", - "MetricName": "tma_info_memory_mix_offcore_mwrite_any_pki" - }, - { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", - "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", - "MetricGroup": "Mem;Metric", - "MetricName": "tma_info_memory_mix_uc_load_pki" - }, - { - "BriefDescription": "\"Bus lock\" per kilo instruction.", - "MetricExpr": "1000 * SQ_MISC.BUS_LOCK / INST_RETIRED.ANY", - "MetricGroup": "Mem;Metric", - "MetricName": "tma_info_memory_mix_bus_lock_pki" - }, - { - "BriefDescription": "Average CPU Utilization (percentage).", - "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", - "MetricGroup": "HPC;Summary;Metric", - "MetricName": "tma_info_system_cpu_utilization" - }, - { - "BriefDescription": "Average number of utilized CPUs.", - "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", - "MetricGroup": "Summary;Metric", - "MetricName": "tma_info_system_cpus_utilized" - }, - { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", - "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", - "MetricGroup": "Summary;Power;System_Metric", - "MetricName": "tma_info_system_core_frequency" - }, - { - "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz].", - "MetricExpr": "tma_info_system_socket_clks / 1e9 / tma_info_system_time", - "MetricGroup": "SoC;System_Metric", - "MetricName": "tma_info_system_uncore_frequency" - }, - { - "BriefDescription": "Giga Floating Point Operations Per Second.", - "MetricExpr": "( ( ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIRED2.SCALAR_HALF ) + 2 * ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED2.COMPLEX_SCALAR_HALF ) + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * ( FP_ARITH_INST_RETIRED2.128B_PACKED_HALF + FP_ARITH_INST_RETIRED.8_FLOPS ) + 16 * ( FP_ARITH_INST_RETIRED2.256B_PACKED_HALF + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) + 32 * FP_ARITH_INST_RETIRED2.512B_PACKED_HALF ) ) / ( 1000000000 ) ) / tma_info_system_time", - "MetricGroup": "Cor;Flops;HPC;Metric", - "MetricName": "tma_info_system_gflops", - "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." - }, - { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", - "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", - "MetricGroup": "Power;Core_Metric", - "MetricName": "tma_info_system_turbo_utilization" - }, - { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", - "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED if #SMT_on else 0", - "MetricGroup": "SMT;Core_Metric", - "MetricName": "tma_info_system_smt_2t_utilization" - }, - { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", - "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", - "MetricGroup": "OS;Metric", - "MetricName": "tma_info_system_kernel_utilization", - "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" - }, - { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", - "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", - "MetricGroup": "OS;Metric", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states.", - "MetricExpr": "CPU_CLK_UNHALTED.C0_WAIT / tma_info_thread_clks", - "MetricGroup": "C0Wait;Metric", - "MetricName": "tma_info_system_c0_wait", - "MetricThreshold": "tma_info_system_c0_wait > 0.05" - }, - { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", - "MetricExpr": "( 64 * ( UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR ) / ( 1000000000 ) ) / tma_info_system_time", - "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", - "MetricName": "tma_info_system_dram_bw_use", - "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_bottleneck_cache_memory_bandwidth." - }, - { - "BriefDescription": "Average latency of data read request to external memory (in nanoseconds).", - "MetricExpr": "( 1000000000 ) * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( tma_info_system_socket_clks / tma_info_system_time )", - "MetricGroup": "Mem;MemoryLat;SoC;NanoSeconds", - "MetricName": "tma_info_system_mem_read_latency", - "PublicDescription": "Average latency of data read request to external memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetches. ([RKL+]memory-controller only)." - }, - { - "BriefDescription": "Average number of parallel data read requests to external memory.", - "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD\\,thresh\\=0x1@", - "MetricGroup": "Mem;MemoryBW;SoC;System_Metric", - "MetricName": "tma_info_system_mem_parallel_reads", - "PublicDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches." - }, - { - "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds].", - "MetricExpr": "( 1000000000 ) * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR ) / cha_0@event\\=0x0@", - "MetricGroup": "MemOffcore;MemoryLat;SoC;Server;NanoSeconds", - "MetricName": "tma_info_system_mem_dram_read_latency", - "PublicDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]. Accounts for demand loads and L1/L2 data-read prefetches." - }, - { - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec].", - "MetricExpr": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / ( 1000000000 ) / tma_info_system_time", - "MetricGroup": "IoBW;MemOffcore;SoC;Server;GB/sec", - "MetricName": "tma_info_system_io_read_bw", - "PublicDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]. Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU." - }, - { - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec].", - "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_ITOM + UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR ) * 64 / ( 1000000000 ) / tma_info_system_time", - "MetricGroup": "IoBW;MemOffcore;SoC;Server;GB/sec", - "MetricName": "tma_info_system_io_write_bw", - "PublicDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]. Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU." - }, - { - "BriefDescription": "Cross-socket Ultra Path Interconnect (UPI) data transmit bandwidth for data only [MB / sec].", - "MetricExpr": "UNC_UPI_TxL_FLITS.ALL_DATA * 64 / 9 / 1000000", - "MetricGroup": "SoC;Server;MB/sec", - "MetricName": "tma_info_system_upi_data_transmit_bw" - }, - { - "BriefDescription": "Total package Power in Watts.", - "MetricExpr": "( power@energy\\-pkg@ * ( 61 ) + 15.6 * power@energy\\-ram@ ) / ( ( duration_time ) * ( 1000000 ) )", - "MetricGroup": "Power;SoC;System_Metric", - "MetricName": "tma_info_system_power" - }, - { - "BriefDescription": "Run duration time in seconds.", - "MetricExpr": "duration_time", - "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time", - "MetricThreshold": "tma_info_system_time < 1" - }, - { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", - "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", - "MetricGroup": "Summary;Clocks", - "MetricName": "tma_info_system_mux", - "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" - }, - { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", - "MetricExpr": "cha_0@event\\=0x0@", - "MetricGroup": "SoC;Count", - "MetricName": "tma_info_system_socket_clks" - }, - { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", - "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", - "MetricGroup": "Branches;OS;Inst_Metric", - "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" - }, - { - "BriefDescription": "Average Off-core access BW for Reads-to-Core (R2C).", - "MetricExpr": "64 * OCR.READS_TO_CORE.ANY_RESPONSE / 1e9 / tma_info_system_time", - "MetricGroup": "HPC;Mem;MemoryBW;SoC;GB/sec", - "MetricName": "tma_info_memory_soc_r2c_offcore_bw", - "PublicDescription": "Average Off-core access BW for Reads-to-Core (R2C). R2C account for demand or prefetch load/RFO/code access that fill data into the Core caches." - }, - { - "BriefDescription": "Average L3-cache miss BW for Reads-to-Core (R2C).", - "MetricExpr": "64 * OCR.READS_TO_CORE.L3_MISS / 1e9 / tma_info_system_time", - "MetricGroup": "HPC;Mem;MemoryBW;SoC;GB/sec", - "MetricName": "tma_info_memory_soc_r2c_l3m_bw", - "PublicDescription": "Average L3-cache miss BW for Reads-to-Core (R2C). This covering going to DRAM or other memory off-chip memory tears. See R2C_Offcore_BW." - }, - { - "BriefDescription": "Average DRAM BW for Reads-to-Core (R2C) covering for memory attached to local- and remote-socket.", - "MetricExpr": "64 * OCR.READS_TO_CORE.DRAM / 1e9 / tma_info_system_time", - "MetricGroup": "HPC;Mem;MemoryBW;SoC;GB/sec", - "MetricName": "tma_info_memory_soc_r2c_dram_bw", - "PublicDescription": "Average DRAM BW for Reads-to-Core (R2C) covering for memory attached to local- and remote-socket. See R2C_Offcore_BW." - }, - { - "BriefDescription": "Average HBM BW for Reads-to-Core.", - "MetricExpr": "if 0 > 2 else 64 * OCR.DEMAND_DATA_RD.PMM / 1e9 / tma_info_system_time", - "MetricGroup": "HPC;Mem;MemoryBW;Server;SoC;GB/sec", - "MetricName": "tma_info_memory_soc_r2c_hbm_bw", - "PublicDescription": "Average HBM BW for Reads-to-Core. See R2C_Offcore_BW." - } -] \ No newline at end of file diff --git a/SPR/metrics/sapphirerapids_metrics.json b/SPR/metrics/sapphirerapids_metrics.json index eca5ff50..3fc7e13d 100644 --- a/SPR/metrics/sapphirerapids_metrics.json +++ b/SPR/metrics/sapphirerapids_metrics.json @@ -2,7 +2,7 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for 4th Generation Intel(R) Xeon(R) Processor Scalable Family based on Sapphire Rapids microarchitecture0", - "DatePublished": "11/06/2024", + "DatePublished": "11/15/2024", "Version": "1.0", "Legend": "", "TmaVersion": "5.01", @@ -1320,7 +1320,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Mispredictions > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Mispredictions" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Mispredictions > 20", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP", @@ -1408,7 +1416,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Big_Code > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Big_Code" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Big_Code > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB", @@ -1557,7 +1573,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Instruction_Fetch_BW" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", @@ -1746,7 +1770,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore", @@ -1975,7 +2007,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Latency" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;Mem;MemoryLat;Offcore", @@ -2164,7 +2204,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Data_TLBs" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore", @@ -2377,7 +2425,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Synchronization > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Synchronization" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Synchronization > 10", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;LockCont;Mem;Offcore", @@ -2482,7 +2538,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Compute_Bound_Est" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20", + "ThresholdIssues": "$issueComp" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Cor", @@ -2687,7 +2751,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Irregular_Overhead > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Irregular_Overhead" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Irregular_Overhead > 10", + "ThresholdIssues": "$issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BvIO;Cor;Ret", @@ -3096,7 +3168,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Other_Bottlenecks" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;Cor;Offcore", @@ -3132,7 +3212,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Branching_Overhead > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Branching_Overhead" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Bottleneck_Branching_Overhead > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Ret", @@ -3196,7 +3284,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Useful_Work > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Useful_Work" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Useful_Work > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;Ret", @@ -3240,7 +3336,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;BvIO;TmaL1;PGO", @@ -3289,7 +3393,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Frontend;TmaL2", @@ -3318,7 +3434,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss", @@ -3351,7 +3483,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -3380,7 +3532,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -3409,7 +3581,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB", @@ -3442,7 +3630,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3471,7 +3679,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3508,7 +3736,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3545,7 +3797,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -3578,7 +3854,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -3635,7 +3927,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Mispredicts_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP", @@ -3692,7 +4004,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Clears_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;MachineClears", @@ -3721,7 +4053,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Unknown_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat", @@ -3758,7 +4110,23 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC, $issueMS, $issueMV, $issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MicroSeq", @@ -3787,16 +4155,32 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "FetchLat", - "LocateWith": "" - }, - { - "MetricName": "DSB_Switches", - "LegacyName": "metric_TMA_....DSB_Switches(%)", - "ParentCategory": "Fetch_Latency", + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LCP(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" + }, + "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", + "MetricGroup": "FetchLat", + "LocateWith": "" + }, + { + "MetricName": "DSB_Switches", + "LegacyName": "metric_TMA_....DSB_Switches(%)", + "ParentCategory": "Fetch_Latency", "Level": 3, "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty.", "UnitOfMeasure": "percent", @@ -3816,7 +4200,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchLat", @@ -3865,7 +4265,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;Frontend;TmaL2", @@ -3911,7 +4319,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -3957,7 +4377,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Decoder0_Alone(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -4003,7 +4439,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;FetchBW", @@ -4057,7 +4505,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 5 & b > 20", + "BaseFormula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -4101,7 +4561,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", @@ -4142,7 +4610,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2", @@ -4195,7 +4675,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;BrMispredicts", @@ -4244,7 +4740,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueMC, $issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2", @@ -4301,7 +4809,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Nukes(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;Machine_Clears", @@ -4337,7 +4861,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;TmaL1", @@ -4378,7 +4910,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20", + "BaseFormula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2", @@ -4411,7 +4955,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1, $issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -4452,7 +5012,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -4493,7 +5073,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4522,7 +5126,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4563,7 +5191,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4604,7 +5260,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4645,7 +5329,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4674,7 +5386,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Fwd_Blk(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4724,13 +5456,33 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvML;MemoryLat", - "LocateWith": " MEM_LOAD_RETIRED.L1_HIT" - }, - { + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L1_Latency_Dependency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" + }, + "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", + "MetricGroup": "BvML;MemoryLat", + "LocateWith": " MEM_LOAD_RETIRED.L1_HIT" + }, + { "MetricName": "Lock_Latency", "LegacyName": "metric_TMA_......Lock_Latency(%)", "ParentCategory": "L1_Bound", @@ -4769,7 +5521,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Lock_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Offcore", @@ -4806,7 +5578,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......Split_Loads(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Loads(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......Split_Loads(%) > 30", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4835,7 +5615,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......FB_Full(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FB_Full(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......FB_Full(%) > 30", + "ThresholdIssues": "$issueBW, $issueSL, $issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW", @@ -4868,7 +5656,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem", @@ -4918,7 +5722,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L2_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryLat", @@ -4951,7 +5775,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -5013,7 +5853,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Contested_Accesses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -5075,7 +5935,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Data_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;Offcore;Snoop", @@ -5125,7 +6005,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L3_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -5158,7 +6058,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......SQ_Full(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 30 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -5187,7 +6107,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -5216,7 +6152,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Bandwidth(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -5245,7 +6201,31 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_........MBA_Stalls(%) > 10 & metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........MBA_Stalls(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......MEM_Bandwidth(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........MBA_Stalls(%) > 10 & metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBW;Offcore;Server", @@ -5278,7 +6258,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat;Offcore", @@ -5328,7 +6328,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Local_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Local_MEM(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Local_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Server", @@ -5378,7 +6402,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Remote_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Remote_MEM(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Remote_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Server;Snoop", @@ -5432,7 +6480,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Remote_Cache(%) > 5 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Remote_Cache(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Remote_Cache(%) > 5 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Offcore;Server;Snoop", @@ -5461,7 +6533,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -5502,7 +6590,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO, $issueSL, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore", @@ -5548,7 +6656,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......False_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -5590,7 +6718,27 @@ "Category": "TMA", "CountDomain": "Core_Utilization", "Threshold": { - "Formula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5619,7 +6767,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Streaming_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBW;Offcore", @@ -5665,7 +6833,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -5711,7 +6899,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5753,7 +6965,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5807,7 +7043,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5861,7 +7125,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5915,7 +7207,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -5956,7 +7276,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2;Compute", @@ -5985,7 +7317,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB", @@ -6014,7 +7362,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6047,7 +7415,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......INT_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6080,7 +7468,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;PortsUtil", @@ -6109,7 +7513,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Slow_Pause(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6138,7 +7562,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......C01_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......C01_Wait(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......C01_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", @@ -6167,7 +7611,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......C02_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......C02_Wait(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......C02_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "C0Wait", @@ -6196,7 +7660,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Memory_Fence(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Memory_Fence(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Memory_Fence(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6238,7 +7722,23 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_....AMX_Busy(%) > 50 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....AMX_Busy(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 50 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....AMX_Busy(%) > 50 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Compute;HPC;Server", @@ -6311,7 +7811,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -6356,7 +7872,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_0(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -6385,7 +7921,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_........Mixing_Vectors(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Mixing_Vectors(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Mixing_Vectors(%) > 5", + "ThresholdIssues": "$issueMV" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6414,7 +7958,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_1(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -6443,7 +8007,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_2(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -6472,7 +8056,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_3m(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 40 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;PortsUtil", @@ -6526,7 +8130,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........ALU_Op_Utilization(%) > 40" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........ALU_Op_Utilization(%)" + } + ], + "Formula": "a > 40", + "BaseFormula": "metric_TMA_........ALU_Op_Utilization(%) > 40", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6568,7 +8180,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_0(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_0(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_0(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -6610,7 +8230,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_1(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_1(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_1(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6652,7 +8280,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_6(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_6(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_6(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6694,7 +8330,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Load_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Load_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6740,7 +8384,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Store_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Store_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6776,7 +8428,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Retiring(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 70 | b > 10", + "BaseFormula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;TmaL1", @@ -6817,7 +8481,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -6882,7 +8554,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 20 & b > 60", + "BaseFormula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -6927,7 +8611,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......X87_Use(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -6976,7 +8676,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Scalar(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -7025,7 +8741,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -7078,7 +8810,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -7131,7 +8883,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -7184,7 +8956,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_512b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -7245,7 +9037,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -7294,7 +9098,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......Int_Vector_128b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Int_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 60", + "BaseFormula": "metric_TMA_......Int_Vector_128b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;IntVector;Pipeline", @@ -7347,7 +9167,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......Int_Vector_256b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Int_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Int_Operations(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 60", + "BaseFormula": "metric_TMA_......Int_Vector_256b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;IntVector;Pipeline", @@ -7396,7 +9232,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Memory_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -7445,7 +9293,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Fused_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -7498,7 +9358,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Non_Fused_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -7599,7 +9471,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 30 & b > 60", + "BaseFormula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -7648,7 +9532,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Nop_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Pipeline", @@ -7697,7 +9597,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Shuffles_256b(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Shuffles_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Shuffles_256b(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC;Pipeline", @@ -7738,7 +9654,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -7787,7 +9711,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Few_Uops_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -7816,7 +9752,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueMC, $issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -7845,7 +9793,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Assists(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO", @@ -7874,7 +9838,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........Page_Faults(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Page_Faults(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Page_Faults(%) > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -7903,7 +9875,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........FP_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........FP_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -7932,7 +9912,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........AVX_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........AVX_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........AVX_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -7965,7 +9953,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......CISC(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -9443,6 +11447,11 @@ "BaseFormula": " uops_executed.thread / ( ( uops_executed.core_cycles_ge_1 / 2 ) if smt_on else uops_executed.thread:c1 )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "LocateWith": "" @@ -9658,6 +11667,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -11499,7 +13513,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Useless_HWPF" + } + ], + "Formula": "a > 0.15", + "BaseFormula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", diff --git a/SPR/metrics/sapphirerapidshbm_metrics.json b/SPR/metrics/sapphirerapidshbm_metrics.json deleted file mode 100644 index 7b39b333..00000000 --- a/SPR/metrics/sapphirerapidshbm_metrics.json +++ /dev/null @@ -1,12856 +0,0 @@ -{ - "Header": { - "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", - "Info": "Performance Monitoring Metrics for 4th Generation Intel(R) Xeon(R) Processor Scalable Family based on Sapphire Rapids microarchitecture0", - "DatePublished": "11/06/2024", - "Version": "1.0", - "Legend": "", - "TmaVersion": "5.01", - "TmaFlavor": "Full" - }, - "Metrics": [ - { - "MetricName": "cpu_operating_frequency", - "LegacyName": "metric_CPU operating frequency (in GHz)", - "Level": 1, - "BriefDescription": "CPU operating frequency (in GHz)", - "UnitOfMeasure": "GHz", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "b" - } - ], - "Constants": [ - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "c" - } - ], - "Formula": "(a / b * c) / 1000000000", - "Category": "Freq", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "cpu_utilization", - "LegacyName": "metric_CPU utilization %", - "Level": 1, - "BriefDescription": "Percentage of time spent in the active CPU power state C0", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "a" - }, - { - "Name": "TSC", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * a / b", - "Category": "Util", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "cpi", - "LegacyName": "metric_CPI", - "Level": 1, - "BriefDescription": "Cycles per instruction retired; indicating how much time each executed instruction took; in units of cycles.", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "CPI", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "loads_per_instr", - "LegacyName": "metric_loads per instr", - "Level": 1, - "BriefDescription": "The ratio of number of completed memory load instructions to the total number completed instructions", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "MEM_INST_RETIRED.ALL_LOADS", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "stores_per_instr", - "LegacyName": "metric_stores per instr", - "Level": 1, - "BriefDescription": "The ratio of number of completed memory store instructions to the total number completed instructions", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "MEM_INST_RETIRED.ALL_STORES", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "l1d_mpi", - "LegacyName": "metric_L1D MPI (includes data+rfo w/ prefetches)", - "Level": 1, - "BriefDescription": "Ratio of number of requests missing L1 data cache (includes data+rfo w/ prefetches) to the total number of completed instructions", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "L1D.REPLACEMENT", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "MPI, D-side", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "l1d_demand_data_read_hits_per_instr", - "LegacyName": "metric_L1D demand data read hits per instr", - "Level": 1, - "BriefDescription": "Ratio of number of demand load requests hitting in L1 data cache to the total number of completed instructions ", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "MEM_LOAD_RETIRED.L1_HIT", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "D-side", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "l1_i_code_read_misses_with_prefetches_per_instr", - "LegacyName": "metric_L1-I code read misses (w/ prefetches) per instr", - "Level": 1, - "BriefDescription": "Ratio of number of code read requests missing in L1 instruction cache (includes prefetches) to the total number of completed instructions", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "L2_RQSTS.ALL_CODE_RD", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "MPI, I-side", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "l2_demand_data_read_hits_per_instr", - "LegacyName": "metric_L2 demand data read hits per instr", - "Level": 1, - "BriefDescription": "Ratio of number of completed demand load requests hitting in L2 cache to the total number of completed instructions ", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "MEM_LOAD_RETIRED.L2_HIT", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "D-side", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "l2_mpi", - "LegacyName": "metric_L2 MPI (includes code+data+rfo w/ prefetches)", - "Level": 1, - "BriefDescription": "Ratio of number of requests missing L2 cache (includes code+data+rfo w/ prefetches) to the total number of completed instructions", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "L2_LINES_IN.ALL", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "MPI", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "l2_demand_data_read_mpi", - "LegacyName": "metric_L2 demand data read MPI", - "Level": 1, - "BriefDescription": "Ratio of number of completed data read request missing L2 cache to the total number of completed instructions", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "MEM_LOAD_RETIRED.L2_MISS", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "MPI, D-side", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "l2_demand_code_mpi", - "LegacyName": "metric_L2 demand code MPI", - "Level": 1, - "BriefDescription": "Ratio of number of code read request missing L2 cache to the total number of completed instructions", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "L2_RQSTS.CODE_RD_MISS", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "MPI, I-side", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "llc_data_read_mpi_demand_plus_prefetch", - "LegacyName": "metric_LLC data read MPI (demand+prefetch)", - "Level": 1, - "BriefDescription": "Ratio of number of data read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_LLCPREFDATA", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD", - "Alias": "b" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF", - "Alias": "c" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "(a + b + c) / d", - "Category": "MPI, D-side", - "ResolutionLevels": "SOCKET, SYSTEM" - }, - { - "MetricName": "llc_code_read_mpi_demand_plus_prefetch", - "LegacyName": "metric_LLC code read MPI (demand+prefetch)", - "Level": 1, - "BriefDescription": "Ratio of number of code read requests missing last level core cache (includes demand w/ prefetches) to the total number of completed instructions", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_CRD", - "Alias": "b" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "b / d", - "Category": "MPI, I-side", - "ResolutionLevels": "SOCKET, SYSTEM" - }, - { - "MetricName": "llc_demand_data_read_miss_latency", - "LegacyName": "metric_Average LLC demand data read miss latency (in ns)", - "Level": 1, - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) in nano seconds", - "UnitOfMeasure": "ns", - "Events": [ - { - "Name": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD", - "Alias": "b" - }, - { - "Name": "UNC_CHA_CLOCKTICKS", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "CHAS_PER_SOCKET", - "Alias": "d" - }, - { - "Name": "SOCKET_COUNT", - "Alias": "socket_count" - } - ], - "Formula": "( 1000000000 * (a / b) / (c / (d * socket_count) ) ) * DURATIONTIMEINSECONDS", - "Category": "Latency", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "llc_demand_data_read_miss_latency_for_local_requests", - "LegacyName": "metric_Average LLC demand data read miss latency for LOCAL requests (in ns)", - "Level": 1, - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to local memory in nano seconds", - "UnitOfMeasure": "ns", - "Events": [ - { - "Name": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_LOCAL", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL", - "Alias": "b" - }, - { - "Name": "UNC_CHA_CLOCKTICKS", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "CHAS_PER_SOCKET", - "Alias": "d" - }, - { - "Name": "SOCKET_COUNT", - "Alias": "socket_count" - } - ], - "Formula": "( 1000000000 * (a / b) / (c / (d * socket_count) ) ) * DURATIONTIMEINSECONDS", - "Category": "Latency", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "llc_demand_data_read_miss_latency_for_remote_requests", - "LegacyName": "metric_Average LLC demand data read miss latency for REMOTE requests (in ns)", - "Level": 1, - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to remote memory in nano seconds", - "UnitOfMeasure": "ns", - "Events": [ - { - "Name": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE", - "Alias": "b" - }, - { - "Name": "UNC_CHA_CLOCKTICKS", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "CHAS_PER_SOCKET", - "Alias": "d" - }, - { - "Name": "SOCKET_COUNT", - "Alias": "socket_count" - } - ], - "Formula": "( 1000000000 * (a / b) / (c / (d * socket_count) ) ) * DURATIONTIMEINSECONDS", - "Category": "Latency", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "llc_demand_data_read_miss_to_pmem_latency", - "LegacyName": "metric_Average LLC demand data read miss to DCPMEM latency (in ns)", - "Level": 1, - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to Intel(R) Optane(TM) Persistent Memory(PMEM) in nano seconds", - "UnitOfMeasure": "ns", - "Events": [ - { - "Name": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PMM", - "Alias": "b" - }, - { - "Name": "UNC_CHA_CLOCKTICKS", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "CHAS_PER_SOCKET", - "Alias": "d" - }, - { - "Name": "SOCKET_COUNT", - "Alias": "socket_count" - } - ], - "Formula": "( 1000000000 * (a / b) / (c / (d * socket_count) ) ) * DURATIONTIMEINSECONDS", - "Category": "Latency", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "llc_demand_data_read_miss_to_dram_latency", - "LegacyName": "metric_Average LLC demand data read miss to DRAM latency (in ns)", - "Level": 1, - "BriefDescription": "Average latency of a last level cache (LLC) demand data read miss (read memory access) addressed to DRAM in nano seconds", - "UnitOfMeasure": "ns", - "Events": [ - { - "Name": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR", - "Alias": "b" - }, - { - "Name": "UNC_CHA_CLOCKTICKS", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "CHAS_PER_SOCKET", - "Alias": "d" - }, - { - "Name": "SOCKET_COUNT", - "Alias": "socket_count" - } - ], - "Formula": "( 1000000000 * (a / b) / (c / (d * socket_count) ) ) * DURATIONTIMEINSECONDS", - "Category": "Latency", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "itlb_2nd_level_mpi", - "LegacyName": "metric_ITLB (2nd level) MPI", - "Level": 1, - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the ITLB (Instruction TLB) and further levels of TLB.", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "ITLB_MISSES.WALK_COMPLETED", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "MPI, I-side", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "itlb_2nd_level_large_page_mpi", - "LegacyName": "metric_ITLB (2nd level) large page MPI", - "Level": 1, - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte and 4 megabyte page sizes) caused by a code fetch to the total number of completed instructions. This implies it missed in the Instruction Translation Lookaside Buffer (ITLB) and further levels of TLB.", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "ITLB_MISSES.WALK_COMPLETED_2M_4M", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "MPI", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "dtlb_2nd_level_load_mpi", - "LegacyName": "metric_DTLB (2nd level) load MPI", - "Level": 1, - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB.", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "DTLB_LOAD_MISSES.WALK_COMPLETED", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "MPI, D-side", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "dtlb_2nd_level_2mb_large_page_load_mpi", - "LegacyName": "metric_DTLB (2nd level) 2MB large page load MPI", - "Level": 1, - "BriefDescription": "Ratio of number of completed page walks (for 2 megabyte page sizes) caused by demand data loads to the total number of completed instructions. This implies it missed in the Data Translation Lookaside Buffer (DTLB) and further levels of TLB.", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "MPI, D-side", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "dtlb_2nd_level_store_mpi", - "LegacyName": "metric_DTLB (2nd level) store MPI", - "Level": 1, - "BriefDescription": "Ratio of number of completed page walks (for all page sizes) caused by demand data stores to the total number of completed instructions. This implies it missed in the DTLB and further levels of TLB.", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "DTLB_STORE_MISSES.WALK_COMPLETED", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "Category": "MPI, D-side", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "numa_reads_addressed_to_local_dram", - "LegacyName": "metric_NUMA %_Reads addressed to local DRAM", - "Level": 1, - "BriefDescription": "Memory read that miss the last level cache (LLC) addressed to local DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL", - "Alias": "b" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE", - "Alias": "c" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * (a + b) / (a + b + c + d)", - "Category": "NUMA", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "numa_reads_addressed_to_remote_dram", - "LegacyName": "metric_NUMA %_Reads addressed to remote DRAM", - "Level": 1, - "BriefDescription": "Memory reads that miss the last level cache (LLC) addressed to remote DRAM as a percentage of total memory read accesses, does not include LLC prefetches.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_LOCAL", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_LOCAL", - "Alias": "b" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE", - "Alias": "c" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PREF_REMOTE", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * (c + d) / (a + b + c + d)", - "Category": "NUMA", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "uncore_frequency", - "LegacyName": "metric_uncore frequency GHz", - "Level": 1, - "BriefDescription": "Uncore operating frequency in GHz", - "UnitOfMeasure": "GHz", - "Events": [ - { - "Name": "UNC_CHA_CLOCKTICKS", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "CHAS_PER_SOCKET", - "Alias": "b" - }, - { - "Name": "SOCKET_COUNT", - "Alias": "socket_count" - } - ], - "Formula": "(a / (b * socket_count) / 1000000000) / DURATIONTIMEINSECONDS", - "Category": "Freq", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "upi_data_transmit_bw", - "LegacyName": "metric_UPI Data transmit BW (MB/sec) (only data)", - "Level": 1, - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data transmit bandwidth (MB/sec)", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_UPI_TxL_FLITS.ALL_DATA", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * (64 / 9.0) / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW", - "ResolutionLevels": "UPI, SOCKET, SYSTEM" - }, - { - "MetricName": "memory_bandwidth_read", - "LegacyName": "metric_memory bandwidth read (MB/sec)", - "Level": 1, - "BriefDescription": "DDR memory read bandwidth (MB/sec)", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_M_CAS_COUNT.RD", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW, MC", - "ResolutionLevels": "CHANNEL, IMC, SOCKET, SYSTEM" - }, - { - "MetricName": "memory_bandwidth_write", - "LegacyName": "metric_memory bandwidth write (MB/sec)", - "Level": 1, - "BriefDescription": "DDR memory write bandwidth (MB/sec)", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_M_CAS_COUNT.WR", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW, MC", - "ResolutionLevels": "CHANNEL, IMC, SOCKET, SYSTEM" - }, - { - "MetricName": "memory_bandwidth_total", - "LegacyName": "metric_memory bandwidth total (MB/sec)", - "Level": 1, - "BriefDescription": "DDR memory bandwidth (MB/sec)", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_M_CAS_COUNT.RD", - "Alias": "a" - }, - { - "Name": "UNC_M_CAS_COUNT.WR", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "((a + b) * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW, MC", - "ResolutionLevels": "CHANNEL, IMC, SOCKET, SYSTEM" - }, - { - "MetricName": "memory_extra_write_bw_due_to_directory_updates", - "LegacyName": "metric_memory extra write b/w due to directory updates (MB/sec)", - "Level": 1, - "BriefDescription": "Memory write bandwidth (MB/sec) caused by directory updates; includes DDR and Intel(R) Optane(TM) Persistent Memory(PMEM).", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_CHA_DIR_UPDATE.HA", - "Alias": "a" - }, - { - "Name": "UNC_CHA_DIR_UPDATE.TOR", - "Alias": "b" - }, - { - "Name": "UNC_M2M_DIRECTORY_UPDATE.ANY", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "((a + b + c) * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW, MC", - "ResolutionLevels": "SOCKET, SYSTEM" - }, - { - "MetricName": "pmem_memory_bandwidth_read", - "LegacyName": "metric_DCPMEM_memory bandwidth read (MB/sec)", - "Level": 1, - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory read bandwidth (MB/sec)", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_M_PMM_RPQ_INSERTS", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "", - "ResolutionLevels": "CHANNEL, IMC, SOCKET, SYSTEM" - }, - { - "MetricName": "pmem_memory_bandwidth_write", - "LegacyName": "metric_DCPMEM_memory bandwidth write (MB/sec)", - "Level": 1, - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory write bandwidth (MB/sec)", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_M_PMM_WPQ_INSERTS", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "", - "ResolutionLevels": "CHANNEL, IMC, SOCKET, SYSTEM" - }, - { - "MetricName": "pmem_memory_bandwidth_total", - "LegacyName": "metric_DCPMEM_memory bandwidth total (MB/sec)", - "Level": 1, - "BriefDescription": "Intel(R) Optane(TM) Persistent Memory(PMEM) memory bandwidth (MB/sec)", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_M_PMM_RPQ_INSERTS", - "Alias": "a" - }, - { - "Name": "UNC_M_PMM_WPQ_INSERTS", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "((a + b) * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "", - "ResolutionLevels": "CHANNEL, IMC, SOCKET, SYSTEM" - }, - { - "MetricName": "io_bandwidth_read", - "LegacyName": "metric_IO_bandwidth_disk_or_network_writes (MB/sec)", - "Level": 1, - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU.", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW, IO", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "io_bandwidth_write", - "LegacyName": "metric_IO_bandwidth_disk_or_network_reads (MB/sec)", - "Level": 1, - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU.", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IO_ITOM", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "((a + b) * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW, IO", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "io_percent_of_inbound_reads_that_miss_l3", - "LegacyName": "metric_IO % of inbound reads that miss L3", - "Level": 1, - "BriefDescription": "Percentage of inbound reads initiated by end device controllers that miss the L3 cache.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IO_MISS_PCIRDCUR", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * a / b", - "Category": "IO", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "io_percent_of_inbound_partial_writes_that_miss_l3", - "LegacyName": "metric_IO % of inbound partial writes that miss L3", - "Level": 1, - "BriefDescription": "Percentage of inbound partial cacheline writes initiated by end device controllers that miss the L3 cache.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IO_MISS_ITOMCACHENEAR", - "Alias": "b" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IO_RFO", - "Alias": "c" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IO_MISS_RFO", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * ((b + d) / (a + c) )", - "Category": "IO", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "io_percent_of_inbound_full_writes_that_miss_l3", - "LegacyName": "metric_IO % of inbound full writes that miss L3", - "Level": 1, - "BriefDescription": "Percentage of inbound full cacheline writes initiated by end device controllers that miss the L3 cache.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IO_ITOM", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IO_MISS_ITOM", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * (b / a)", - "Category": "IO", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "percent_uops_delivered_from_decoded_icache", - "LegacyName": "metric_% Uops delivered from decoded Icache (DSB)", - "Level": 1, - "BriefDescription": "Uops delivered from decoded instruction cache (decoded stream buffer or DSB) as a percent of total uops delivered to Instruction Decode Queue", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "IDQ.DSB_UOPS", - "Alias": "a" - }, - { - "Name": "IDQ.MITE_UOPS", - "Alias": "b" - }, - { - "Name": "IDQ.MS_UOPS", - "Alias": "c" - }, - { - "Name": "LSD.UOPS", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * (a / (a + b + c + d) )", - "Category": "I-side", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "percent_uops_delivered_from_legacy_decode_pipeline", - "LegacyName": "metric_% Uops delivered from legacy decode pipeline (MITE)", - "Level": 1, - "BriefDescription": "Uops delivered from legacy decode pipeline (Micro-instruction Translation Engine or MITE) as a percent of total uops delivered to Instruction Decode Queue", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "IDQ.DSB_UOPS", - "Alias": "a" - }, - { - "Name": "IDQ.MITE_UOPS", - "Alias": "b" - }, - { - "Name": "IDQ.MS_UOPS", - "Alias": "c" - }, - { - "Name": "LSD.UOPS", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * (b / (a + b + c + d) )", - "Category": "I-side", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "percent_uops_delivered_from_microcode_sequencer", - "LegacyName": "metric_% Uops delivered from microcode sequencer (MS)", - "Level": 1, - "BriefDescription": "Uops delivered from microcode sequencer (MS) as a percent of total uops delivered to Instruction Decode Queue", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "IDQ.DSB_UOPS", - "Alias": "a" - }, - { - "Name": "IDQ.MITE_UOPS", - "Alias": "b" - }, - { - "Name": "IDQ.MS_UOPS", - "Alias": "c" - }, - { - "Name": "LSD.UOPS", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * (c / (a + b + c + d) )", - "Category": "I-side", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM" - }, - { - "MetricName": "io_bandwidth_read_local", - "LegacyName": "metric_IO bandwidth read local (MB/sec)", - "Level": 1, - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the local CPU socket.", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_LOCAL", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW,IO", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "io_bandwidth_read_remote", - "LegacyName": "metric_IO bandwidth read remote (MB/sec)", - "Level": 1, - "BriefDescription": "Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from a remote CPU socket.", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR_REMOTE", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW,IO", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "io_bandwidth_write_local", - "LegacyName": "metric_IO bandwidth write local (MB/sec)", - "Level": 1, - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the local CPU socket.", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IO_ITOM_LOCAL", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_LOCAL", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "((a + b) * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW,IO", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "io_bandwidth_write_remote", - "LegacyName": "metric_IO bandwidth write remote (MB/sec)", - "Level": 1, - "BriefDescription": "Bandwidth of IO writes that are initiated by end device controllers that are writing memory to a remote CPU socket.", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IO_ITOM_REMOTE", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR_REMOTE", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "((a + b) * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW,IO", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "llc_miss_local_memory_bandwidth_read", - "LegacyName": "metric_llc_miss_local_memory_bandwidth_read_MB/s", - "Level": 1, - "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to local memory.", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_CHA_REQUESTS.READS_LOCAL", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "llc_miss_local_memory_bandwidth_write", - "LegacyName": "metric_llc_miss_local_memory_bandwidth_write_MB/s", - "Level": 1, - "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to local memory.", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_CHA_REQUESTS.WRITES_LOCAL", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "llc_miss_remote_memory_bandwidth_read", - "LegacyName": "metric_llc_miss_remote_memory_bandwidth_read_MB/s", - "Level": 1, - "BriefDescription": "Bandwidth (MB/sec) of read requests that miss the last level cache (LLC) and go to remote memory.", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_CHA_REQUESTS.READS_REMOTE", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "llc_miss_remote_memory_bandwidth_write", - "LegacyName": "metric_llc_miss_remote_memory_bandwidth_write_MB/s", - "Level": 1, - "BriefDescription": "Bandwidth (MB/sec) of write requests that miss the last level cache (LLC) and go to remote memory.", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_CHA_REQUESTS.WRITES_REMOTE", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * 64 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW", - "ResolutionLevels": "CHA, SOCKET, SYSTEM" - }, - { - "MetricName": "upi_data_receive_bw", - "LegacyName": "metric_UPI Data receive BW (MB/sec) (only data)", - "Level": 1, - "BriefDescription": "Intel(R) Ultra Path Interconnect (UPI) data receive bandwidth (MB/sec)", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_UPI_RxL_FLITS.ALL_DATA", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * (64 / 9.0) / 1000000) / DURATIONTIMEINSECONDS", - "Category": "BW", - "ResolutionLevels": "UPI, SOCKET, SYSTEM" - }, - { - "MetricName": "iio_bandwidth_read", - "LegacyName": "metric_IIO_bandwidth_read (MB/sec)", - "Level": 1, - "BriefDescription": "Bandwidth observed by the integrated I/O traffic controller (IIO) of IO reads that are initiated by end device controllers that are requesting memory from the CPU.", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_IIO_DATA_REQ_OF_CPU.MEM_READ.ALL_PARTS", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * 4 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "IO, BW", - "ResolutionLevels": "IIO, SOCKET, SYSTEM" - }, - { - "MetricName": "iio_bandwidth_write", - "LegacyName": "metric_IIO_bandwidth_write (MB/sec)", - "Level": 1, - "BriefDescription": "Bandwidth observed by the integrated I/O traffic controller (IIO) of IO writes that are initiated by end device controllers that are writing memory to the CPU.", - "UnitOfMeasure": "MB/sec", - "Events": [ - { - "Name": "UNC_IIO_DATA_REQ_OF_CPU.MEM_WRITE.ALL_PARTS", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "(a * 4 / 1000000) / DURATIONTIMEINSECONDS", - "Category": "IO, BW", - "ResolutionLevels": "IIO, SOCKET, SYSTEM" - }, - { - "MetricName": "Bottleneck_Mispredictions", - "LegacyName": "metric_TMA_Bottleneck_Mispredictions", - "Level": 1, - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UOPS_RETIRED.MS", - "Alias": "a" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BRANCH_MISPREDICTS", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "e" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "f" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "g" - }, - { - "Name": "BR_MISP_RETIRED.ALL_BRANCHES", - "Alias": "h" - }, - { - "Name": "INT_MISC.CLEARS_COUNT", - "Alias": "i" - }, - { - "Name": "MACHINE_CLEARS.COUNT", - "Alias": "j" - }, - { - "Name": "PERF_METRICS.FETCH_LATENCY", - "Alias": "k" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "l" - }, - { - "Name": "INT_MISC.CLEAR_RESTEER_CYCLES", - "Alias": "m" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "n" - }, - { - "Name": "ICACHE_DATA.STALLS", - "Alias": "o" - }, - { - "Name": "ICACHE_TAG.STALLS", - "Alias": "p" - }, - { - "Name": "INT_MISC.UNKNOWN_BRANCH_CYCLES", - "Alias": "q" - }, - { - "Name": "UOPS_RETIRED.MS:c1:e1", - "Alias": "r" - }, - { - "Name": "UOPS_RETIRED.SLOTS", - "Alias": "s" - }, - { - "Name": "UOPS_ISSUED.ANY", - "Alias": "t" - }, - { - "Name": "DECODE.LCP", - "Alias": "u" - }, - { - "Name": "DSB2MITE_SWITCHES.PENALTY_CYCLES", - "Alias": "v" - } - ], - "Constants": [], - "Formula": "100 * ( 1 - ( 10 * ( a / ( b ) ) * ( max( ( c / ( d + e + f + g ) ) * ( 1 - h / ( i - j ) ) , 0.0001 ) ) / ( c / ( d + e + f + g ) ) ) ) * ( ( c / ( d + e + f + g ) ) + ( ( k / ( d + e + f + g ) - l / ( b ) ) ) * ( ( ( c / ( d + e + f + g ) ) / ( max( 1 - ( ( d / ( d + e + f + g ) - l / ( b ) ) + ( g / ( d + e + f + g ) ) + ( f / ( d + e + f + g ) ) ) , 0 ) ) ) * m / ( n ) ) / ( ( o / ( n ) ) + ( p / ( n ) ) + ( m / ( n ) + ( q / ( n ) ) ) + ( ( 3 ) * r / ( s / t ) / ( n ) ) + ( u / ( n ) ) + ( v / ( n ) ) ) )", - "BaseFormula": " 100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "Formula": "metric_TMA_Bottleneck_Mispredictions > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP", - "LocateWith": "" - }, - { - "MetricName": "Bottleneck_Big_Code", - "LegacyName": "metric_TMA_Bottleneck_Big_Code", - "Level": 1, - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.FETCH_LATENCY", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - }, - { - "Name": "ICACHE_TAG.STALLS", - "Alias": "h" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "i" - }, - { - "Name": "ICACHE_DATA.STALLS", - "Alias": "j" - }, - { - "Name": "INT_MISC.UNKNOWN_BRANCH_CYCLES", - "Alias": "k" - }, - { - "Name": "INT_MISC.CLEAR_RESTEER_CYCLES", - "Alias": "l" - }, - { - "Name": "UOPS_RETIRED.MS:c1:e1", - "Alias": "m" - }, - { - "Name": "UOPS_RETIRED.SLOTS", - "Alias": "n" - }, - { - "Name": "UOPS_ISSUED.ANY", - "Alias": "o" - }, - { - "Name": "DECODE.LCP", - "Alias": "p" - }, - { - "Name": "DSB2MITE_SWITCHES.PENALTY_CYCLES", - "Alias": "q" - } - ], - "Constants": [], - "Formula": "100 * ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( h / ( i ) ) + ( j / ( i ) ) + ( k / ( i ) ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) )", - "BaseFormula": " 100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "Formula": "metric_TMA_Bottleneck_Big_Code > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Bottleneck_Instruction_Fetch_BW", - "LegacyName": "metric_TMA_Bottleneck_Instruction_Fetch_BW", - "Level": 1, - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "a" - }, - { - "Name": "IDQ.MITE_CYCLES_ANY", - "Alias": "a_a" - }, - { - "Name": "IDQ.MITE_CYCLES_OK", - "Alias": "a_b" - }, - { - "Name": "IDQ.DSB_CYCLES_ANY", - "Alias": "a_c" - }, - { - "Name": "IDQ.DSB_CYCLES_OK", - "Alias": "a_d" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "e" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "f" - }, - { - "Name": "UOPS_RETIRED.MS", - "Alias": "g" - }, - { - "Name": "PERF_METRICS.BRANCH_MISPREDICTS", - "Alias": "h" - }, - { - "Name": "BR_MISP_RETIRED.ALL_BRANCHES", - "Alias": "i" - }, - { - "Name": "INT_MISC.CLEARS_COUNT", - "Alias": "j" - }, - { - "Name": "MACHINE_CLEARS.COUNT", - "Alias": "k" - }, - { - "Name": "PERF_METRICS.FETCH_LATENCY", - "Alias": "l" - }, - { - "Name": "INT_MISC.CLEAR_RESTEER_CYCLES", - "Alias": "m" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "n" - }, - { - "Name": "ICACHE_DATA.STALLS", - "Alias": "o" - }, - { - "Name": "ICACHE_TAG.STALLS", - "Alias": "p" - }, - { - "Name": "INT_MISC.UNKNOWN_BRANCH_CYCLES", - "Alias": "q" - }, - { - "Name": "UOPS_RETIRED.MS:c1:e1", - "Alias": "r" - }, - { - "Name": "UOPS_RETIRED.SLOTS", - "Alias": "s" - }, - { - "Name": "UOPS_ISSUED.ANY", - "Alias": "t" - }, - { - "Name": "DECODE.LCP", - "Alias": "u" - }, - { - "Name": "DSB2MITE_SWITCHES.PENALTY_CYCLES", - "Alias": "v" - }, - { - "Name": "INST_RETIRED.REP_ITERATION", - "Alias": "w" - }, - { - "Name": "UOPS_RETIRED.MS:c1", - "Alias": "x" - }, - { - "Name": "IDQ.MS_CYCLES_ANY", - "Alias": "y" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "z" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( a / ( a + b + c + d ) - e / ( f ) ) - ( 1 - ( 10 * ( g / ( f ) ) * ( max( ( h / ( a + b + c + d ) ) * ( 1 - i / ( j - k ) ) , 0.0001 ) ) / ( h / ( a + b + c + d ) ) ) ) * ( ( l / ( a + b + c + d ) - e / ( f ) ) ) * ( ( ( h / ( a + b + c + d ) ) / ( max( 1 - ( ( a / ( a + b + c + d ) - e / ( f ) ) + ( d / ( a + b + c + d ) ) + ( c / ( a + b + c + d ) ) ) , 0 ) ) ) * m / ( n ) ) / ( ( o / ( n ) ) + ( p / ( n ) ) + ( m / ( n ) + ( q / ( n ) ) ) + ( ( 3 ) * r / ( s / t ) / ( n ) ) + ( u / ( n ) ) + ( v / ( n ) ) ) - ( ( 1 - w / x ) * ( ( ( l / ( a + b + c + d ) - e / ( f ) ) ) * ( ( ( 3 ) * r / ( s / t ) / ( n ) ) + ( m / ( n ) + ( q / ( n ) ) ) * ( ( ( 1 - ( ( h / ( a + b + c + d ) ) / ( max( 1 - ( ( a / ( a + b + c + d ) - e / ( f ) ) + ( d / ( a + b + c + d ) ) + ( c / ( a + b + c + d ) ) ) , 0 ) ) ) ) * m / ( n ) ) + ( ( ( h / ( a + b + c + d ) ) / ( max( 1 - ( ( a / ( a + b + c + d ) - e / ( f ) ) + ( d / ( a + b + c + d ) ) + ( c / ( a + b + c + d ) ) ) , 0 ) ) ) * m / ( n ) ) * ( max( ( h / ( a + b + c + d ) ) * ( 1 - i / ( j - k ) ) , 0.0001 ) ) / ( h / ( a + b + c + d ) ) ) / ( ( ( ( h / ( a + b + c + d ) ) / ( max( 1 - ( ( a / ( a + b + c + d ) - e / ( f ) ) + ( d / ( a + b + c + d ) ) + ( c / ( a + b + c + d ) ) ) , 0 ) ) ) * m / ( n ) ) + ( ( 1 - ( ( h / ( a + b + c + d ) ) / ( max( 1 - ( ( a / ( a + b + c + d ) - e / ( f ) ) + ( d / ( a + b + c + d ) ) + ( c / ( a + b + c + d ) ) ) , 0 ) ) ) ) * m / ( n ) ) + ( q / ( n ) ) ) ) / ( ( o / ( n ) ) + ( p / ( n ) ) + ( m / ( n ) + ( q / ( n ) ) ) + ( ( 3 ) * r / ( s / t ) / ( n ) ) + ( u / ( n ) ) + ( v / ( n ) ) ) + ( max( 0 , ( a / ( a + b + c + d ) - e / ( f ) ) - ( ( l / ( a + b + c + d ) - e / ( f ) ) ) ) ) * ( max( y , x / ( s / t ) ) / ( z if smt_on else ( n ) ) / 2 ) / ( ( ( a_a - a_b ) / ( z if smt_on else ( n ) ) / 2 ) + ( ( a_c - a_d ) / ( z if smt_on else ( n ) ) / 2 ) + ( max( y , x / ( s / t ) ) / ( z if smt_on else ( n ) ) / 2 ) ) ) ) ) - ( 100 * ( ( l / ( a + b + c + d ) - e / ( f ) ) ) * ( ( p / ( n ) ) + ( o / ( n ) ) + ( q / ( n ) ) ) / ( ( o / ( n ) ) + ( p / ( n ) ) + ( m / ( n ) + ( q / ( n ) ) ) + ( ( 3 ) * r / ( s / t ) / ( n ) ) + ( u / ( n ) ) + ( v / ( n ) ) ) )", - "BaseFormula": " 100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( 1 - inst_retired.rep_iteration / uops_retired.ms:c1 ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_ms ) ) ) ) - tma_big_code", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "Formula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvFB;Fed;FetchBW;Frontend", - "LocateWith": "" - }, - { - "MetricName": "Bottleneck_Cache_Memory_Bandwidth", - "LegacyName": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth", - "Level": 1, - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.MEMORY_BOUND", - "Alias": "a" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD", - "Alias": "a_a" - }, - { - "Name": "MEM_LOAD_RETIRED.L3_HIT", - "Alias": "a_b" - }, - { - "Name": "L1D_PEND_MISS.FB_FULL", - "Alias": "a_c" - }, - { - "Name": "DTLB_LOAD_MISSES.STLB_HIT:c1", - "Alias": "a_d" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_ACTIVE", - "Alias": "a_e" - }, - { - "Name": "CYCLE_ACTIVITY.CYCLES_MEM_ANY", - "Alias": "a_f" - }, - { - "Name": "MEMORY_ACTIVITY.CYCLES_L1D_MISS", - "Alias": "a_g" - }, - { - "Name": "LD_BLOCKS.STORE_FORWARD", - "Alias": "a_h" - }, - { - "Name": "MEM_INST_RETIRED.ALL_LOADS", - "Alias": "a_i" - }, - { - "Name": "MEM_INST_RETIRED.LOCK_LOADS", - "Alias": "a_k" - }, - { - "Name": "L2_RQSTS.ALL_RFO", - "Alias": "a_l" - }, - { - "Name": "MEM_INST_RETIRED.ALL_STORES", - "Alias": "a_m" - }, - { - "Name": "L2_RQSTS.RFO_HIT", - "Alias": "a_n" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO", - "Alias": "a_o" - }, - { - "Name": "L1D_PEND_MISS.PENDING", - "Alias": "a_p" - }, - { - "Name": "MEM_LOAD_COMPLETED.L1_MISS_ANY", - "Alias": "a_q" - }, - { - "Name": "LD_BLOCKS.NO_SR", - "Alias": "a_r" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L3_MISS", - "Alias": "f" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "g" - }, - { - "Name": "OCR.DEMAND_DATA_RD.PMM", - "Alias": "h" - }, - { - "Name": "OCR.READS_TO_CORE.L3_MISS", - "Alias": "i" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_LOADS", - "Alias": "j" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L1D_MISS", - "Alias": "k" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L2_MISS", - "Alias": "l" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_STORES", - "Alias": "m" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD:c4", - "Alias": "n" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", - "Alias": "o" - }, - { - "Name": "XQ.FULL_CYCLES", - "Alias": "p" - }, - { - "Name": "L1D_PEND_MISS.L2_STALLS", - "Alias": "q" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "r" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD", - "Alias": "u" - }, - { - "Name": "OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM", - "Alias": "v" - }, - { - "Name": "OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD", - "Alias": "w" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS", - "Alias": "x" - }, - { - "Name": "MEM_LOAD_RETIRED.FB_HIT", - "Alias": "y" - }, - { - "Name": "MEM_LOAD_RETIRED.L1_MISS", - "Alias": "z" - } - ], - "Constants": [ - { - "Name": "20", - "Alias": "dependentloadsweight" - }, - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - }, - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "s" - } - ], - "Formula": "( 100 * ( ( ( a / ( b + c + d + e ) ) * ( ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( ( min( g , n ) ) / ( g ) ) / ( ( ( min( g , n ) ) / ( g ) ) + ( ( min( g , o ) ) / ( g ) - ( ( min( g , n ) ) / ( g ) ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( f / ( g ) ) * h / i ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( min( g , n ) ) / ( g ) ) / ( ( ( min( g , n ) ) / ( g ) ) + ( ( min( g , o ) ) / ( g ) - ( ( min( g , n ) ) / ( g ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( ( l - f ) / ( g ) ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( ( p + q ) / ( g ) ) / ( ( ( ( ( 81 * ( ( ( g ) / r ) * s / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / r ) * s / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( u * ( v / ( v + w ) ) ) + ( ( 79 * ( ( ( g ) / r ) * s / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / r ) * s / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( x ) ) * ( 1 + ( y / z ) / 2 ) / ( g ) ) + ( ( ( 79 * ( ( ( g ) / r ) * s / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / r ) * s / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_a + u * ( 1 - ( v / ( v + w ) ) ) ) * ( 1 + ( y / z ) / 2 ) / ( g ) ) + ( ( ( 37 * ( ( ( g ) / r ) * s / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / r ) * s / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_b * ( 1 + ( y / z ) / 2 ) ) / ( g ) ) + ( ( p + q ) / ( g ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( max( ( j - k ) / ( g ) , 0 ) ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( a_c / ( g ) ) / ( ( min( ( 7 ) * a_d + a_e , max( a_f - a_g , 0 ) ) / ( g ) ) + ( 13 * a_h / ( g ) ) + ( min( 2 * ( a_i - y - z ) * dependentloadsweight / 100 , max( a_f - a_g , 0 ) ) / ( g ) ) + ( ( 16 * max( 0 , a_k - a_l ) + ( a_k / a_m ) * ( ( 10 ) * a_n + ( min( g , a_o ) ) ) ) / ( g ) ) + ( ( a_p / a_q ) * a_r / ( g ) ) + ( a_c / ( g ) ) ) ) ) ) )", - "BaseFormula": " ( 100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * tma_hbm_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) ) )", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth" - } - ], - "Formula": "a > 20", - "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20", - "ThresholdIssues": "$issueBW" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Bottleneck_Cache_Memory_Latency", - "LegacyName": "metric_TMA_Bottleneck_Cache_Memory_Latency", - "Level": 1, - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.MEMORY_BOUND", - "Alias": "a" - }, - { - "Name": "XQ.FULL_CYCLES", - "Alias": "a_a" - }, - { - "Name": "L1D_PEND_MISS.L2_STALLS", - "Alias": "a_b" - }, - { - "Name": "MEM_INST_RETIRED.ALL_LOADS", - "Alias": "a_c" - }, - { - "Name": "CYCLE_ACTIVITY.CYCLES_MEM_ANY", - "Alias": "a_e" - }, - { - "Name": "MEMORY_ACTIVITY.CYCLES_L1D_MISS", - "Alias": "a_f" - }, - { - "Name": "DTLB_LOAD_MISSES.STLB_HIT:c1", - "Alias": "a_g" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_ACTIVE", - "Alias": "a_h" - }, - { - "Name": "LD_BLOCKS.STORE_FORWARD", - "Alias": "a_i" - }, - { - "Name": "MEM_INST_RETIRED.LOCK_LOADS", - "Alias": "a_j" - }, - { - "Name": "L2_RQSTS.ALL_RFO", - "Alias": "a_k" - }, - { - "Name": "MEM_INST_RETIRED.ALL_STORES", - "Alias": "a_l" - }, - { - "Name": "L2_RQSTS.RFO_HIT", - "Alias": "a_m" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO", - "Alias": "a_n" - }, - { - "Name": "L1D_PEND_MISS.PENDING", - "Alias": "a_o" - }, - { - "Name": "MEM_LOAD_COMPLETED.L1_MISS_ANY", - "Alias": "a_p" - }, - { - "Name": "LD_BLOCKS.NO_SR", - "Alias": "a_q" - }, - { - "Name": "L1D_PEND_MISS.FB_FULL", - "Alias": "a_r" - }, - { - "Name": "MEM_INST_RETIRED.SPLIT_STORES", - "Alias": "a_s" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "a_t" - }, - { - "Name": "MEM_STORE_RETIRED.L2_HIT", - "Alias": "a_u" - }, - { - "Name": "OCR.DEMAND_RFO.L3_MISS:ocr_msr_val=0x103b800002", - "Alias": "a_v" - }, - { - "Name": "OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM", - "Alias": "a_w" - }, - { - "Name": "OCR.STREAMING_WR.ANY_RESPONSE", - "Alias": "a_x" - }, - { - "Name": "DTLB_STORE_MISSES.STLB_HIT:c1", - "Alias": "a_y" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", - "Alias": "a_z" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L3_MISS", - "Alias": "f" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "g" - }, - { - "Name": "OCR.DEMAND_DATA_RD.PMM", - "Alias": "h" - }, - { - "Name": "OCR.READS_TO_CORE.L3_MISS", - "Alias": "i" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_LOADS", - "Alias": "j" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L1D_MISS", - "Alias": "k" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L2_MISS", - "Alias": "l" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_STORES", - "Alias": "m" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", - "Alias": "n" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD:c4", - "Alias": "o" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "p" - }, - { - "Name": "MEM_LOAD_RETIRED.L3_HIT", - "Alias": "s" - }, - { - "Name": "MEM_LOAD_RETIRED.FB_HIT", - "Alias": "t" - }, - { - "Name": "MEM_LOAD_RETIRED.L1_MISS", - "Alias": "u" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD", - "Alias": "v" - }, - { - "Name": "OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM", - "Alias": "w" - }, - { - "Name": "OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD", - "Alias": "x" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS", - "Alias": "y" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD", - "Alias": "z" - } - ], - "Constants": [ - { - "Name": "20", - "Alias": "dependentloadsweight" - }, - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - }, - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "q" - }, - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( ( a / ( b + c + d + e ) ) * ( ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( ( min( g , n ) ) / ( g ) - ( ( min( g , o ) ) / ( g ) ) ) / ( ( ( min( g , o ) ) / ( g ) ) + ( ( min( g , n ) ) / ( g ) - ( ( min( g , o ) ) / ( g ) ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( f / ( g ) ) * h / i ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( min( g , n ) ) / ( g ) - ( ( min( g , o ) ) / ( g ) ) ) / ( ( ( min( g , o ) ) / ( g ) ) + ( ( min( g , n ) ) / ( g ) - ( ( min( g , o ) ) / ( g ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( ( l - f ) / ( g ) ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( ( ( 37 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( s * ( 1 + ( t / u ) / 2 ) ) / ( g ) ) / ( ( ( ( ( 81 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( v * ( w / ( w + x ) ) ) + ( ( 79 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( y ) ) * ( 1 + ( t / u ) / 2 ) / ( g ) ) + ( ( ( 79 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( z + v * ( 1 - ( w / ( w + x ) ) ) ) * ( 1 + ( t / u ) / 2 ) / ( g ) ) + ( ( ( 37 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( s * ( 1 + ( t / u ) / 2 ) ) / ( g ) ) + ( ( a_a + a_b ) / ( g ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( k - l ) / ( g ) ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( max( ( j - k ) / ( g ) , 0 ) ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( min( 2 * ( a_c - t - u ) * dependentloadsweight / 100 , max( a_e - a_f , 0 ) ) / ( g ) ) / ( ( min( ( 7 ) * a_g + a_h , max( a_e - a_f , 0 ) ) / ( g ) ) + ( 13 * a_i / ( g ) ) + ( min( 2 * ( a_c - t - u ) * dependentloadsweight / 100 , max( a_e - a_f , 0 ) ) / ( g ) ) + ( ( 16 * max( 0 , a_j - a_k ) + ( a_j / a_l ) * ( ( 10 ) * a_m + ( min( g , a_n ) ) ) ) / ( g ) ) + ( ( a_o / a_p ) * a_q / ( g ) ) + ( a_r / ( g ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( max( ( j - k ) / ( g ) , 0 ) ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( ( 16 * max( 0 , a_j - a_k ) + ( a_j / a_l ) * ( ( 10 ) * a_m + ( min( g , a_n ) ) ) ) / ( g ) ) / ( ( min( ( 7 ) * a_g + a_h , max( a_e - a_f , 0 ) ) / ( g ) ) + ( 13 * a_i / ( g ) ) + ( min( 2 * ( a_c - t - u ) * dependentloadsweight / 100 , max( a_e - a_f , 0 ) ) / ( g ) ) + ( ( 16 * max( 0 , a_j - a_k ) + ( a_j / a_l ) * ( ( 10 ) * a_m + ( min( g , a_n ) ) ) ) / ( g ) ) + ( ( a_o / a_p ) * a_q / ( g ) ) + ( a_r / ( g ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( max( ( j - k ) / ( g ) , 0 ) ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( ( a_o / a_p ) * a_q / ( g ) ) / ( ( min( ( 7 ) * a_g + a_h , max( a_e - a_f , 0 ) ) / ( g ) ) + ( 13 * a_i / ( g ) ) + ( min( 2 * ( a_c - t - u ) * dependentloadsweight / 100 , max( a_e - a_f , 0 ) ) / ( g ) ) + ( ( 16 * max( 0 , a_j - a_k ) + ( a_j / a_l ) * ( ( 10 ) * a_m + ( min( g , a_n ) ) ) ) / ( g ) ) + ( ( a_o / a_p ) * a_q / ( g ) ) + ( a_r / ( g ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( m / ( g ) ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( a_s / ( a_t if smt_on else ( g ) ) ) / ( ( ( ( a_u * ( 10 ) * ( 1 - ( a_j / a_l ) ) ) + ( 1 - ( a_j / a_l ) ) * ( min( g , a_n ) ) ) / ( g ) ) + ( ( ( 170 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_v + ( 81 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_w ) / ( g ) ) + ( a_s / ( a_t if smt_on else ( g ) ) ) + ( 9 * a_x / ( g ) ) + ( ( ( 7 ) * a_y + a_z ) / ( a_t if smt_on else ( g ) ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( m / ( g ) ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( ( ( a_u * ( 10 ) * ( 1 - ( a_j / a_l ) ) ) + ( 1 - ( a_j / a_l ) ) * ( min( g , a_n ) ) ) / ( g ) ) / ( ( ( ( a_u * ( 10 ) * ( 1 - ( a_j / a_l ) ) ) + ( 1 - ( a_j / a_l ) ) * ( min( g , a_n ) ) ) / ( g ) ) + ( ( ( 170 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_v + ( 81 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_w ) / ( g ) ) + ( a_s / ( a_t if smt_on else ( g ) ) ) + ( 9 * a_x / ( g ) ) + ( ( ( 7 ) * a_y + a_z ) / ( a_t if smt_on else ( g ) ) ) ) ) ) )", - "BaseFormula": " 100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * tma_hbm_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Bottleneck_Cache_Memory_Latency" - } - ], - "Formula": "a > 20", - "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20", - "ThresholdIssues": "$issueLat" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvML;Mem;MemoryLat;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Bottleneck_Memory_Data_TLBs", - "LegacyName": "metric_TMA_Bottleneck_Memory_Data_TLBs", - "Level": 1, - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.MEMORY_BOUND", - "Alias": "a" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO", - "Alias": "a_a" - }, - { - "Name": "L1D_PEND_MISS.PENDING", - "Alias": "a_b" - }, - { - "Name": "MEM_LOAD_COMPLETED.L1_MISS_ANY", - "Alias": "a_c" - }, - { - "Name": "LD_BLOCKS.NO_SR", - "Alias": "a_d" - }, - { - "Name": "L1D_PEND_MISS.FB_FULL", - "Alias": "a_e" - }, - { - "Name": "DTLB_STORE_MISSES.STLB_HIT:c1", - "Alias": "a_f" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", - "Alias": "a_g" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "a_h" - }, - { - "Name": "MEM_STORE_RETIRED.L2_HIT", - "Alias": "a_i" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "a_j" - }, - { - "Name": "OCR.DEMAND_RFO.L3_MISS:ocr_msr_val=0x103b800002", - "Alias": "a_m" - }, - { - "Name": "OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM", - "Alias": "a_n" - }, - { - "Name": "MEM_INST_RETIRED.SPLIT_STORES", - "Alias": "a_o" - }, - { - "Name": "OCR.STREAMING_WR.ANY_RESPONSE", - "Alias": "a_p" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_LOADS", - "Alias": "f" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L1D_MISS", - "Alias": "g" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "h" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L2_MISS", - "Alias": "i" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L3_MISS", - "Alias": "j" - }, - { - "Name": "OCR.DEMAND_DATA_RD.PMM", - "Alias": "k" - }, - { - "Name": "OCR.READS_TO_CORE.L3_MISS", - "Alias": "l" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_STORES", - "Alias": "m" - }, - { - "Name": "DTLB_LOAD_MISSES.STLB_HIT:c1", - "Alias": "n" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_ACTIVE", - "Alias": "o" - }, - { - "Name": "CYCLE_ACTIVITY.CYCLES_MEM_ANY", - "Alias": "p" - }, - { - "Name": "MEMORY_ACTIVITY.CYCLES_L1D_MISS", - "Alias": "q" - }, - { - "Name": "LD_BLOCKS.STORE_FORWARD", - "Alias": "r" - }, - { - "Name": "MEM_INST_RETIRED.ALL_LOADS", - "Alias": "s" - }, - { - "Name": "MEM_LOAD_RETIRED.FB_HIT", - "Alias": "t" - }, - { - "Name": "MEM_LOAD_RETIRED.L1_MISS", - "Alias": "u" - }, - { - "Name": "MEM_INST_RETIRED.LOCK_LOADS", - "Alias": "w" - }, - { - "Name": "L2_RQSTS.ALL_RFO", - "Alias": "x" - }, - { - "Name": "MEM_INST_RETIRED.ALL_STORES", - "Alias": "y" - }, - { - "Name": "L2_RQSTS.RFO_HIT", - "Alias": "z" - } - ], - "Constants": [ - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "a_k" - }, - { - "Name": "20", - "Alias": "dependentloadsweight" - }, - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - }, - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( a / ( b + c + d + e ) ) * ( ( max( ( f - g ) / ( h ) , 0 ) ) / max( ( a / ( b + c + d + e ) ) , ( ( max( ( f - g ) / ( h ) , 0 ) ) + ( ( g - i ) / ( h ) ) + ( ( i - j ) / ( h ) ) + ( ( j / ( h ) ) * k / l ) + ( ( j / ( h ) ) - ( ( j / ( h ) ) * k / l ) ) + ( m / ( h ) ) ) ) ) * ( ( min( ( 7 ) * n + o , max( p - q , 0 ) ) / ( h ) ) / max( ( max( ( f - g ) / ( h ) , 0 ) ) , ( ( min( ( 7 ) * n + o , max( p - q , 0 ) ) / ( h ) ) + ( 13 * r / ( h ) ) + ( min( 2 * ( s - t - u ) * dependentloadsweight / 100 , max( p - q , 0 ) ) / ( h ) ) + ( ( 16 * max( 0 , w - x ) + ( w / y ) * ( ( 10 ) * z + ( min( h , a_a ) ) ) ) / ( h ) ) + ( ( a_b / a_c ) * a_d / ( h ) ) + ( a_e / ( h ) ) ) ) ) + ( ( a / ( b + c + d + e ) ) * ( ( m / ( h ) ) / ( ( max( ( f - g ) / ( h ) , 0 ) ) + ( ( g - i ) / ( h ) ) + ( ( i - j ) / ( h ) ) + ( ( j / ( h ) ) * k / l ) + ( ( j / ( h ) ) - ( ( j / ( h ) ) * k / l ) ) + ( m / ( h ) ) ) ) * ( ( ( ( 7 ) * a_f + a_g ) / ( a_h if smt_on else ( h ) ) ) / ( ( ( ( a_i * ( 10 ) * ( 1 - ( w / y ) ) ) + ( 1 - ( w / y ) ) * ( min( h , a_a ) ) ) / ( h ) ) + ( ( ( 170 * ( ( ( h ) / a_j ) * a_k / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_m + ( 81 * ( ( ( h ) / a_j ) * a_k / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_n ) / ( h ) ) + ( a_o / ( a_h if smt_on else ( h ) ) ) + ( 9 * a_p / ( h ) ) + ( ( ( 7 ) * a_f + a_g ) / ( a_h if smt_on else ( h ) ) ) ) ) ) )", - "BaseFormula": " 100 * ( tma_memory_bound * ( tma_l1_bound / max( tma_memory_bound , ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) ) * ( tma_dtlb_load / max( tma_l1_bound , ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Bottleneck_Memory_Data_TLBs" - } - ], - "Formula": "a > 20", - "BaseFormula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20", - "ThresholdIssues": "$issueTLB" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Bottleneck_Memory_Synchronization", - "LegacyName": "metric_TMA_Bottleneck_Memory_Synchronization", - "Level": 1, - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.MEMORY_BOUND", - "Alias": "a" - }, - { - "Name": "OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD", - "Alias": "a_a" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS", - "Alias": "a_b" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD", - "Alias": "a_c" - }, - { - "Name": "MEM_LOAD_RETIRED.L3_HIT", - "Alias": "a_d" - }, - { - "Name": "XQ.FULL_CYCLES", - "Alias": "a_e" - }, - { - "Name": "L1D_PEND_MISS.L2_STALLS", - "Alias": "a_f" - }, - { - "Name": "OCR.DEMAND_RFO.L3_MISS:ocr_msr_val=0x103b800002", - "Alias": "a_g" - }, - { - "Name": "OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM", - "Alias": "a_h" - }, - { - "Name": "MEM_STORE_RETIRED.L2_HIT", - "Alias": "a_i" - }, - { - "Name": "MEM_INST_RETIRED.LOCK_LOADS", - "Alias": "a_j" - }, - { - "Name": "MEM_INST_RETIRED.ALL_STORES", - "Alias": "a_k" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO", - "Alias": "a_l" - }, - { - "Name": "MEM_INST_RETIRED.SPLIT_STORES", - "Alias": "a_m" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "a_n" - }, - { - "Name": "OCR.STREAMING_WR.ANY_RESPONSE", - "Alias": "a_o" - }, - { - "Name": "DTLB_STORE_MISSES.STLB_HIT:c1", - "Alias": "a_p" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", - "Alias": "a_q" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "a_r" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "a_s" - }, - { - "Name": "PERF_METRICS.BRANCH_MISPREDICTS", - "Alias": "a_t" - }, - { - "Name": "MACHINE_CLEARS.MEMORY_ORDERING", - "Alias": "a_u" - }, - { - "Name": "MACHINE_CLEARS.COUNT", - "Alias": "a_v" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L3_MISS", - "Alias": "f" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "g" - }, - { - "Name": "OCR.DEMAND_DATA_RD.PMM", - "Alias": "h" - }, - { - "Name": "OCR.READS_TO_CORE.L3_MISS", - "Alias": "i" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_LOADS", - "Alias": "j" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L1D_MISS", - "Alias": "k" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L2_MISS", - "Alias": "l" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_STORES", - "Alias": "m" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", - "Alias": "n" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD:c4", - "Alias": "o" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "p" - }, - { - "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM", - "Alias": "s" - }, - { - "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD", - "Alias": "t" - }, - { - "Name": "MEM_LOAD_RETIRED.FB_HIT", - "Alias": "u" - }, - { - "Name": "MEM_LOAD_RETIRED.L1_MISS", - "Alias": "v" - }, - { - "Name": "MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM", - "Alias": "w" - }, - { - "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM", - "Alias": "x" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD", - "Alias": "y" - }, - { - "Name": "OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM", - "Alias": "z" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - }, - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "q" - }, - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( a / ( b + c + d + e ) ) * ( ( ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( ( min( g , n ) ) / ( g ) - ( ( min( g , o ) ) / ( g ) ) ) / ( ( ( min( g , o ) ) / ( g ) ) + ( ( min( g , n ) ) / ( g ) - ( ( min( g , o ) ) / ( g ) ) ) ) ) * ( ( ( ( 170 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * s + ( ( 170 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * t ) * ( 1 + ( u / v ) / 2 ) / ( g ) ) / ( ( ( ( 109 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * w * ( 1 + ( u / v ) / 2 ) / ( g ) ) + ( ( ( 190 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * x * ( 1 + ( u / v ) / 2 ) / ( g ) ) + ( ( ( ( 170 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * s + ( ( 170 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * t ) * ( 1 + ( u / v ) / 2 ) / ( g ) ) ) + ( ( ( l - f ) / ( g ) ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( ( ( ( 81 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( y * ( z / ( z + a_a ) ) ) + ( ( 79 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_b ) ) * ( 1 + ( u / v ) / 2 ) / ( g ) ) + ( ( ( 79 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_c + y * ( 1 - ( z / ( z + a_a ) ) ) ) * ( 1 + ( u / v ) / 2 ) / ( g ) ) ) / ( ( ( ( ( 81 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( y * ( z / ( z + a_a ) ) ) + ( ( 79 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_b ) ) * ( 1 + ( u / v ) / 2 ) / ( g ) ) + ( ( ( 79 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_c + y * ( 1 - ( z / ( z + a_a ) ) ) ) * ( 1 + ( u / v ) / 2 ) / ( g ) ) + ( ( ( 37 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_d * ( 1 + ( u / v ) / 2 ) ) / ( g ) ) + ( ( a_e + a_f ) / ( g ) ) ) + ( ( m / ( g ) ) / ( ( max( ( j - k ) / ( g ) , 0 ) ) + ( ( k - l ) / ( g ) ) + ( ( l - f ) / ( g ) ) + ( ( f / ( g ) ) * h / i ) + ( ( f / ( g ) ) - ( ( f / ( g ) ) * h / i ) ) + ( m / ( g ) ) ) ) * ( ( ( 170 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_g + ( 81 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_h ) / ( g ) ) / ( ( ( ( ( a_i * ( 10 ) * ( 1 - ( a_j / a_k ) ) ) + ( 1 - ( a_j / a_k ) ) * ( min( g , a_l ) ) ) / ( g ) ) + ( ( ( 170 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_g + ( 81 * ( ( ( g ) / p ) * q / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * a_h ) / ( g ) ) + ( a_m / ( a_n if smt_on else ( g ) ) ) + ( 9 * a_o / ( g ) ) + ( ( ( 7 ) * a_p + a_q ) / ( a_n if smt_on else ( g ) ) ) ) - ( ( ( a_i * ( 10 ) * ( 1 - ( a_j / a_k ) ) ) + ( 1 - ( a_j / a_k ) ) * ( min( g , a_l ) ) ) / ( g ) ) ) ) + ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - a_r / ( a_s ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( a_t / ( b + c + d + e ) ) ) ) * ( 1 - ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - a_r / ( a_s ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( a_t / ( b + c + d + e ) ) ) ) * ( 1 - a_u / a_v ) , 0.0001 ) ) / ( ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - a_r / ( a_s ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( a_t / ( b + c + d + e ) ) ) ) * ( 1 - a_u / a_v ) , 0.0001 ) ) ) ) )", - "BaseFormula": " 100 * ( tma_memory_bound * ( ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) * tma_remote_cache / ( tma_local_mem + tma_remote_mem + tma_remote_cache ) + ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_hbm_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Bottleneck_Memory_Synchronization" - } - ], - "Formula": "a > 10", - "BaseFormula": "metric_TMA_Bottleneck_Memory_Synchronization > 10", - "ThresholdIssues": "$issueSyncxn" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvMS;LockCont;Mem;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Bottleneck_Compute_Bound_Est", - "LegacyName": "metric_TMA_Bottleneck_Compute_Bound_Est", - "Level": 1, - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.MEMORY_BOUND", - "Alias": "e" - }, - { - "Name": "ARITH.DIV_ACTIVE", - "Alias": "f" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "g" - }, - { - "Name": "RESOURCE_STALLS.SCOREBOARD", - "Alias": "h" - }, - { - "Name": "CPU_CLK_UNHALTED.C02", - "Alias": "i" - }, - { - "Name": "EXE.AMX_BUSY", - "Alias": "j" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "k" - }, - { - "Name": "EXE_ACTIVITY.EXE_BOUND_0_PORTS", - "Alias": "l" - }, - { - "Name": "RS.EMPTY_RESOURCE", - "Alias": "m" - }, - { - "Name": "CYCLE_ACTIVITY.STALLS_TOTAL", - "Alias": "n" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_LOADS", - "Alias": "o" - }, - { - "Name": "EXE_ACTIVITY.1_PORTS_UTIL", - "Alias": "p" - }, - { - "Name": "EXE_ACTIVITY.2_3_PORTS_UTIL", - "Alias": "q" - }, - { - "Name": "UOPS_EXECUTED.CYCLES_GE_3", - "Alias": "r" - }, - { - "Name": "EXE_ACTIVITY.2_PORTS_UTIL", - "Alias": "s" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( ( max( 0 , ( a / ( b + c + d + a ) ) - ( e / ( b + c + d + a ) ) ) ) * ( f / ( g ) ) / ( ( f / ( g ) ) + ( h / ( g ) + ( i / ( g ) ) ) + ( j / ( k if smt_on else ( g ) ) ) + ( ( ( ( l + max( m - h , 0 ) ) / ( g ) * ( n - o ) / ( g ) ) * ( g ) + ( p + ( d / ( b + c + d + a ) ) * q ) ) / ( g ) if ( f < ( n - o ) ) else ( p + ( d / ( b + c + d + a ) ) * q ) / ( g ) ) ) ) + ( ( max( 0 , ( a / ( b + c + d + a ) ) - ( e / ( b + c + d + a ) ) ) ) * ( j / ( k if smt_on else ( g ) ) ) / ( ( f / ( g ) ) + ( h / ( g ) + ( i / ( g ) ) ) + ( j / ( k if smt_on else ( g ) ) ) + ( ( ( ( l + max( m - h , 0 ) ) / ( g ) * ( n - o ) / ( g ) ) * ( g ) + ( p + ( d / ( b + c + d + a ) ) * q ) ) / ( g ) if ( f < ( n - o ) ) else ( p + ( d / ( b + c + d + a ) ) * q ) / ( g ) ) ) ) + ( ( max( 0 , ( a / ( b + c + d + a ) ) - ( e / ( b + c + d + a ) ) ) ) * ( ( ( ( ( l + max( m - h , 0 ) ) / ( g ) * ( n - o ) / ( g ) ) * ( g ) + ( p + ( d / ( b + c + d + a ) ) * q ) ) / ( g ) if ( f < ( n - o ) ) else ( p + ( d / ( b + c + d + a ) ) * q ) / ( g ) ) / ( ( f / ( g ) ) + ( h / ( g ) + ( i / ( g ) ) ) + ( j / ( k if smt_on else ( g ) ) ) + ( ( ( ( l + max( m - h , 0 ) ) / ( g ) * ( n - o ) / ( g ) ) * ( g ) + ( p + ( d / ( b + c + d + a ) ) * q ) ) / ( g ) if ( f < ( n - o ) ) else ( p + ( d / ( b + c + d + a ) ) * q ) / ( g ) ) ) ) * ( ( r / ( g ) ) / ( ( ( l + max( m - h , 0 ) ) / ( g ) * ( n - o ) / ( g ) ) + ( p / ( g ) ) + ( s / ( g ) ) + ( r / ( g ) ) ) ) ) )", - "BaseFormula": " 100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( tma_core_bound * tma_amx_busy / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "Formula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvCB;Cor", - "LocateWith": "" - }, - { - "MetricName": "Bottleneck_Irregular_Overhead", - "LegacyName": "metric_TMA_Bottleneck_Irregular_Overhead", - "Level": 1, - "BriefDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.REP_ITERATION", - "Alias": "a" - }, - { - "Name": "IDQ.MITE_CYCLES_OK", - "Alias": "a_a" - }, - { - "Name": "IDQ.DSB_CYCLES_ANY", - "Alias": "a_b" - }, - { - "Name": "IDQ.DSB_CYCLES_OK", - "Alias": "a_c" - }, - { - "Name": "UOPS_RETIRED.MS", - "Alias": "a_d" - }, - { - "Name": "MACHINE_CLEARS.MEMORY_ORDERING", - "Alias": "a_e" - }, - { - "Name": "PERF_METRICS.MEMORY_BOUND", - "Alias": "a_f" - }, - { - "Name": "RESOURCE_STALLS.SCOREBOARD", - "Alias": "a_g" - }, - { - "Name": "CPU_CLK_UNHALTED.C02", - "Alias": "a_h" - }, - { - "Name": "RS.EMPTY_RESOURCE", - "Alias": "a_i" - }, - { - "Name": "EXE_ACTIVITY.EXE_BOUND_0_PORTS", - "Alias": "a_j" - }, - { - "Name": "CYCLE_ACTIVITY.STALLS_TOTAL", - "Alias": "a_k" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_LOADS", - "Alias": "a_l" - }, - { - "Name": "ARITH.DIV_ACTIVE", - "Alias": "a_m" - }, - { - "Name": "EXE.AMX_BUSY", - "Alias": "a_n" - }, - { - "Name": "EXE_ACTIVITY.1_PORTS_UTIL", - "Alias": "a_o" - }, - { - "Name": "EXE_ACTIVITY.2_3_PORTS_UTIL", - "Alias": "a_p" - }, - { - "Name": "PERF_METRICS.HEAVY_OPERATIONS", - "Alias": "a_q" - }, - { - "Name": "ASSISTS.ANY", - "Alias": "a_r" - }, - { - "Name": "UOPS_RETIRED.MS:c1", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.FETCH_LATENCY", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "e" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "f" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "g" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "h" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "i" - }, - { - "Name": "UOPS_RETIRED.MS:c1:e1", - "Alias": "j" - }, - { - "Name": "UOPS_RETIRED.SLOTS", - "Alias": "k" - }, - { - "Name": "UOPS_ISSUED.ANY", - "Alias": "l" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "m" - }, - { - "Name": "INT_MISC.CLEAR_RESTEER_CYCLES", - "Alias": "n" - }, - { - "Name": "INT_MISC.UNKNOWN_BRANCH_CYCLES", - "Alias": "o" - }, - { - "Name": "PERF_METRICS.BRANCH_MISPREDICTS", - "Alias": "p" - }, - { - "Name": "BR_MISP_RETIRED.ALL_BRANCHES", - "Alias": "q" - }, - { - "Name": "INT_MISC.CLEARS_COUNT", - "Alias": "r" - }, - { - "Name": "MACHINE_CLEARS.COUNT", - "Alias": "s" - }, - { - "Name": "ICACHE_DATA.STALLS", - "Alias": "t" - }, - { - "Name": "ICACHE_TAG.STALLS", - "Alias": "u" - }, - { - "Name": "DECODE.LCP", - "Alias": "v" - }, - { - "Name": "DSB2MITE_SWITCHES.PENALTY_CYCLES", - "Alias": "w" - }, - { - "Name": "IDQ.MS_CYCLES_ANY", - "Alias": "x" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "y" - }, - { - "Name": "IDQ.MITE_CYCLES_ANY", - "Alias": "z" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( ( 1 - a / b ) * ( ( ( c / ( d + e + f + g ) - h / ( i ) ) ) * ( ( ( 3 ) * j / ( k / l ) / ( m ) ) + ( n / ( m ) + ( o / ( m ) ) ) * ( ( ( 1 - ( ( p / ( d + e + f + g ) ) / ( max( 1 - ( ( d / ( d + e + f + g ) - h / ( i ) ) + ( g / ( d + e + f + g ) ) + ( f / ( d + e + f + g ) ) ) , 0 ) ) ) ) * n / ( m ) ) + ( ( ( p / ( d + e + f + g ) ) / ( max( 1 - ( ( d / ( d + e + f + g ) - h / ( i ) ) + ( g / ( d + e + f + g ) ) + ( f / ( d + e + f + g ) ) ) , 0 ) ) ) * n / ( m ) ) * ( max( ( p / ( d + e + f + g ) ) * ( 1 - q / ( r - s ) ) , 0.0001 ) ) / ( p / ( d + e + f + g ) ) ) / ( ( ( ( p / ( d + e + f + g ) ) / ( max( 1 - ( ( d / ( d + e + f + g ) - h / ( i ) ) + ( g / ( d + e + f + g ) ) + ( f / ( d + e + f + g ) ) ) , 0 ) ) ) * n / ( m ) ) + ( ( 1 - ( ( p / ( d + e + f + g ) ) / ( max( 1 - ( ( d / ( d + e + f + g ) - h / ( i ) ) + ( g / ( d + e + f + g ) ) + ( f / ( d + e + f + g ) ) ) , 0 ) ) ) ) * n / ( m ) ) + ( o / ( m ) ) ) ) / ( ( t / ( m ) ) + ( u / ( m ) ) + ( n / ( m ) + ( o / ( m ) ) ) + ( ( 3 ) * j / ( k / l ) / ( m ) ) + ( v / ( m ) ) + ( w / ( m ) ) ) + ( max( 0 , ( d / ( d + e + f + g ) - h / ( i ) ) - ( ( c / ( d + e + f + g ) - h / ( i ) ) ) ) ) * ( max( x , b / ( k / l ) ) / ( y if smt_on else ( m ) ) / 2 ) / ( ( ( z - a_a ) / ( y if smt_on else ( m ) ) / 2 ) + ( ( a_b - a_c ) / ( y if smt_on else ( m ) ) / 2 ) + ( max( x , b / ( k / l ) ) / ( y if smt_on else ( m ) ) / 2 ) ) ) ) + ( 10 * ( a_d / ( i ) ) * ( max( ( p / ( d + e + f + g ) ) * ( 1 - q / ( r - s ) ) , 0.0001 ) ) / ( p / ( d + e + f + g ) ) ) * ( p / ( d + e + f + g ) ) + ( ( max( 0 , ( max( 1 - ( ( d / ( d + e + f + g ) - h / ( i ) ) + ( g / ( d + e + f + g ) ) + ( f / ( d + e + f + g ) ) ) , 0 ) ) - ( p / ( d + e + f + g ) ) ) ) * ( max( ( max( 0 , ( max( 1 - ( ( d / ( d + e + f + g ) - h / ( i ) ) + ( g / ( d + e + f + g ) ) + ( f / ( d + e + f + g ) ) ) , 0 ) ) - ( p / ( d + e + f + g ) ) ) ) * ( 1 - a_e / s ) , 0.0001 ) ) / ( ( max( ( max( 0 , ( max( 1 - ( ( d / ( d + e + f + g ) - h / ( i ) ) + ( g / ( d + e + f + g ) ) + ( f / ( d + e + f + g ) ) ) , 0 ) ) - ( p / ( d + e + f + g ) ) ) ) * ( 1 - a_e / s ) , 0.0001 ) ) ) ) + ( ( max( 0 , ( g / ( d + e + f + g ) ) - ( a_f / ( d + e + f + g ) ) ) ) * ( ( a_g / ( m ) + ( a_h / ( m ) ) ) + a_i / ( m ) * ( ( a_j + max( a_i - a_g , 0 ) ) / ( m ) * ( a_k - a_l ) / ( m ) ) ) / ( ( a_m / ( m ) ) + ( a_g / ( m ) + ( a_h / ( m ) ) ) + ( a_n / ( y if smt_on else ( m ) ) ) + ( ( ( ( a_j + max( a_i - a_g , 0 ) ) / ( m ) * ( a_k - a_l ) / ( m ) ) * ( m ) + ( a_o + ( f / ( d + e + f + g ) ) * a_p ) ) / ( m ) if ( a_m < ( a_k - a_l ) ) else ( a_o + ( f / ( d + e + f + g ) ) * a_p ) / ( m ) ) ) ) + ( ( ( ( a_d / ( i ) ) / ( ( max( 0 , ( a_q / ( d + e + f + g ) ) - ( a_d / ( i ) ) ) ) + ( a_d / ( i ) ) ) ) * ( ( ( ( 99 * 3 + 63 + 30 ) / 5 ) * a_r / ( i ) ) / ( a_d / ( i ) ) ) ) * ( a_q / ( d + e + f + g ) ) ) )", - "BaseFormula": " 100 * ( ( ( 1 - inst_retired.rep_iteration / uops_retired.ms:c1 ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_ms ) ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + rs.empty_resource / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "Formula": "metric_TMA_Bottleneck_Irregular_Overhead > 10" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Bad;BvIO;Cor;Ret", - "LocateWith": "" - }, - { - "MetricName": "Bottleneck_Other_Bottlenecks", - "LegacyName": "metric_TMA_Bottleneck_Other_Bottlenecks", - "Level": 1, - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end. Examples include data-dependencies (Core Bound when Low ILP) and other unlisted memory-related stalls.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.FETCH_LATENCY", - "Alias": "a" - }, - { - "Name": "IDQ.MITE_CYCLES_ANY", - "Alias": "a_a" - }, - { - "Name": "IDQ.MITE_CYCLES_OK", - "Alias": "a_b" - }, - { - "Name": "IDQ.DSB_CYCLES_ANY", - "Alias": "a_c" - }, - { - "Name": "IDQ.DSB_CYCLES_OK", - "Alias": "a_d" - }, - { - "Name": "PERF_METRICS.MEMORY_BOUND", - "Alias": "a_e" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L3_MISS", - "Alias": "a_f" - }, - { - "Name": "OCR.DEMAND_DATA_RD.PMM", - "Alias": "a_g" - }, - { - "Name": "OCR.READS_TO_CORE.L3_MISS", - "Alias": "a_h" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_LOADS", - "Alias": "a_i" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L1D_MISS", - "Alias": "a_j" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L2_MISS", - "Alias": "a_k" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_STORES", - "Alias": "a_l" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD:c4", - "Alias": "a_m" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", - "Alias": "a_n" - }, - { - "Name": "XQ.FULL_CYCLES", - "Alias": "a_o" - }, - { - "Name": "L1D_PEND_MISS.L2_STALLS", - "Alias": "a_p" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "a_q" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD", - "Alias": "a_t" - }, - { - "Name": "OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM", - "Alias": "a_u" - }, - { - "Name": "OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD", - "Alias": "a_v" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS", - "Alias": "a_w" - }, - { - "Name": "MEM_LOAD_RETIRED.FB_HIT", - "Alias": "a_x" - }, - { - "Name": "MEM_LOAD_RETIRED.L1_MISS", - "Alias": "a_y" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD", - "Alias": "a_z" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "MEM_LOAD_RETIRED.L3_HIT", - "Alias": "b_a" - }, - { - "Name": "L1D_PEND_MISS.FB_FULL", - "Alias": "b_b" - }, - { - "Name": "DTLB_LOAD_MISSES.STLB_HIT:c1", - "Alias": "b_c" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_ACTIVE", - "Alias": "b_d" - }, - { - "Name": "CYCLE_ACTIVITY.CYCLES_MEM_ANY", - "Alias": "b_e" - }, - { - "Name": "MEMORY_ACTIVITY.CYCLES_L1D_MISS", - "Alias": "b_f" - }, - { - "Name": "LD_BLOCKS.STORE_FORWARD", - "Alias": "b_g" - }, - { - "Name": "MEM_INST_RETIRED.ALL_LOADS", - "Alias": "b_h" - }, - { - "Name": "MEM_INST_RETIRED.LOCK_LOADS", - "Alias": "b_j" - }, - { - "Name": "L2_RQSTS.ALL_RFO", - "Alias": "b_k" - }, - { - "Name": "MEM_INST_RETIRED.ALL_STORES", - "Alias": "b_l" - }, - { - "Name": "L2_RQSTS.RFO_HIT", - "Alias": "b_m" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO", - "Alias": "b_n" - }, - { - "Name": "L1D_PEND_MISS.PENDING", - "Alias": "b_o" - }, - { - "Name": "MEM_LOAD_COMPLETED.L1_MISS_ANY", - "Alias": "b_p" - }, - { - "Name": "LD_BLOCKS.NO_SR", - "Alias": "b_q" - }, - { - "Name": "MEM_INST_RETIRED.SPLIT_STORES", - "Alias": "b_r" - }, - { - "Name": "MEM_STORE_RETIRED.L2_HIT", - "Alias": "b_s" - }, - { - "Name": "OCR.DEMAND_RFO.L3_MISS:ocr_msr_val=0x103b800002", - "Alias": "b_t" - }, - { - "Name": "OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM", - "Alias": "b_u" - }, - { - "Name": "OCR.STREAMING_WR.ANY_RESPONSE", - "Alias": "b_v" - }, - { - "Name": "DTLB_STORE_MISSES.STLB_HIT:c1", - "Alias": "b_w" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", - "Alias": "b_x" - }, - { - "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM", - "Alias": "b_y" - }, - { - "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD", - "Alias": "b_z" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM", - "Alias": "c_a" - }, - { - "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM", - "Alias": "c_b" - }, - { - "Name": "MACHINE_CLEARS.MEMORY_ORDERING", - "Alias": "c_c" - }, - { - "Name": "ARITH.DIV_ACTIVE", - "Alias": "c_d" - }, - { - "Name": "RESOURCE_STALLS.SCOREBOARD", - "Alias": "c_e" - }, - { - "Name": "CPU_CLK_UNHALTED.C02", - "Alias": "c_f" - }, - { - "Name": "EXE.AMX_BUSY", - "Alias": "c_g" - }, - { - "Name": "EXE_ACTIVITY.EXE_BOUND_0_PORTS", - "Alias": "c_h" - }, - { - "Name": "RS.EMPTY_RESOURCE", - "Alias": "c_i" - }, - { - "Name": "CYCLE_ACTIVITY.STALLS_TOTAL", - "Alias": "c_j" - }, - { - "Name": "EXE_ACTIVITY.1_PORTS_UTIL", - "Alias": "c_k" - }, - { - "Name": "EXE_ACTIVITY.2_3_PORTS_UTIL", - "Alias": "c_l" - }, - { - "Name": "UOPS_EXECUTED.CYCLES_GE_3", - "Alias": "c_m" - }, - { - "Name": "EXE_ACTIVITY.2_PORTS_UTIL", - "Alias": "c_n" - }, - { - "Name": "PERF_METRICS.HEAVY_OPERATIONS", - "Alias": "c_o" - }, - { - "Name": "ASSISTS.ANY", - "Alias": "c_p" - }, - { - "Name": "BR_INST_RETIRED.ALL_BRANCHES", - "Alias": "c_q" - }, - { - "Name": "BR_INST_RETIRED.NEAR_CALL", - "Alias": "c_r" - }, - { - "Name": "INST_RETIRED.NOP", - "Alias": "c_s" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - }, - { - "Name": "ICACHE_TAG.STALLS", - "Alias": "h" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "i" - }, - { - "Name": "ICACHE_DATA.STALLS", - "Alias": "j" - }, - { - "Name": "INT_MISC.UNKNOWN_BRANCH_CYCLES", - "Alias": "k" - }, - { - "Name": "INT_MISC.CLEAR_RESTEER_CYCLES", - "Alias": "l" - }, - { - "Name": "UOPS_RETIRED.MS:c1:e1", - "Alias": "m" - }, - { - "Name": "UOPS_RETIRED.SLOTS", - "Alias": "n" - }, - { - "Name": "UOPS_ISSUED.ANY", - "Alias": "o" - }, - { - "Name": "DECODE.LCP", - "Alias": "p" - }, - { - "Name": "DSB2MITE_SWITCHES.PENALTY_CYCLES", - "Alias": "q" - }, - { - "Name": "UOPS_RETIRED.MS", - "Alias": "r" - }, - { - "Name": "PERF_METRICS.BRANCH_MISPREDICTS", - "Alias": "s" - }, - { - "Name": "BR_MISP_RETIRED.ALL_BRANCHES", - "Alias": "t" - }, - { - "Name": "INT_MISC.CLEARS_COUNT", - "Alias": "u" - }, - { - "Name": "MACHINE_CLEARS.COUNT", - "Alias": "v" - }, - { - "Name": "INST_RETIRED.REP_ITERATION", - "Alias": "w" - }, - { - "Name": "UOPS_RETIRED.MS:c1", - "Alias": "x" - }, - { - "Name": "IDQ.MS_CYCLES_ANY", - "Alias": "y" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "z" - } - ], - "Constants": [ - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "a_r" - }, - { - "Name": "20", - "Alias": "dependentloadsweight" - }, - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - }, - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 - ( ( 100 * ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( h / ( i ) ) + ( j / ( i ) ) + ( k / ( i ) ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) ) + ( 100 * ( ( b / ( b + c + d + e ) - f / ( g ) ) - ( 1 - ( 10 * ( r / ( g ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) ) * ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) - ( ( 1 - w / x ) * ( ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) * ( ( ( 1 - ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) ) * l / ( i ) ) + ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) / ( ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) + ( ( 1 - ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) ) * l / ( i ) ) + ( k / ( i ) ) ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) + ( max( 0 , ( b / ( b + c + d + e ) - f / ( g ) ) - ( ( a / ( b + c + d + e ) - f / ( g ) ) ) ) ) * ( max( y , x / ( n / o ) ) / ( z if smt_on else ( i ) ) / 2 ) / ( ( ( a_a - a_b ) / ( z if smt_on else ( i ) ) / 2 ) + ( ( a_c - a_d ) / ( z if smt_on else ( i ) ) / 2 ) + ( max( y , x / ( n / o ) ) / ( z if smt_on else ( i ) ) / 2 ) ) ) ) ) - ( 100 * ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( h / ( i ) ) + ( j / ( i ) ) + ( k / ( i ) ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) ) ) + ( 100 * ( 1 - ( 10 * ( r / ( g ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) ) * ( ( s / ( b + c + d + e ) ) + ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) ) ) + ( ( 100 * ( ( ( a_e / ( b + c + d + e ) ) * ( ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( ( min( i , a_m ) ) / ( i ) ) / ( ( ( min( i , a_m ) ) / ( i ) ) + ( ( min( i , a_n ) ) / ( i ) - ( ( min( i , a_m ) ) / ( i ) ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( a_f / ( i ) ) * a_g / a_h ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( min( i , a_m ) ) / ( i ) ) / ( ( ( min( i , a_m ) ) / ( i ) ) + ( ( min( i , a_n ) ) / ( i ) - ( ( min( i , a_m ) ) / ( i ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( ( a_k - a_f ) / ( i ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( ( a_o + a_p ) / ( i ) ) / ( ( ( ( ( 81 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_t * ( a_u / ( a_u + a_v ) ) ) + ( ( 79 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_w ) ) * ( 1 + ( a_x / a_y ) / 2 ) / ( i ) ) + ( ( ( 79 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_z + a_t * ( 1 - ( a_u / ( a_u + a_v ) ) ) ) * ( 1 + ( a_x / a_y ) / 2 ) / ( i ) ) + ( ( ( 37 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( b_a * ( 1 + ( a_x / a_y ) / 2 ) ) / ( i ) ) + ( ( a_o + a_p ) / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( b_b / ( i ) ) / ( ( min( ( 7 ) * b_c + b_d , max( b_e - b_f , 0 ) ) / ( i ) ) + ( 13 * b_g / ( i ) ) + ( min( 2 * ( b_h - a_x - a_y ) * dependentloadsweight / 100 , max( b_e - b_f , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_j - b_k ) + ( b_j / b_l ) * ( ( 10 ) * b_m + ( min( i , b_n ) ) ) ) / ( i ) ) + ( ( b_o / b_p ) * b_q / ( i ) ) + ( b_b / ( i ) ) ) ) ) ) ) ) + ( 100 * ( ( ( a_e / ( b + c + d + e ) ) * ( ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( ( min( i , a_n ) ) / ( i ) - ( ( min( i , a_m ) ) / ( i ) ) ) / ( ( ( min( i , a_m ) ) / ( i ) ) + ( ( min( i , a_n ) ) / ( i ) - ( ( min( i , a_m ) ) / ( i ) ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( a_f / ( i ) ) * a_g / a_h ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( min( i , a_n ) ) / ( i ) - ( ( min( i , a_m ) ) / ( i ) ) ) / ( ( ( min( i , a_m ) ) / ( i ) ) + ( ( min( i , a_n ) ) / ( i ) - ( ( min( i , a_m ) ) / ( i ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( ( a_k - a_f ) / ( i ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( ( ( 37 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( b_a * ( 1 + ( a_x / a_y ) / 2 ) ) / ( i ) ) / ( ( ( ( ( 81 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_t * ( a_u / ( a_u + a_v ) ) ) + ( ( 79 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_w ) ) * ( 1 + ( a_x / a_y ) / 2 ) / ( i ) ) + ( ( ( 79 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_z + a_t * ( 1 - ( a_u / ( a_u + a_v ) ) ) ) * ( 1 + ( a_x / a_y ) / 2 ) / ( i ) ) + ( ( ( 37 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( b_a * ( 1 + ( a_x / a_y ) / 2 ) ) / ( i ) ) + ( ( a_o + a_p ) / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( a_j - a_k ) / ( i ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( min( 2 * ( b_h - a_x - a_y ) * dependentloadsweight / 100 , max( b_e - b_f , 0 ) ) / ( i ) ) / ( ( min( ( 7 ) * b_c + b_d , max( b_e - b_f , 0 ) ) / ( i ) ) + ( 13 * b_g / ( i ) ) + ( min( 2 * ( b_h - a_x - a_y ) * dependentloadsweight / 100 , max( b_e - b_f , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_j - b_k ) + ( b_j / b_l ) * ( ( 10 ) * b_m + ( min( i , b_n ) ) ) ) / ( i ) ) + ( ( b_o / b_p ) * b_q / ( i ) ) + ( b_b / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( ( 16 * max( 0 , b_j - b_k ) + ( b_j / b_l ) * ( ( 10 ) * b_m + ( min( i , b_n ) ) ) ) / ( i ) ) / ( ( min( ( 7 ) * b_c + b_d , max( b_e - b_f , 0 ) ) / ( i ) ) + ( 13 * b_g / ( i ) ) + ( min( 2 * ( b_h - a_x - a_y ) * dependentloadsweight / 100 , max( b_e - b_f , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_j - b_k ) + ( b_j / b_l ) * ( ( 10 ) * b_m + ( min( i , b_n ) ) ) ) / ( i ) ) + ( ( b_o / b_p ) * b_q / ( i ) ) + ( b_b / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( ( b_o / b_p ) * b_q / ( i ) ) / ( ( min( ( 7 ) * b_c + b_d , max( b_e - b_f , 0 ) ) / ( i ) ) + ( 13 * b_g / ( i ) ) + ( min( 2 * ( b_h - a_x - a_y ) * dependentloadsweight / 100 , max( b_e - b_f , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_j - b_k ) + ( b_j / b_l ) * ( ( 10 ) * b_m + ( min( i , b_n ) ) ) ) / ( i ) ) + ( ( b_o / b_p ) * b_q / ( i ) ) + ( b_b / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( a_l / ( i ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( b_r / ( z if smt_on else ( i ) ) ) / ( ( ( ( b_s * ( 10 ) * ( 1 - ( b_j / b_l ) ) ) + ( 1 - ( b_j / b_l ) ) * ( min( i , b_n ) ) ) / ( i ) ) + ( ( ( 170 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_t + ( 81 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_u ) / ( i ) ) + ( b_r / ( z if smt_on else ( i ) ) ) + ( 9 * b_v / ( i ) ) + ( ( ( 7 ) * b_w + b_x ) / ( z if smt_on else ( i ) ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( a_l / ( i ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( ( ( b_s * ( 10 ) * ( 1 - ( b_j / b_l ) ) ) + ( 1 - ( b_j / b_l ) ) * ( min( i , b_n ) ) ) / ( i ) ) / ( ( ( ( b_s * ( 10 ) * ( 1 - ( b_j / b_l ) ) ) + ( 1 - ( b_j / b_l ) ) * ( min( i , b_n ) ) ) / ( i ) ) + ( ( ( 170 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_t + ( 81 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_u ) / ( i ) ) + ( b_r / ( z if smt_on else ( i ) ) ) + ( 9 * b_v / ( i ) ) + ( ( ( 7 ) * b_w + b_x ) / ( z if smt_on else ( i ) ) ) ) ) ) ) ) + ( 100 * ( ( a_e / ( b + c + d + e ) ) * ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) / max( ( a_e / ( b + c + d + e ) ) , ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) ) * ( ( min( ( 7 ) * b_c + b_d , max( b_e - b_f , 0 ) ) / ( i ) ) / max( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) , ( ( min( ( 7 ) * b_c + b_d , max( b_e - b_f , 0 ) ) / ( i ) ) + ( 13 * b_g / ( i ) ) + ( min( 2 * ( b_h - a_x - a_y ) * dependentloadsweight / 100 , max( b_e - b_f , 0 ) ) / ( i ) ) + ( ( 16 * max( 0 , b_j - b_k ) + ( b_j / b_l ) * ( ( 10 ) * b_m + ( min( i , b_n ) ) ) ) / ( i ) ) + ( ( b_o / b_p ) * b_q / ( i ) ) + ( b_b / ( i ) ) ) ) ) + ( ( a_e / ( b + c + d + e ) ) * ( ( a_l / ( i ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( ( ( 7 ) * b_w + b_x ) / ( z if smt_on else ( i ) ) ) / ( ( ( ( b_s * ( 10 ) * ( 1 - ( b_j / b_l ) ) ) + ( 1 - ( b_j / b_l ) ) * ( min( i , b_n ) ) ) / ( i ) ) + ( ( ( 170 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_t + ( 81 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_u ) / ( i ) ) + ( b_r / ( z if smt_on else ( i ) ) ) + ( 9 * b_v / ( i ) ) + ( ( ( 7 ) * b_w + b_x ) / ( z if smt_on else ( i ) ) ) ) ) ) ) ) + ( 100 * ( ( a_e / ( b + c + d + e ) ) * ( ( ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( ( min( i , a_n ) ) / ( i ) - ( ( min( i , a_m ) ) / ( i ) ) ) / ( ( ( min( i , a_m ) ) / ( i ) ) + ( ( min( i , a_n ) ) / ( i ) - ( ( min( i , a_m ) ) / ( i ) ) ) ) ) * ( ( ( ( 170 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_y + ( ( 170 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_z ) * ( 1 + ( a_x / a_y ) / 2 ) / ( i ) ) / ( ( ( ( 109 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * c_a * ( 1 + ( a_x / a_y ) / 2 ) / ( i ) ) + ( ( ( 190 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * c_b * ( 1 + ( a_x / a_y ) / 2 ) / ( i ) ) + ( ( ( ( 170 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_y + ( ( 170 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * b_z ) * ( 1 + ( a_x / a_y ) / 2 ) / ( i ) ) ) + ( ( ( a_k - a_f ) / ( i ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( ( ( ( 81 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_t * ( a_u / ( a_u + a_v ) ) ) + ( ( 79 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_w ) ) * ( 1 + ( a_x / a_y ) / 2 ) / ( i ) ) + ( ( ( 79 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_z + a_t * ( 1 - ( a_u / ( a_u + a_v ) ) ) ) * ( 1 + ( a_x / a_y ) / 2 ) / ( i ) ) ) / ( ( ( ( ( 81 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_t * ( a_u / ( a_u + a_v ) ) ) + ( ( 79 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_w ) ) * ( 1 + ( a_x / a_y ) / 2 ) / ( i ) ) + ( ( ( 79 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( a_z + a_t * ( 1 - ( a_u / ( a_u + a_v ) ) ) ) * ( 1 + ( a_x / a_y ) / 2 ) / ( i ) ) + ( ( ( 37 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( b_a * ( 1 + ( a_x / a_y ) / 2 ) ) / ( i ) ) + ( ( a_o + a_p ) / ( i ) ) ) + ( ( a_l / ( i ) ) / ( ( max( ( a_i - a_j ) / ( i ) , 0 ) ) + ( ( a_j - a_k ) / ( i ) ) + ( ( a_k - a_f ) / ( i ) ) + ( ( a_f / ( i ) ) * a_g / a_h ) + ( ( a_f / ( i ) ) - ( ( a_f / ( i ) ) * a_g / a_h ) ) + ( a_l / ( i ) ) ) ) * ( ( ( 170 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_t + ( 81 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_u ) / ( i ) ) / ( ( ( ( ( b_s * ( 10 ) * ( 1 - ( b_j / b_l ) ) ) + ( 1 - ( b_j / b_l ) ) * ( min( i , b_n ) ) ) / ( i ) ) + ( ( ( 170 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_t + ( 81 * ( ( ( i ) / a_q ) * a_r / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * b_u ) / ( i ) ) + ( b_r / ( z if smt_on else ( i ) ) ) + ( 9 * b_v / ( i ) ) + ( ( ( 7 ) * b_w + b_x ) / ( z if smt_on else ( i ) ) ) ) - ( ( ( b_s * ( 10 ) * ( 1 - ( b_j / b_l ) ) ) + ( 1 - ( b_j / b_l ) ) * ( min( i , b_n ) ) ) / ( i ) ) ) ) + ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - c_c / v ) , 0.0001 ) ) / ( ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - c_c / v ) , 0.0001 ) ) ) ) ) ) + ( 100 * ( ( ( max( 0 , ( e / ( b + c + d + e ) ) - ( a_e / ( b + c + d + e ) ) ) ) * ( c_d / ( i ) ) / ( ( c_d / ( i ) ) + ( c_e / ( i ) + ( c_f / ( i ) ) ) + ( c_g / ( z if smt_on else ( i ) ) ) + ( ( ( ( c_h + max( c_i - c_e , 0 ) ) / ( i ) * ( c_j - a_i ) / ( i ) ) * ( i ) + ( c_k + ( d / ( b + c + d + e ) ) * c_l ) ) / ( i ) if ( c_d < ( c_j - a_i ) ) else ( c_k + ( d / ( b + c + d + e ) ) * c_l ) / ( i ) ) ) ) + ( ( max( 0 , ( e / ( b + c + d + e ) ) - ( a_e / ( b + c + d + e ) ) ) ) * ( c_g / ( z if smt_on else ( i ) ) ) / ( ( c_d / ( i ) ) + ( c_e / ( i ) + ( c_f / ( i ) ) ) + ( c_g / ( z if smt_on else ( i ) ) ) + ( ( ( ( c_h + max( c_i - c_e , 0 ) ) / ( i ) * ( c_j - a_i ) / ( i ) ) * ( i ) + ( c_k + ( d / ( b + c + d + e ) ) * c_l ) ) / ( i ) if ( c_d < ( c_j - a_i ) ) else ( c_k + ( d / ( b + c + d + e ) ) * c_l ) / ( i ) ) ) ) + ( ( max( 0 , ( e / ( b + c + d + e ) ) - ( a_e / ( b + c + d + e ) ) ) ) * ( ( ( ( ( c_h + max( c_i - c_e , 0 ) ) / ( i ) * ( c_j - a_i ) / ( i ) ) * ( i ) + ( c_k + ( d / ( b + c + d + e ) ) * c_l ) ) / ( i ) if ( c_d < ( c_j - a_i ) ) else ( c_k + ( d / ( b + c + d + e ) ) * c_l ) / ( i ) ) / ( ( c_d / ( i ) ) + ( c_e / ( i ) + ( c_f / ( i ) ) ) + ( c_g / ( z if smt_on else ( i ) ) ) + ( ( ( ( c_h + max( c_i - c_e , 0 ) ) / ( i ) * ( c_j - a_i ) / ( i ) ) * ( i ) + ( c_k + ( d / ( b + c + d + e ) ) * c_l ) ) / ( i ) if ( c_d < ( c_j - a_i ) ) else ( c_k + ( d / ( b + c + d + e ) ) * c_l ) / ( i ) ) ) ) * ( ( c_m / ( i ) ) / ( ( ( c_h + max( c_i - c_e , 0 ) ) / ( i ) * ( c_j - a_i ) / ( i ) ) + ( c_k / ( i ) ) + ( c_n / ( i ) ) + ( c_m / ( i ) ) ) ) ) ) ) + ( 100 * ( ( ( 1 - w / x ) * ( ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) * ( ( ( 1 - ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) ) * l / ( i ) ) + ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) / ( ( ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * l / ( i ) ) + ( ( 1 - ( ( s / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) ) * l / ( i ) ) + ( k / ( i ) ) ) ) / ( ( j / ( i ) ) + ( h / ( i ) ) + ( l / ( i ) + ( k / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) + ( max( 0 , ( b / ( b + c + d + e ) - f / ( g ) ) - ( ( a / ( b + c + d + e ) - f / ( g ) ) ) ) ) * ( max( y , x / ( n / o ) ) / ( z if smt_on else ( i ) ) / 2 ) / ( ( ( a_a - a_b ) / ( z if smt_on else ( i ) ) / 2 ) + ( ( a_c - a_d ) / ( z if smt_on else ( i ) ) / 2 ) + ( max( y , x / ( n / o ) ) / ( z if smt_on else ( i ) ) / 2 ) ) ) ) + ( 10 * ( r / ( g ) ) * ( max( ( s / ( b + c + d + e ) ) * ( 1 - t / ( u - v ) ) , 0.0001 ) ) / ( s / ( b + c + d + e ) ) ) * ( s / ( b + c + d + e ) ) + ( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - c_c / v ) , 0.0001 ) ) / ( ( max( ( max( 0 , ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) - ( s / ( b + c + d + e ) ) ) ) * ( 1 - c_c / v ) , 0.0001 ) ) ) ) + ( ( max( 0 , ( e / ( b + c + d + e ) ) - ( a_e / ( b + c + d + e ) ) ) ) * ( ( c_e / ( i ) + ( c_f / ( i ) ) ) + c_i / ( i ) * ( ( c_h + max( c_i - c_e , 0 ) ) / ( i ) * ( c_j - a_i ) / ( i ) ) ) / ( ( c_d / ( i ) ) + ( c_e / ( i ) + ( c_f / ( i ) ) ) + ( c_g / ( z if smt_on else ( i ) ) ) + ( ( ( ( c_h + max( c_i - c_e , 0 ) ) / ( i ) * ( c_j - a_i ) / ( i ) ) * ( i ) + ( c_k + ( d / ( b + c + d + e ) ) * c_l ) ) / ( i ) if ( c_d < ( c_j - a_i ) ) else ( c_k + ( d / ( b + c + d + e ) ) * c_l ) / ( i ) ) ) ) + ( ( ( ( r / ( g ) ) / ( ( max( 0 , ( c_o / ( b + c + d + e ) ) - ( r / ( g ) ) ) ) + ( r / ( g ) ) ) ) * ( ( ( ( 99 * 3 + 63 + 30 ) / 5 ) * c_p / ( g ) ) / ( r / ( g ) ) ) ) * ( c_o / ( b + c + d + e ) ) ) ) ) + ( 100 * ( ( c_q + 2 * c_r + c_s ) / ( g ) ) ) + ( 100 * ( ( d / ( b + c + d + e ) ) - ( ( c_q + 2 * c_r + c_s ) / ( g ) ) - ( ( ( ( r / ( g ) ) / ( ( max( 0 , ( c_o / ( b + c + d + e ) ) - ( r / ( g ) ) ) ) + ( r / ( g ) ) ) ) * ( ( ( ( 99 * 3 + 63 + 30 ) / 5 ) * c_p / ( g ) ) / ( r / ( g ) ) ) ) * ( c_o / ( b + c + d + e ) ) ) ) ) )", - "BaseFormula": " 100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Bottleneck_Other_Bottlenecks" - } - ], - "Formula": "a > 20", - "BaseFormula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvOB;Cor;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Bottleneck_Branching_Overhead", - "LegacyName": "metric_TMA_Bottleneck_Branching_Overhead", - "Level": 1, - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "BR_INST_RETIRED.ALL_BRANCHES", - "Alias": "a" - }, - { - "Name": "BR_INST_RETIRED.NEAR_CALL", - "Alias": "b" - }, - { - "Name": "INST_RETIRED.NOP", - "Alias": "c" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * ( ( a + 2 * b + c ) / ( d ) )", - "BaseFormula": " 100 * ( ( br_inst_retired.all_branches + 2 * br_inst_retired.near_call + inst_retired.nop ) / tma_info_thread_slots )", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "Formula": "metric_TMA_Bottleneck_Branching_Overhead > 5" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvBO;Ret", - "LocateWith": "" - }, - { - "MetricName": "Bottleneck_Useful_Work", - "LegacyName": "metric_TMA_Bottleneck_Useful_Work", - "Level": 1, - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "BR_INST_RETIRED.ALL_BRANCHES", - "Alias": "e" - }, - { - "Name": "BR_INST_RETIRED.NEAR_CALL", - "Alias": "f" - }, - { - "Name": "INST_RETIRED.NOP", - "Alias": "g" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "h" - }, - { - "Name": "UOPS_RETIRED.MS", - "Alias": "i" - }, - { - "Name": "PERF_METRICS.HEAVY_OPERATIONS", - "Alias": "j" - }, - { - "Name": "ASSISTS.ANY", - "Alias": "k" - } - ], - "Constants": [], - "Formula": "100 * ( ( a / ( b + c + a + d ) ) - ( ( e + 2 * f + g ) / ( h ) ) - ( ( ( ( i / ( h ) ) / ( ( max( 0 , ( j / ( b + c + a + d ) ) - ( i / ( h ) ) ) ) + ( i / ( h ) ) ) ) * ( ( ( ( 99 * 3 + 63 + 30 ) / 5 ) * k / ( h ) ) / ( i / ( h ) ) ) ) * ( j / ( b + c + a + d ) ) ) )", - "BaseFormula": " 100 * ( tma_retiring - ( ( br_inst_retired.all_branches + 2 * br_inst_retired.near_call + inst_retired.nop ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "Formula": "metric_TMA_Bottleneck_Useful_Work > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvUW;Ret", - "LocateWith": "" - }, - { - "MetricName": "Frontend_Bound", - "LegacyName": "metric_TMA_Frontend_Bound(%)", - "Level": 1, - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "e" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "f" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( a + b + c + d ) - e / ( f ) )", - "BaseFormula": " perf_metrics.frontend_bound / ( perf_metrics.frontend_bound + perf_metrics.bad_speculation + perf_metrics.retiring + perf_metrics.backend_bound ) - int_misc.uop_dropping / tma_info_thread_slots", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvFB;BvIO;TmaL1;PGO", - "LocateWith": " FRONTEND_RETIRED.LATENCY_GE_4" - }, - { - "MetricName": "Fetch_Latency", - "LegacyName": "metric_TMA_..Fetch_Latency(%)", - "ParentCategory": "Frontend_Bound", - "Level": 2, - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.FETCH_LATENCY", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - } - ], - "Constants": [], - "Formula": "100 * ( ( a / ( b + c + d + e ) - f / ( g ) ) )", - "BaseFormula": " ( perf_metrics.fetch_latency / ( perf_metrics.frontend_bound + perf_metrics.bad_speculation + perf_metrics.retiring + perf_metrics.backend_bound ) - int_misc.uop_dropping / tma_info_thread_slots )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Frontend;TmaL2", - "LocateWith": " FRONTEND_RETIRED.LATENCY_GE_16;FRONTEND_RETIRED.LATENCY_GE_8" - }, - { - "MetricName": "ICache_Misses", - "LegacyName": "metric_TMA_....ICache_Misses(%)", - "ParentCategory": "Fetch_Latency", - "Level": 3, - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ICACHE_DATA.STALLS", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " icache_data.stalls / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss", - "LocateWith": " FRONTEND_RETIRED.L2_MISS;FRONTEND_RETIRED.L1I_MISS" - }, - { - "MetricName": "Code_L2_Hit", - "LegacyName": "metric_TMA_......Code_L2_Hit(%)", - "ParentCategory": "ICache_Misses", - "Level": 4, - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ICACHE_DATA.STALLS", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_CODE_RD", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "100 * ( max( 0 , ( a / ( b ) ) - ( c / ( b ) ) ) )", - "BaseFormula": " max( 0 , tma_icache_misses - tma_code_l2_miss )", - "Category": "TMA", - "CountDomain": "Clocks_Retired", - "Threshold": { - "Formula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "IcMiss;FetchLat;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Code_L2_Miss", - "LegacyName": "metric_TMA_......Code_L2_Miss(%)", - "ParentCategory": "ICache_Misses", - "Level": 4, - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_CODE_RD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " offcore_requests_outstanding.cycles_with_demand_code_rd / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Retired", - "Threshold": { - "Formula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "IcMiss;FetchLat;Offcore", - "LocateWith": "" - }, - { - "MetricName": "ITLB_Misses", - "LegacyName": "metric_TMA_....ITLB_Misses(%)", - "ParentCategory": "Fetch_Latency", - "Level": 3, - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ICACHE_TAG.STALLS", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " icache_tag.stalls / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB", - "LocateWith": " FRONTEND_RETIRED.STLB_MISS;FRONTEND_RETIRED.ITLB_MISS" - }, - { - "MetricName": "Code_STLB_Hit", - "LegacyName": "metric_TMA_......Code_STLB_Hit(%)", - "ParentCategory": "ITLB_Misses", - "Level": 4, - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ICACHE_TAG.STALLS", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - }, - { - "Name": "ITLB_MISSES.WALK_ACTIVE", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "100 * ( max( 0 , ( a / ( b ) ) - ( c / ( b ) ) ) )", - "BaseFormula": " max( 0 , tma_itlb_misses - tma_code_stlb_miss )", - "Category": "TMA", - "CountDomain": "Clocks_Retired", - "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "FetchLat;MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Code_STLB_Miss", - "LegacyName": "metric_TMA_......Code_STLB_Miss(%)", - "ParentCategory": "ITLB_Misses", - "Level": 4, - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ITLB_MISSES.WALK_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " itlb_misses.walk_active / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Retired", - "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "FetchLat;MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Code_STLB_Miss_4K", - "LegacyName": "metric_TMA_........Code_STLB_Miss_4K(%)", - "ParentCategory": "Code_STLB_Miss", - "Level": 5, - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ITLB_MISSES.WALK_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - }, - { - "Name": "ITLB_MISSES.WALK_COMPLETED_4K", - "Alias": "c" - }, - { - "Name": "ITLB_MISSES.WALK_COMPLETED_2M_4M", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * ( ( a / ( b ) ) * c / ( c + d ) )", - "BaseFormula": " tma_code_stlb_miss * itlb_misses.walk_completed_4k / ( itlb_misses.walk_completed_4k + itlb_misses.walk_completed_2m_4m )", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "FetchLat;MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Code_STLB_Miss_2M", - "LegacyName": "metric_TMA_........Code_STLB_Miss_2M(%)", - "ParentCategory": "Code_STLB_Miss", - "Level": 5, - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ITLB_MISSES.WALK_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - }, - { - "Name": "ITLB_MISSES.WALK_COMPLETED_2M_4M", - "Alias": "c" - }, - { - "Name": "ITLB_MISSES.WALK_COMPLETED_4K", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * ( ( a / ( b ) ) * c / ( d + c ) )", - "BaseFormula": " tma_code_stlb_miss * itlb_misses.walk_completed_2m_4m / ( itlb_misses.walk_completed_4k + itlb_misses.walk_completed_2m_4m )", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "FetchLat;MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Branch_Resteers", - "LegacyName": "metric_TMA_....Branch_Resteers(%)", - "ParentCategory": "Fetch_Latency", - "Level": 3, - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "INT_MISC.CLEAR_RESTEER_CYCLES", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - }, - { - "Name": "INT_MISC.UNKNOWN_BRANCH_CYCLES", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) + ( c / ( b ) ) )", - "BaseFormula": " int_misc.clear_resteer_cycles / tma_info_thread_clks + tma_unknown_branches", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "FetchLat", - "LocateWith": "BR_MISP_RETIRED.ALL_BRANCHES" - }, - { - "MetricName": "Mispredicts_Resteers", - "LegacyName": "metric_TMA_......Mispredicts_Resteers(%)", - "ParentCategory": "Branch_Resteers", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. ", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.BRANCH_MISPREDICTS", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - }, - { - "Name": "INT_MISC.CLEAR_RESTEER_CYCLES", - "Alias": "h" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "i" - } - ], - "Constants": [], - "Formula": "100 * ( ( ( a / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) * h / ( i ) )", - "BaseFormula": " ( tma_branch_mispredicts / tma_bad_speculation ) * int_misc.clear_resteer_cycles / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BadSpec;BrMispredicts;BvMP", - "LocateWith": " INT_MISC.CLEAR_RESTEER_CYCLES" - }, - { - "MetricName": "Clears_Resteers", - "LegacyName": "metric_TMA_......Clears_Resteers(%)", - "ParentCategory": "Branch_Resteers", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. ", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.BRANCH_MISPREDICTS", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - }, - { - "Name": "INT_MISC.CLEAR_RESTEER_CYCLES", - "Alias": "h" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "i" - } - ], - "Constants": [], - "Formula": "100 * ( ( 1 - ( ( a / ( b + c + d + e ) ) / ( max( 1 - ( ( b / ( b + c + d + e ) - f / ( g ) ) + ( e / ( b + c + d + e ) ) + ( d / ( b + c + d + e ) ) ) , 0 ) ) ) ) * h / ( i ) )", - "BaseFormula": " ( 1 - ( tma_branch_mispredicts / tma_bad_speculation ) ) * int_misc.clear_resteer_cycles / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BadSpec;MachineClears", - "LocateWith": " INT_MISC.CLEAR_RESTEER_CYCLES" - }, - { - "MetricName": "Unknown_Branches", - "LegacyName": "metric_TMA_......Unknown_Branches(%)", - "ParentCategory": "Branch_Resteers", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "INT_MISC.UNKNOWN_BRANCH_CYCLES", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " int_misc.unknown_branch_cycles / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BigFootprint;BvBC;FetchLat", - "LocateWith": " FRONTEND_RETIRED.UNKNOWN_BRANCH" - }, - { - "MetricName": "MS_Switches", - "LegacyName": "metric_TMA_....MS_Switches(%)", - "ParentCategory": "Fetch_Latency", - "Level": 3, - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UOPS_RETIRED.MS:c1:e1", - "Alias": "a" - }, - { - "Name": "UOPS_RETIRED.SLOTS", - "Alias": "b" - }, - { - "Name": "UOPS_ISSUED.ANY", - "Alias": "c" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * ( ( 3 ) * a / ( b / c ) / ( d ) )", - "BaseFormula": " ( 3 ) * uops_retired.ms:c1:e1 / ( uops_retired.slots / uops_issued.any ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "FetchLat;MicroSeq", - "LocateWith": " FRONTEND_RETIRED.MS_FLOWS " - }, - { - "MetricName": "LCP", - "LegacyName": "metric_TMA_....LCP(%)", - "ParentCategory": "Fetch_Latency", - "Level": 3, - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. ", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "DECODE.LCP", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " decode.lcp / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "FetchLat", - "LocateWith": "" - }, - { - "MetricName": "DSB_Switches", - "LegacyName": "metric_TMA_....DSB_Switches(%)", - "ParentCategory": "Fetch_Latency", - "Level": 3, - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "DSB2MITE_SWITCHES.PENALTY_CYCLES", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " dsb2mite_switches.penalty_cycles / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "DSBmiss;FetchLat", - "LocateWith": " FRONTEND_RETIRED.DSB_MISS" - }, - { - "MetricName": "Fetch_Bandwidth", - "LegacyName": "metric_TMA_..Fetch_Bandwidth(%)", - "ParentCategory": "Frontend_Bound", - "Level": 2, - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "e" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "f" - }, - { - "Name": "PERF_METRICS.FETCH_LATENCY", - "Alias": "g" - } - ], - "Constants": [], - "Formula": "100 * ( max( 0 , ( a / ( a + b + c + d ) - e / ( f ) ) - ( ( g / ( a + b + c + d ) - e / ( f ) ) ) ) )", - "BaseFormula": " max( 0 , tma_frontend_bound - tma_fetch_latency )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_..Fetch_Bandwidth(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "FetchBW;Frontend;TmaL2", - "LocateWith": " FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATENCY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2" - }, - { - "MetricName": "MITE", - "LegacyName": "metric_TMA_....MITE(%)", - "ParentCategory": "Fetch_Bandwidth", - "Level": 3, - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "IDQ.MITE_CYCLES_ANY", - "Alias": "a" - }, - { - "Name": "IDQ.MITE_CYCLES_OK", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "c" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "d" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( a - b ) / ( c if smt_on else ( d ) ) / 2 )", - "BaseFormula": " ( idq.mite_cycles_any - idq.mite_cycles_ok ) / tma_info_core_core_clks / 2", - "Category": "TMA", - "CountDomain": "Slots_Estimated", - "Threshold": { - "Formula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "DSBmiss;FetchBW", - "LocateWith": " FRONTEND_RETIRED.ANY_DSB_MISS" - }, - { - "MetricName": "Decoder0_Alone", - "LegacyName": "metric_TMA_......Decoder0_Alone(%)", - "ParentCategory": "MITE", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "INST_DECODED.DECODERS:c1", - "Alias": "a" - }, - { - "Name": "INST_DECODED.DECODERS:c2", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "c" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "d" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( a - b ) / ( c if smt_on else ( d ) ) / 2 )", - "BaseFormula": " ( inst_decoded.decoders:c1 - inst_decoded.decoders:c2 ) / tma_info_core_core_clks / 2", - "Category": "TMA", - "CountDomain": "Slots_Estimated", - "Threshold": { - "Formula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "DSBmiss;FetchBW", - "LocateWith": "" - }, - { - "MetricName": "DSB", - "LegacyName": "metric_TMA_....DSB(%)", - "ParentCategory": "Fetch_Bandwidth", - "Level": 3, - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "IDQ.DSB_CYCLES_ANY", - "Alias": "a" - }, - { - "Name": "IDQ.DSB_CYCLES_OK", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "c" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "d" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( a - b ) / ( c if smt_on else ( d ) ) / 2 )", - "BaseFormula": " ( idq.dsb_cycles_any - idq.dsb_cycles_ok ) / tma_info_core_core_clks / 2", - "Category": "TMA", - "CountDomain": "Slots_Estimated", - "Threshold": { - "Formula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "DSB;FetchBW", - "LocateWith": "" - }, - { - "MetricName": "MS", - "LegacyName": "metric_TMA_....MS(%)", - "ParentCategory": "Fetch_Bandwidth", - "Level": 3, - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "IDQ.MS_CYCLES_ANY", - "Alias": "a" - }, - { - "Name": "UOPS_RETIRED.MS:c1", - "Alias": "b" - }, - { - "Name": "UOPS_RETIRED.SLOTS", - "Alias": "c" - }, - { - "Name": "UOPS_ISSUED.ANY", - "Alias": "d" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "e" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "f" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( max( a , b / ( c / d ) ) / ( e if smt_on else ( f ) ) / 2 )", - "BaseFormula": " max( idq.ms_cycles_any , uops_retired.ms:c1 / ( uops_retired.slots / uops_issued.any ) ) / tma_info_core_core_clks / 2", - "Category": "TMA", - "CountDomain": "Slots_Estimated", - "Threshold": { - "Formula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MicroSeq", - "LocateWith": "" - }, - { - "MetricName": "Bad_Speculation", - "LegacyName": "metric_TMA_Bad_Speculation(%)", - "Level": 1, - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "e" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "f" - } - ], - "Constants": [], - "Formula": "100 * ( max( 1 - ( ( a / ( a + b + c + d ) - e / ( f ) ) + ( d / ( a + b + c + d ) ) + ( c / ( a + b + c + d ) ) ) , 0 ) )", - "BaseFormula": " max( 1 - ( tma_frontend_bound + tma_backend_bound + tma_retiring ) , 0 )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_Bad_Speculation(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "TmaL1", - "LocateWith": "#NA" - }, - { - "MetricName": "Branch_Mispredicts", - "LegacyName": "metric_TMA_..Branch_Mispredicts(%)", - "ParentCategory": "Bad_Speculation", - "Level": 2, - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.BRANCH_MISPREDICTS", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b + c + d + e ) )", - "BaseFormula": " perf_metrics.branch_mispredicts / ( perf_metrics.frontend_bound + perf_metrics.bad_speculation + perf_metrics.retiring + perf_metrics.backend_bound )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2", - "LocateWith": " TOPDOWN.BR_MISPREDICT_SLOTS" - }, - { - "MetricName": "Other_Mispredicts", - "LegacyName": "metric_TMA_....Other_Mispredicts(%)", - "ParentCategory": "Branch_Mispredicts", - "Level": 3, - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.BRANCH_MISPREDICTS", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - }, - { - "Name": "BR_MISP_RETIRED.ALL_BRANCHES", - "Alias": "f" - }, - { - "Name": "INT_MISC.CLEARS_COUNT", - "Alias": "g" - }, - { - "Name": "MACHINE_CLEARS.COUNT", - "Alias": "h" - } - ], - "Constants": [], - "Formula": "100 * ( max( ( a / ( b + c + d + e ) ) * ( 1 - f / ( g - h ) ) , 0.0001 ) )", - "BaseFormula": " max( tma_branch_mispredicts * ( 1 - br_misp_retired.all_branches / ( int_misc.clears_count - machine_clears.count ) ) , 0.0001 )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvIO;BrMispredicts", - "LocateWith": "" - }, - { - "MetricName": "Machine_Clears", - "LegacyName": "metric_TMA_..Machine_Clears(%)", - "ParentCategory": "Bad_Speculation", - "Level": 2, - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "e" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "f" - }, - { - "Name": "PERF_METRICS.BRANCH_MISPREDICTS", - "Alias": "g" - } - ], - "Constants": [], - "Formula": "100 * ( max( 0 , ( max( 1 - ( ( a / ( a + b + c + d ) - e / ( f ) ) + ( d / ( a + b + c + d ) ) + ( c / ( a + b + c + d ) ) ) , 0 ) ) - ( g / ( a + b + c + d ) ) ) )", - "BaseFormula": " max( 0 , tma_bad_speculation - tma_branch_mispredicts )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2", - "LocateWith": "MACHINE_CLEARS.COUNT" - }, - { - "MetricName": "Other_Nukes", - "LegacyName": "metric_TMA_....Other_Nukes(%)", - "ParentCategory": "Machine_Clears", - "Level": 3, - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "e" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "f" - }, - { - "Name": "PERF_METRICS.BRANCH_MISPREDICTS", - "Alias": "g" - }, - { - "Name": "MACHINE_CLEARS.MEMORY_ORDERING", - "Alias": "h" - }, - { - "Name": "MACHINE_CLEARS.COUNT", - "Alias": "i" - } - ], - "Constants": [], - "Formula": "100 * ( max( ( max( 0 , ( max( 1 - ( ( a / ( a + b + c + d ) - e / ( f ) ) + ( d / ( a + b + c + d ) ) + ( c / ( a + b + c + d ) ) ) , 0 ) ) - ( g / ( a + b + c + d ) ) ) ) * ( 1 - h / i ) , 0.0001 ) )", - "BaseFormula": " max( tma_machine_clears * ( 1 - machine_clears.memory_ordering / machine_clears.count ) , 0.0001 )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvIO;Machine_Clears", - "LocateWith": "" - }, - { - "MetricName": "Backend_Bound", - "LegacyName": "metric_TMA_Backend_Bound(%)", - "Level": 1, - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b + c + d + a ) )", - "BaseFormula": " perf_metrics.backend_bound / ( perf_metrics.frontend_bound + perf_metrics.bad_speculation + perf_metrics.retiring + perf_metrics.backend_bound )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvOB;TmaL1", - "LocateWith": " TOPDOWN.BACKEND_BOUND_SLOTS" - }, - { - "MetricName": "Memory_Bound", - "LegacyName": "metric_TMA_..Memory_Bound(%)", - "ParentCategory": "Backend_Bound", - "Level": 2, - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two).", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.MEMORY_BOUND", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b + c + d + e ) )", - "BaseFormula": " perf_metrics.memory_bound / ( perf_metrics.frontend_bound + perf_metrics.bad_speculation + perf_metrics.retiring + perf_metrics.backend_bound )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Backend;TmaL2", - "LocateWith": "#NA" - }, - { - "MetricName": "L1_Bound", - "LegacyName": "metric_TMA_....L1_Bound(%)", - "ParentCategory": "Memory_Bound", - "Level": 3, - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "EXE_ACTIVITY.BOUND_ON_LOADS", - "Alias": "a" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L1D_MISS", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "100 * ( max( ( a - b ) / ( c ) , 0 ) )", - "BaseFormula": " max( ( exe_activity.bound_on_loads - memory_activity.stalls_l1d_miss ) / tma_info_thread_clks , 0 )", - "Category": "TMA", - "CountDomain": "Stalls", - "Threshold": { - "Formula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", - "LocateWith": " MEM_LOAD_RETIRED.L1_HIT" - }, - { - "MetricName": "DTLB_Load", - "LegacyName": "metric_TMA_......DTLB_Load(%)", - "ParentCategory": "L1_Bound", - "Level": 4, - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "DTLB_LOAD_MISSES.STLB_HIT:c1", - "Alias": "a" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_ACTIVE", - "Alias": "b" - }, - { - "Name": "CYCLE_ACTIVITY.CYCLES_MEM_ANY", - "Alias": "c" - }, - { - "Name": "MEMORY_ACTIVITY.CYCLES_L1D_MISS", - "Alias": "d" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "e" - } - ], - "Constants": [], - "Formula": "100 * ( min( ( 7 ) * a + b , max( c - d , 0 ) ) / ( e ) )", - "BaseFormula": " min( ( 7 ) * dtlb_load_misses.stlb_hit:c1 + dtlb_load_misses.walk_active , max( cycle_activity.cycles_mem_any - memory_activity.cycles_l1d_miss , 0 ) ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvMT;MemoryTLB", - "LocateWith": " MEM_INST_RETIRED.STLB_MISS_LOADS" - }, - { - "MetricName": "Load_STLB_Hit", - "LegacyName": "metric_TMA_........Load_STLB_Hit(%)", - "ParentCategory": "DTLB_Load", - "Level": 5, - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "DTLB_LOAD_MISSES.STLB_HIT:c1", - "Alias": "a" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_ACTIVE", - "Alias": "b" - }, - { - "Name": "CYCLE_ACTIVITY.CYCLES_MEM_ANY", - "Alias": "c" - }, - { - "Name": "MEMORY_ACTIVITY.CYCLES_L1D_MISS", - "Alias": "d" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "e" - } - ], - "Constants": [], - "Formula": "100 * ( ( min( ( 7 ) * a + b , max( c - d , 0 ) ) / ( e ) ) - ( b / ( e ) ) )", - "BaseFormula": " tma_dtlb_load - tma_load_stlb_miss", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Load_STLB_Miss", - "LegacyName": "metric_TMA_........Load_STLB_Miss(%)", - "ParentCategory": "DTLB_Load", - "Level": 5, - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "DTLB_LOAD_MISSES.WALK_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " dtlb_load_misses.walk_active / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Calculated", - "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Load_STLB_Miss_4K", - "LegacyName": "metric_TMA_..........Load_STLB_Miss_4K(%)", - "ParentCategory": "Load_STLB_Miss", - "Level": 6, - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "DTLB_LOAD_MISSES.WALK_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_COMPLETED_4K", - "Alias": "c" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M", - "Alias": "d" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_COMPLETED_1G", - "Alias": "e" - } - ], - "Constants": [], - "Formula": "100 * ( ( a / ( b ) ) * c / ( c + d + e ) )", - "BaseFormula": " tma_load_stlb_miss * dtlb_load_misses.walk_completed_4k / ( dtlb_load_misses.walk_completed_4k + dtlb_load_misses.walk_completed_2m_4m + dtlb_load_misses.walk_completed_1g )", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Load_STLB_Miss_2M", - "LegacyName": "metric_TMA_..........Load_STLB_Miss_2M(%)", - "ParentCategory": "Load_STLB_Miss", - "Level": 6, - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "DTLB_LOAD_MISSES.WALK_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M", - "Alias": "c" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_COMPLETED_4K", - "Alias": "d" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_COMPLETED_1G", - "Alias": "e" - } - ], - "Constants": [], - "Formula": "100 * ( ( a / ( b ) ) * c / ( d + c + e ) )", - "BaseFormula": " tma_load_stlb_miss * dtlb_load_misses.walk_completed_2m_4m / ( dtlb_load_misses.walk_completed_4k + dtlb_load_misses.walk_completed_2m_4m + dtlb_load_misses.walk_completed_1g )", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Load_STLB_Miss_1G", - "LegacyName": "metric_TMA_..........Load_STLB_Miss_1G(%)", - "ParentCategory": "Load_STLB_Miss", - "Level": 6, - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "DTLB_LOAD_MISSES.WALK_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_COMPLETED_1G", - "Alias": "c" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_COMPLETED_4K", - "Alias": "d" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M", - "Alias": "e" - } - ], - "Constants": [], - "Formula": "100 * ( ( a / ( b ) ) * c / ( d + e + c ) )", - "BaseFormula": " tma_load_stlb_miss * dtlb_load_misses.walk_completed_1g / ( dtlb_load_misses.walk_completed_4k + dtlb_load_misses.walk_completed_2m_4m + dtlb_load_misses.walk_completed_1g )", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Store_Fwd_Blk", - "LegacyName": "metric_TMA_......Store_Fwd_Blk(%)", - "ParentCategory": "L1_Bound", - "Level": 4, - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "LD_BLOCKS.STORE_FORWARD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( 13 * a / ( b ) )", - "BaseFormula": " 13 * ld_blocks.store_forward / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": "" - }, - { - "MetricName": "L1_Latency_Dependency", - "LegacyName": "metric_TMA_......L1_Latency_Dependency(%)", - "ParentCategory": "L1_Bound", - "Level": 4, - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "MEM_INST_RETIRED.ALL_LOADS", - "Alias": "a" - }, - { - "Name": "MEM_LOAD_RETIRED.FB_HIT", - "Alias": "b" - }, - { - "Name": "MEM_LOAD_RETIRED.L1_MISS", - "Alias": "c" - }, - { - "Name": "CYCLE_ACTIVITY.CYCLES_MEM_ANY", - "Alias": "e" - }, - { - "Name": "MEMORY_ACTIVITY.CYCLES_L1D_MISS", - "Alias": "f" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "g" - } - ], - "Constants": [ - { - "Name": "20", - "Alias": "dependentloadsweight" - } - ], - "Formula": "100 * ( min( 2 * ( a - b - c ) * dependentloadsweight / 100 , max( e - f , 0 ) ) / ( g ) )", - "BaseFormula": " min( 2 * ( mem_inst_retired.all_loads - mem_load_retired.fb_hit - mem_load_retired.l1_miss ) * 20 / 100 , max( cycle_activity.cycles_mem_any - memory_activity.cycles_l1d_miss , 0 ) ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvML;MemoryLat", - "LocateWith": " MEM_LOAD_RETIRED.L1_HIT" - }, - { - "MetricName": "Lock_Latency", - "LegacyName": "metric_TMA_......Lock_Latency(%)", - "ParentCategory": "L1_Bound", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "MEM_INST_RETIRED.LOCK_LOADS", - "Alias": "a" - }, - { - "Name": "L2_RQSTS.ALL_RFO", - "Alias": "b" - }, - { - "Name": "MEM_INST_RETIRED.ALL_STORES", - "Alias": "c" - }, - { - "Name": "L2_RQSTS.RFO_HIT", - "Alias": "d" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "e" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO", - "Alias": "f" - } - ], - "Constants": [], - "Formula": "100 * ( ( 16 * max( 0 , a - b ) + ( a / c ) * ( ( 10 ) * d + ( min( e , f ) ) ) ) / ( e ) )", - "BaseFormula": " ( 16 * max( 0 , mem_inst_retired.lock_loads - l2_rqsts.all_rfo ) + ( mem_inst_retired.lock_loads / mem_inst_retired.all_stores ) * ( ( 10 ) * l2_rqsts.rfo_hit + ( min( cpu_clk_unhalted.thread , offcore_requests_outstanding.cycles_with_demand_rfo ) ) ) ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "LockCont;Offcore", - "LocateWith": " MEM_INST_RETIRED.LOCK_LOADS" - }, - { - "MetricName": "Split_Loads", - "LegacyName": "metric_TMA_......Split_Loads(%)", - "ParentCategory": "L1_Bound", - "Level": 4, - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. ", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "L1D_PEND_MISS.PENDING", - "Alias": "a" - }, - { - "Name": "MEM_LOAD_COMPLETED.L1_MISS_ANY", - "Alias": "b" - }, - { - "Name": "LD_BLOCKS.NO_SR", - "Alias": "c" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * ( ( a / b ) * c / ( d ) )", - "BaseFormula": " tma_info_memory_load_miss_real_latency * ld_blocks.no_sr / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Calculated", - "Threshold": { - "Formula": "metric_TMA_......Split_Loads(%) > 30" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": " MEM_INST_RETIRED.SPLIT_LOADS" - }, - { - "MetricName": "FB_Full", - "LegacyName": "metric_TMA_......FB_Full(%)", - "ParentCategory": "L1_Bound", - "Level": 4, - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory).", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "L1D_PEND_MISS.FB_FULL", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " l1d_pend_miss.fb_full / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Calculated", - "Threshold": { - "Formula": "metric_TMA_......FB_Full(%) > 30" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvMB;MemoryBW", - "LocateWith": "" - }, - { - "MetricName": "L2_Bound", - "LegacyName": "metric_TMA_....L2_Bound(%)", - "ParentCategory": "Memory_Bound", - "Level": 3, - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "MEMORY_ACTIVITY.STALLS_L1D_MISS", - "Alias": "a" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L2_MISS", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "100 * ( ( a - b ) / ( c ) )", - "BaseFormula": " ( memory_activity.stalls_l1d_miss - memory_activity.stalls_l2_miss ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Stalls", - "Threshold": { - "Formula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem", - "LocateWith": " MEM_LOAD_RETIRED.L2_HIT" - }, - { - "MetricName": "L2_Hit_Latency", - "LegacyName": "metric_TMA_......L2_Hit_Latency(%)", - "ParentCategory": "L2_Bound", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "b" - }, - { - "Name": "MEM_LOAD_RETIRED.L2_HIT", - "Alias": "e" - }, - { - "Name": "MEM_LOAD_RETIRED.FB_HIT", - "Alias": "f" - }, - { - "Name": "MEM_LOAD_RETIRED.L1_MISS", - "Alias": "g" - } - ], - "Constants": [ - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "c" - }, - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "100 * ( ( 4.4 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * e * ( 1 + ( f / g ) / 2 ) / ( a ) )", - "BaseFormula": " ( 4.4 * tma_info_system_core_frequency ) * mem_load_retired.l2_hit * ( 1 + ( mem_load_retired.fb_hit / mem_load_retired.l1_miss ) / 2 ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Retired", - "Threshold": { - "Formula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryLat", - "LocateWith": " MEM_LOAD_RETIRED.L2_HIT" - }, - { - "MetricName": "L3_Bound", - "LegacyName": "metric_TMA_....L3_Bound(%)", - "ParentCategory": "Memory_Bound", - "Level": 3, - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "MEMORY_ACTIVITY.STALLS_L2_MISS", - "Alias": "a" - }, - { - "Name": "MEMORY_ACTIVITY.STALLS_L3_MISS", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "100 * ( ( a - b ) / ( c ) )", - "BaseFormula": " ( memory_activity.stalls_l2_miss - memory_activity.stalls_l3_miss ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Stalls", - "Threshold": { - "Formula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", - "LocateWith": " MEM_LOAD_RETIRED.L3_HIT" - }, - { - "MetricName": "Contested_Accesses", - "LegacyName": "metric_TMA_......Contested_Accesses(%)", - "ParentCategory": "L3_Bound", - "Level": 4, - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "b" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD", - "Alias": "e" - }, - { - "Name": "OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM", - "Alias": "f" - }, - { - "Name": "OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD", - "Alias": "g" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS", - "Alias": "h" - }, - { - "Name": "MEM_LOAD_RETIRED.FB_HIT", - "Alias": "i" - }, - { - "Name": "MEM_LOAD_RETIRED.L1_MISS", - "Alias": "j" - } - ], - "Constants": [ - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "c" - }, - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "100 * ( ( ( ( 81 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( e * ( f / ( f + g ) ) ) + ( ( 79 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( h ) ) * ( 1 + ( i / j ) / 2 ) / ( a ) )", - "BaseFormula": " ( ( ( 81 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( mem_load_l3_hit_retired.xsnp_fwd * ( ocr.demand_data_rd.l3_hit.snoop_hitm / ( ocr.demand_data_rd.l3_hit.snoop_hitm + ocr.demand_data_rd.l3_hit.snoop_hit_with_fwd ) ) ) + ( ( 79 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( mem_load_l3_hit_retired.xsnp_miss ) ) * ( 1 + ( mem_load_retired.fb_hit / mem_load_retired.l1_miss ) / 2 ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", - "LocateWith": " MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS" - }, - { - "MetricName": "Data_Sharing", - "LegacyName": "metric_TMA_......Data_Sharing(%)", - "ParentCategory": "L3_Bound", - "Level": 4, - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "b" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD", - "Alias": "e" - }, - { - "Name": "MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD", - "Alias": "f" - }, - { - "Name": "OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM", - "Alias": "g" - }, - { - "Name": "OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD", - "Alias": "h" - }, - { - "Name": "MEM_LOAD_RETIRED.FB_HIT", - "Alias": "i" - }, - { - "Name": "MEM_LOAD_RETIRED.L1_MISS", - "Alias": "j" - } - ], - "Constants": [ - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "c" - }, - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "100 * ( ( ( 79 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( e + f * ( 1 - ( g / ( g + h ) ) ) ) * ( 1 + ( i / j ) / 2 ) / ( a ) )", - "BaseFormula": " ( ( 79 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( mem_load_l3_hit_retired.xsnp_no_fwd + mem_load_l3_hit_retired.xsnp_fwd * ( 1 - ( ocr.demand_data_rd.l3_hit.snoop_hitm / ( ocr.demand_data_rd.l3_hit.snoop_hitm + ocr.demand_data_rd.l3_hit.snoop_hit_with_fwd ) ) ) ) * ( 1 + ( mem_load_retired.fb_hit / mem_load_retired.l1_miss ) / 2 ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvMS;Offcore;Snoop", - "LocateWith": " MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD" - }, - { - "MetricName": "L3_Hit_Latency", - "LegacyName": "metric_TMA_......L3_Hit_Latency(%)", - "ParentCategory": "L3_Bound", - "Level": 4, - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "b" - }, - { - "Name": "MEM_LOAD_RETIRED.L3_HIT", - "Alias": "e" - }, - { - "Name": "MEM_LOAD_RETIRED.FB_HIT", - "Alias": "f" - }, - { - "Name": "MEM_LOAD_RETIRED.L1_MISS", - "Alias": "g" - } - ], - "Constants": [ - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "c" - }, - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "100 * ( ( ( 37 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 4.4 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * ( e * ( 1 + ( f / g ) / 2 ) ) / ( a ) )", - "BaseFormula": " ( ( 37 * tma_info_system_core_frequency ) - ( 4.4 * tma_info_system_core_frequency ) ) * ( mem_load_retired.l3_hit * ( 1 + ( mem_load_retired.fb_hit / mem_load_retired.l1_miss ) / 2 ) ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvML;MemoryLat", - "LocateWith": " MEM_LOAD_RETIRED.L3_HIT" - }, - { - "MetricName": "SQ_Full", - "LegacyName": "metric_TMA_......SQ_Full(%)", - "ParentCategory": "L3_Bound", - "Level": 4, - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "XQ.FULL_CYCLES", - "Alias": "a" - }, - { - "Name": "L1D_PEND_MISS.L2_STALLS", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "100 * ( ( a + b ) / ( c ) )", - "BaseFormula": " ( xq.full_cycles + l1d_pend_miss.l2_stalls ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvMB;MemoryBW;Offcore", - "LocateWith": "" - }, - { - "MetricName": "HBM_Bound", - "LegacyName": "metric_TMA_....HBM_Bound(%)", - "Level": 3, - "BriefDescription": "This metric estimates how often the CPU was stalled due to High Bandwidth Memory (HBM) accesses by loads.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "MEMORY_ACTIVITY.STALLS_L3_MISS", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - }, - { - "Name": "OCR.DEMAND_DATA_RD.PMM", - "Alias": "c" - }, - { - "Name": "OCR.READS_TO_CORE.L3_MISS", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * ( ( a / ( b ) ) * c / d )", - "BaseFormula": " ( memory_activity.stalls_l3_miss / tma_info_thread_clks ) * ocr.demand_data_rd.pmm / ocr.reads_to_core.l3_miss", - "Category": "TMA", - "CountDomain": "Stalls", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_....HBM_Bound(%)" - }, - { - "Alias": "b", - "Value": "metric_TMA_..Memory_Bound(%)" - }, - { - "Alias": "c", - "Value": "metric_TMA_Backend_Bound(%)" - } - ], - "Formula": "a > 10 & b > 20 & c > 20", - "BaseFormula": "metric_TMA_....HBM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvMB;BvML;MemoryBound;Offcore;Server;TmaL3mem", - "LocateWith": "" - }, - { - "MetricName": "DRAM_Bound", - "LegacyName": "metric_TMA_....DRAM_Bound(%)", - "ParentCategory": "Memory_Bound", - "Level": 3, - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "MEMORY_ACTIVITY.STALLS_L3_MISS", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - }, - { - "Name": "OCR.DEMAND_DATA_RD.PMM", - "Alias": "c" - }, - { - "Name": "OCR.READS_TO_CORE.L3_MISS", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * ( ( a / ( b ) ) - ( ( a / ( b ) ) * c / d ) )", - "BaseFormula": " ( memory_activity.stalls_l3_miss / tma_info_thread_clks ) - tma_hbm_bound", - "Category": "TMA", - "CountDomain": "Stalls", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_....DRAM_Bound(%)" - }, - { - "Alias": "b", - "Value": "metric_TMA_..Memory_Bound(%)" - }, - { - "Alias": "c", - "Value": "metric_TMA_Backend_Bound(%)" - } - ], - "Formula": "a > 10 & b > 20 & c > 20", - "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryBound;TmaL3mem", - "LocateWith": " MEM_LOAD_RETIRED.L3_MISS" - }, - { - "MetricName": "MEM_Bandwidth", - "LegacyName": "metric_TMA_......MEM_Bandwidth(%)", - "Level": 4, - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that).", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD:c4", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( ( min( a , b ) ) / ( a ) )", - "BaseFormula": " ( min( cpu_clk_unhalted.thread , offcore_requests_outstanding.all_data_rd:c4 ) ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvMB;MemoryBW;Offcore", - "LocateWith": "" - }, - { - "MetricName": "MBA_Stalls", - "LegacyName": "metric_TMA_........MBA_Stalls(%)", - "ParentCategory": "MEM_Bandwidth", - "Level": 5, - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to memory bandwidth Allocation feature (RDT's memory bandwidth throttling).", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "INT_MISC.MBA_STALLS", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " int_misc.mba_stalls / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_........MBA_Stalls(%) > 10 & metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryBW;Offcore;Server", - "LocateWith": "" - }, - { - "MetricName": "MEM_Latency", - "LegacyName": "metric_TMA_......MEM_Latency(%)", - "Level": 4, - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that).", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", - "Alias": "b" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD:c4", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "100 * ( ( min( a , b ) ) / ( a ) - ( ( min( a , c ) ) / ( a ) ) )", - "BaseFormula": " ( min( cpu_clk_unhalted.thread , offcore_requests_outstanding.cycles_with_data_rd ) ) / tma_info_thread_clks - tma_mem_bandwidth", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvML;MemoryLat;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Local_MEM", - "LegacyName": "metric_TMA_........Local_MEM(%)", - "ParentCategory": "MEM_Latency", - "Level": 5, - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from local memory. Caching will improve the latency and increase performance.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "b" - }, - { - "Name": "MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM", - "Alias": "e" - }, - { - "Name": "MEM_LOAD_RETIRED.FB_HIT", - "Alias": "f" - }, - { - "Name": "MEM_LOAD_RETIRED.L1_MISS", - "Alias": "g" - } - ], - "Constants": [ - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "c" - }, - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "100 * ( ( ( 109 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * e * ( 1 + ( f / g ) / 2 ) / ( a ) )", - "BaseFormula": " ( ( 109 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * mem_load_l3_miss_retired.local_dram * ( 1 + ( mem_load_retired.fb_hit / mem_load_retired.l1_miss ) / 2 ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_........Local_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Server", - "LocateWith": " MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM" - }, - { - "MetricName": "Remote_MEM", - "LegacyName": "metric_TMA_........Remote_MEM(%)", - "ParentCategory": "MEM_Latency", - "Level": 5, - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote memory. This is caused often due to non-optimal NUMA allocations. ", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "b" - }, - { - "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM", - "Alias": "e" - }, - { - "Name": "MEM_LOAD_RETIRED.FB_HIT", - "Alias": "f" - }, - { - "Name": "MEM_LOAD_RETIRED.L1_MISS", - "Alias": "g" - } - ], - "Constants": [ - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "c" - }, - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "100 * ( ( ( 190 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * e * ( 1 + ( f / g ) / 2 ) / ( a ) )", - "BaseFormula": " ( ( 190 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * mem_load_l3_miss_retired.remote_dram * ( 1 + ( mem_load_retired.fb_hit / mem_load_retired.l1_miss ) / 2 ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_........Remote_MEM(%) > 10 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Server;Snoop", - "LocateWith": " MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM" - }, - { - "MetricName": "Remote_Cache", - "LegacyName": "metric_TMA_........Remote_Cache(%)", - "ParentCategory": "MEM_Latency", - "Level": 5, - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling loads from remote cache in other sockets including synchronizations issues. This is caused often due to non-optimal NUMA allocations. ", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "b" - }, - { - "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM", - "Alias": "e" - }, - { - "Name": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD", - "Alias": "f" - }, - { - "Name": "MEM_LOAD_RETIRED.FB_HIT", - "Alias": "g" - }, - { - "Name": "MEM_LOAD_RETIRED.L1_MISS", - "Alias": "h" - } - ], - "Constants": [ - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "c" - }, - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "100 * ( ( ( ( 170 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * e + ( ( 170 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) - ( 37 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) ) * f ) * ( 1 + ( g / h ) / 2 ) / ( a ) )", - "BaseFormula": " ( ( ( 170 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * mem_load_l3_miss_retired.remote_hitm + ( ( 170 * tma_info_system_core_frequency ) - ( 37 * tma_info_system_core_frequency ) ) * mem_load_l3_miss_retired.remote_fwd ) * ( 1 + ( mem_load_retired.fb_hit / mem_load_retired.l1_miss ) / 2 ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_........Remote_Cache(%) > 5 & metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Offcore;Server;Snoop", - "LocateWith": " MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM;MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD" - }, - { - "MetricName": "Store_Bound", - "LegacyName": "metric_TMA_....Store_Bound(%)", - "ParentCategory": "Memory_Bound", - "Level": 3, - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "EXE_ACTIVITY.BOUND_ON_STORES", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " exe_activity.bound_on_stores / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Stalls", - "Threshold": { - "Formula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryBound;TmaL3mem", - "LocateWith": " MEM_INST_RETIRED.ALL_STORES" - }, - { - "MetricName": "Store_Latency", - "LegacyName": "metric_TMA_......Store_Latency(%)", - "ParentCategory": "Store_Bound", - "Level": 4, - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "MEM_STORE_RETIRED.L2_HIT", - "Alias": "a" - }, - { - "Name": "MEM_INST_RETIRED.LOCK_LOADS", - "Alias": "b" - }, - { - "Name": "MEM_INST_RETIRED.ALL_STORES", - "Alias": "c" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "d" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO", - "Alias": "e" - } - ], - "Constants": [], - "Formula": "100 * ( ( ( a * ( 10 ) * ( 1 - ( b / c ) ) ) + ( 1 - ( b / c ) ) * ( min( d , e ) ) ) / ( d ) )", - "BaseFormula": " ( ( mem_store_retired.l2_hit * ( 10 ) * ( 1 - ( mem_inst_retired.lock_loads / mem_inst_retired.all_stores ) ) ) + ( 1 - ( mem_inst_retired.lock_loads / mem_inst_retired.all_stores ) ) * ( min( cpu_clk_unhalted.thread , offcore_requests_outstanding.cycles_with_demand_rfo ) ) ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvML;LockCont;MemoryLat;Offcore", - "LocateWith": "" - }, - { - "MetricName": "False_Sharing", - "LegacyName": "metric_TMA_......False_Sharing(%)", - "ParentCategory": "Store_Bound", - "Level": 4, - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. ", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "b" - }, - { - "Name": "OCR.DEMAND_RFO.L3_MISS:ocr_msr_val=0x103b800002", - "Alias": "e" - }, - { - "Name": "OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM", - "Alias": "f" - } - ], - "Constants": [ - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "c" - }, - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "100 * ( ( ( 170 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * e + ( 81 * ( ( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) ) ) * f ) / ( a ) )", - "BaseFormula": " ( ( 170 * tma_info_system_core_frequency ) * ocr.demand_rfo.l3_miss:ocr_msr_val=0x103b800002 + ( 81 * tma_info_system_core_frequency ) * ocr.demand_rfo.l3_hit.snoop_hitm ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", - "LocateWith": " OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM" - }, - { - "MetricName": "Split_Stores", - "LegacyName": "metric_TMA_......Split_Stores(%)", - "ParentCategory": "Store_Bound", - "Level": 4, - "BriefDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "MEM_INST_RETIRED.SPLIT_STORES", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( a / ( b if smt_on else ( c ) ) )", - "BaseFormula": " mem_inst_retired.split_stores / tma_info_core_core_clks", - "Category": "TMA", - "CountDomain": "Core_Utilization", - "Threshold": { - "Formula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": " MEM_INST_RETIRED.SPLIT_STORES" - }, - { - "MetricName": "Streaming_Stores", - "LegacyName": "metric_TMA_......Streaming_Stores(%)", - "ParentCategory": "Store_Bound", - "Level": 4, - "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "OCR.STREAMING_WR.ANY_RESPONSE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( 9 * a / ( b ) )", - "BaseFormula": " 9 * ocr.streaming_wr.any_response / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryBW;Offcore", - "LocateWith": " OCR.STREAMING_WR.ANY_RESPONSE" - }, - { - "MetricName": "DTLB_Store", - "LegacyName": "metric_TMA_......DTLB_Store(%)", - "ParentCategory": "Store_Bound", - "Level": 4, - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "DTLB_STORE_MISSES.STLB_HIT:c1", - "Alias": "a" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "c" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "d" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( ( 7 ) * a + b ) / ( c if smt_on else ( d ) ) )", - "BaseFormula": " ( ( 7 ) * dtlb_store_misses.stlb_hit:c1 + dtlb_store_misses.walk_active ) / tma_info_core_core_clks", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvMT;MemoryTLB", - "LocateWith": " MEM_INST_RETIRED.STLB_MISS_STORES" - }, - { - "MetricName": "Store_STLB_Hit", - "LegacyName": "metric_TMA_........Store_STLB_Hit(%)", - "ParentCategory": "DTLB_Store", - "Level": 5, - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "DTLB_STORE_MISSES.STLB_HIT:c1", - "Alias": "a" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "c" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "d" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( ( ( 7 ) * a + b ) / ( c if smt_on else ( d ) ) ) - ( b / ( c if smt_on else ( d ) ) ) )", - "BaseFormula": " tma_dtlb_store - tma_store_stlb_miss", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Store_STLB_Miss", - "LegacyName": "metric_TMA_........Store_STLB_Miss(%)", - "ParentCategory": "DTLB_Store", - "Level": 5, - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( a / ( b if smt_on else ( c ) ) )", - "BaseFormula": " dtlb_store_misses.walk_active / tma_info_core_core_clks", - "Category": "TMA", - "CountDomain": "Clocks_Calculated", - "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Store_STLB_Miss_4K", - "LegacyName": "metric_TMA_..........Store_STLB_Miss_4K(%)", - "ParentCategory": "Store_STLB_Miss", - "Level": 6, - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_COMPLETED_4K", - "Alias": "d" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M", - "Alias": "e" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_COMPLETED_1G", - "Alias": "f" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( a / ( b if smt_on else ( c ) ) ) * d / ( d + e + f ) )", - "BaseFormula": " tma_store_stlb_miss * dtlb_store_misses.walk_completed_4k / ( dtlb_store_misses.walk_completed_4k + dtlb_store_misses.walk_completed_2m_4m + dtlb_store_misses.walk_completed_1g )", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Store_STLB_Miss_2M", - "LegacyName": "metric_TMA_..........Store_STLB_Miss_2M(%)", - "ParentCategory": "Store_STLB_Miss", - "Level": 6, - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M", - "Alias": "d" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_COMPLETED_4K", - "Alias": "e" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_COMPLETED_1G", - "Alias": "f" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( a / ( b if smt_on else ( c ) ) ) * d / ( e + d + f ) )", - "BaseFormula": " tma_store_stlb_miss * dtlb_store_misses.walk_completed_2m_4m / ( dtlb_store_misses.walk_completed_4k + dtlb_store_misses.walk_completed_2m_4m + dtlb_store_misses.walk_completed_1g )", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Store_STLB_Miss_1G", - "LegacyName": "metric_TMA_..........Store_STLB_Miss_1G(%)", - "ParentCategory": "Store_STLB_Miss", - "Level": 6, - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "DTLB_STORE_MISSES.WALK_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_COMPLETED_1G", - "Alias": "d" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_COMPLETED_4K", - "Alias": "e" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M", - "Alias": "f" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( a / ( b if smt_on else ( c ) ) ) * d / ( e + f + d ) )", - "BaseFormula": " tma_store_stlb_miss * dtlb_store_misses.walk_completed_1g / ( dtlb_store_misses.walk_completed_4k + dtlb_store_misses.walk_completed_2m_4m + dtlb_store_misses.walk_completed_1g )", - "Category": "TMA", - "CountDomain": "Clocks_Estimated", - "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Core_Bound", - "LegacyName": "metric_TMA_..Core_Bound(%)", - "ParentCategory": "Backend_Bound", - "Level": 2, - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations).", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.MEMORY_BOUND", - "Alias": "e" - } - ], - "Constants": [], - "Formula": "100 * ( max( 0 , ( a / ( b + c + d + a ) ) - ( e / ( b + c + d + a ) ) ) )", - "BaseFormula": " max( 0 , tma_backend_bound - tma_memory_bound )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Backend;TmaL2;Compute", - "LocateWith": "" - }, - { - "MetricName": "Divider", - "LegacyName": "metric_TMA_....Divider(%)", - "ParentCategory": "Core_Bound", - "Level": 3, - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ARITH.DIV_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " arith.div_active / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvCB", - "LocateWith": " ARITH.DIVIDER_ACTIVE " - }, - { - "MetricName": "FP_Divider", - "LegacyName": "metric_TMA_......FP_Divider(%)", - "ParentCategory": "Divider", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ARITH.FPDIV_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " arith.fpdiv_active / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": "" - }, - { - "MetricName": "INT_Divider", - "LegacyName": "metric_TMA_......INT_Divider(%)", - "ParentCategory": "Divider", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ARITH.DIV_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - }, - { - "Name": "ARITH.FPDIV_ACTIVE", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "100 * ( ( a / ( b ) ) - ( c / ( b ) ) )", - "BaseFormula": " tma_divider - tma_fp_divider", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": "" - }, - { - "MetricName": "Serializing_Operation", - "LegacyName": "metric_TMA_....Serializing_Operation(%)", - "ParentCategory": "Ports_Utilized_0", - "Level": 3, - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "RESOURCE_STALLS.SCOREBOARD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.C02", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) + ( c / ( b ) ) )", - "BaseFormula": " resource_stalls.scoreboard / tma_info_thread_clks + tma_c02_wait", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvIO;PortsUtil", - "LocateWith": " RESOURCE_STALLS.SCOREBOARD" - }, - { - "MetricName": "Slow_Pause", - "LegacyName": "metric_TMA_......Slow_Pause(%)", - "ParentCategory": "Serializing_Operation", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.PAUSE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " cpu_clk_unhalted.pause / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": " CPU_CLK_UNHALTED.PAUSE_INST" - }, - { - "MetricName": "C01_Wait", - "LegacyName": "metric_TMA_......C01_Wait(%)", - "ParentCategory": "Serializing_Operation", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.1 power-performance optimized state (Faster wakeup time; Smaller power savings).", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.C01", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " cpu_clk_unhalted.c01 / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......C01_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "C0Wait", - "LocateWith": "" - }, - { - "MetricName": "C02_Wait", - "LegacyName": "metric_TMA_......C02_Wait(%)", - "ParentCategory": "Serializing_Operation", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due staying in C0.2 power-performance optimized state (Slower wakeup time; Larger power savings).", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.C02", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " cpu_clk_unhalted.c02 / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......C02_Wait(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "C0Wait", - "LocateWith": "" - }, - { - "MetricName": "Memory_Fence", - "LegacyName": "metric_TMA_......Memory_Fence(%)", - "ParentCategory": "Serializing_Operation", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to LFENCE Instructions.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "MISC2_RETIRED.LFENCE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( 13 * a / ( b ) )", - "BaseFormula": " 13 * misc2_retired.lfence / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......Memory_Fence(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": "" - }, - { - "MetricName": "AMX_Busy", - "LegacyName": "metric_TMA_....AMX_Busy(%)", - "ParentCategory": "Ports_Utilized_0", - "Level": 3, - "BriefDescription": "This metric estimates fraction of cycles where the Advanced Matrix eXtensions (AMX) execution engine was busy with tile (arithmetic) operations", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "EXE.AMX_BUSY", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( a / ( b if smt_on else ( c ) ) )", - "BaseFormula": " exe.amx_busy / tma_info_core_core_clks", - "Category": "TMA", - "CountDomain": "Core_Clocks", - "Threshold": { - "Formula": "metric_TMA_....AMX_Busy(%) > 50 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvCB;Compute;HPC;Server", - "LocateWith": "" - }, - { - "MetricName": "Ports_Utilization", - "LegacyName": "metric_TMA_....Ports_Utilization(%)", - "ParentCategory": "Core_Bound", - "Level": 3, - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "EXE_ACTIVITY.EXE_BOUND_0_PORTS", - "Alias": "a" - }, - { - "Name": "RS.EMPTY_RESOURCE", - "Alias": "b" - }, - { - "Name": "RESOURCE_STALLS.SCOREBOARD", - "Alias": "c" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "d" - }, - { - "Name": "CYCLE_ACTIVITY.STALLS_TOTAL", - "Alias": "e" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_LOADS", - "Alias": "f" - }, - { - "Name": "EXE_ACTIVITY.1_PORTS_UTIL", - "Alias": "g" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "h" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "i" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "j" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "k" - }, - { - "Name": "EXE_ACTIVITY.2_3_PORTS_UTIL", - "Alias": "l" - }, - { - "Name": "ARITH.DIV_ACTIVE", - "Alias": "m" - } - ], - "Constants": [], - "Formula": "100 * ( ( ( ( a + max( b - c , 0 ) ) / ( d ) * ( e - f ) / ( d ) ) * ( d ) + ( g + ( h / ( i + j + h + k ) ) * l ) ) / ( d ) if ( m < ( e - f ) ) else ( g + ( h / ( i + j + h + k ) ) * l ) / ( d ) )", - "BaseFormula": " ( tma_ports_utilized_0 * tma_info_thread_clks + ( exe_activity.1_ports_util + tma_retiring * exe_activity.2_3_ports_util ) ) / tma_info_thread_clks if ( arith.div_active < ( cycle_activity.stalls_total - exe_activity.bound_on_loads ) ) else ( exe_activity.1_ports_util + tma_retiring * exe_activity.2_3_ports_util ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "PortsUtil", - "LocateWith": "" - }, - { - "MetricName": "Ports_Utilized_0", - "LegacyName": "metric_TMA_......Ports_Utilized_0(%)", - "ParentCategory": "Ports_Utilization", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "EXE_ACTIVITY.EXE_BOUND_0_PORTS", - "Alias": "a" - }, - { - "Name": "RS.EMPTY_RESOURCE", - "Alias": "b" - }, - { - "Name": "RESOURCE_STALLS.SCOREBOARD", - "Alias": "c" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "d" - }, - { - "Name": "CYCLE_ACTIVITY.STALLS_TOTAL", - "Alias": "e" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_LOADS", - "Alias": "f" - } - ], - "Constants": [], - "Formula": "100 * ( ( a + max( b - c , 0 ) ) / ( d ) * ( e - f ) / ( d ) )", - "BaseFormula": " ( exe_activity.exe_bound_0_ports + max( rs.empty_resource - resource_stalls.scoreboard , 0 ) ) / tma_info_thread_clks * ( cycle_activity.stalls_total - exe_activity.bound_on_loads ) / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "PortsUtil", - "LocateWith": "" - }, - { - "MetricName": "Mixing_Vectors", - "LegacyName": "metric_TMA_........Mixing_Vectors(%)", - "ParentCategory": "Ports_Utilized_0", - "Level": 5, - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ASSISTS.SSE_AVX_MIX", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( 160 * a / ( b ) )", - "BaseFormula": " 160 * assists.sse_avx_mix / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_........Mixing_Vectors(%) > 5" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": "" - }, - { - "MetricName": "Ports_Utilized_1", - "LegacyName": "metric_TMA_......Ports_Utilized_1(%)", - "ParentCategory": "Ports_Utilization", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "EXE_ACTIVITY.1_PORTS_UTIL", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " exe_activity.1_ports_util / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "PortsUtil", - "LocateWith": " EXE_ACTIVITY.1_PORTS_UTIL" - }, - { - "MetricName": "Ports_Utilized_2", - "LegacyName": "metric_TMA_......Ports_Utilized_2(%)", - "ParentCategory": "Ports_Utilization", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "EXE_ACTIVITY.2_PORTS_UTIL", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " exe_activity.2_ports_util / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "PortsUtil", - "LocateWith": " EXE_ACTIVITY.2_PORTS_UTIL" - }, - { - "MetricName": "Ports_Utilized_3m", - "LegacyName": "metric_TMA_......Ports_Utilized_3m(%)", - "ParentCategory": "Ports_Utilization", - "Level": 4, - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UOPS_EXECUTED.CYCLES_GE_3", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " uops_executed.cycles_ge_3 / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvCB;PortsUtil", - "LocateWith": " UOPS_EXECUTED.CYCLES_GE_3" - }, - { - "MetricName": "ALU_Op_Utilization", - "LegacyName": "metric_TMA_........ALU_Op_Utilization(%)", - "ParentCategory": "Ports_Utilized_3m", - "Level": 5, - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UOPS_DISPATCHED.PORT_0", - "Alias": "a" - }, - { - "Name": "UOPS_DISPATCHED.PORT_1", - "Alias": "b" - }, - { - "Name": "UOPS_DISPATCHED.PORT_5_11", - "Alias": "c" - }, - { - "Name": "UOPS_DISPATCHED.PORT_6", - "Alias": "d" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "e" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "f" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( a + b + c + d ) / ( 5 * ( e if smt_on else ( f ) ) ) )", - "BaseFormula": " ( uops_dispatched.port_0 + uops_dispatched.port_1 + uops_dispatched.port_5_11 + uops_dispatched.port_6 ) / ( 5 * tma_info_core_core_clks )", - "Category": "TMA", - "CountDomain": "Core_Execution", - "Threshold": { - "Formula": "metric_TMA_........ALU_Op_Utilization(%) > 40" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": "" - }, - { - "MetricName": "Port_0", - "LegacyName": "metric_TMA_..........Port_0(%)", - "ParentCategory": "ALU_Op_Utilization", - "Level": 6, - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UOPS_DISPATCHED.PORT_0", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( a / ( b if smt_on else ( c ) ) )", - "BaseFormula": " uops_dispatched.port_0 / tma_info_core_core_clks", - "Category": "TMA", - "CountDomain": "Core_Clocks", - "Threshold": { - "Formula": "metric_TMA_..........Port_0(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Compute", - "LocateWith": " UOPS_DISPATCHED.PORT_0 " - }, - { - "MetricName": "Port_1", - "LegacyName": "metric_TMA_..........Port_1(%)", - "ParentCategory": "ALU_Op_Utilization", - "Level": 6, - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UOPS_DISPATCHED.PORT_1", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( a / ( b if smt_on else ( c ) ) )", - "BaseFormula": " uops_dispatched.port_1 / tma_info_core_core_clks", - "Category": "TMA", - "CountDomain": "Core_Clocks", - "Threshold": { - "Formula": "metric_TMA_..........Port_1(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": " UOPS_DISPATCHED.PORT_1 " - }, - { - "MetricName": "Port_6", - "LegacyName": "metric_TMA_..........Port_6(%)", - "ParentCategory": "ALU_Op_Utilization", - "Level": 6, - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UOPS_DISPATCHED.PORT_6", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( a / ( b if smt_on else ( c ) ) )", - "BaseFormula": " uops_dispatched.port_6 / tma_info_core_core_clks", - "Category": "TMA", - "CountDomain": "Core_Clocks", - "Threshold": { - "Formula": "metric_TMA_..........Port_6(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": " UOPS_DISPATCHED.PORT_1 " - }, - { - "MetricName": "Load_Op_Utilization", - "LegacyName": "metric_TMA_........Load_Op_Utilization(%)", - "ParentCategory": "Ports_Utilized_3m", - "Level": 5, - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UOPS_DISPATCHED.PORT_2_3_10", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( a / ( 3 * ( b if smt_on else ( c ) ) ) )", - "BaseFormula": " uops_dispatched.port_2_3_10 / ( 3 * tma_info_core_core_clks )", - "Category": "TMA", - "CountDomain": "Core_Execution", - "Threshold": { - "Formula": "metric_TMA_........Load_Op_Utilization(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": " UOPS_DISPATCHED.PORT_2_3_10" - }, - { - "MetricName": "Store_Op_Utilization", - "LegacyName": "metric_TMA_........Store_Op_Utilization(%)", - "ParentCategory": "Ports_Utilized_3m", - "Level": 5, - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UOPS_DISPATCHED.PORT_4_9", - "Alias": "a" - }, - { - "Name": "UOPS_DISPATCHED.PORT_7_8", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "c" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "d" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( a + b ) / ( 4 * ( c if smt_on else ( d ) ) ) )", - "BaseFormula": " ( uops_dispatched.port_4_9 + uops_dispatched.port_7_8 ) / ( 4 * tma_info_core_core_clks )", - "Category": "TMA", - "CountDomain": "Core_Execution", - "Threshold": { - "Formula": "metric_TMA_........Store_Op_Utilization(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": " UOPS_DISPATCHED.PORT_7_8" - }, - { - "MetricName": "Retiring", - "LegacyName": "metric_TMA_Retiring(%)", - "Level": 1, - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. ", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b + c + a + d ) )", - "BaseFormula": " perf_metrics.retiring / ( perf_metrics.frontend_bound + perf_metrics.bad_speculation + perf_metrics.retiring + perf_metrics.backend_bound )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvUW;TmaL1", - "LocateWith": " UOPS_RETIRED.SLOTS" - }, - { - "MetricName": "Light_Operations", - "LegacyName": "metric_TMA_..Light_Operations(%)", - "ParentCategory": "Retiring", - "Level": 2, - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .)", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.HEAVY_OPERATIONS", - "Alias": "e" - } - ], - "Constants": [], - "Formula": "100 * ( max( 0 , ( a / ( b + c + a + d ) ) - ( e / ( b + c + a + d ) ) ) )", - "BaseFormula": " max( 0 , tma_retiring - tma_heavy_operations )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Retire;TmaL2", - "LocateWith": "INST_RETIRED.PREC_DIST" - }, - { - "MetricName": "FP_Arith", - "LegacyName": "metric_TMA_....FP_Arith(%)", - "ParentCategory": "Light_Operations", - "Level": 3, - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "UOPS_EXECUTED.X87", - "Alias": "e" - }, - { - "Name": "UOPS_EXECUTED.THREAD", - "Alias": "f" - }, - { - "Name": "FP_ARITH_INST_RETIRED.SCALAR", - "Alias": "g" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.SCALAR", - "Alias": "h" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "i" - }, - { - "Name": "FP_ARITH_INST_RETIRED.VECTOR", - "Alias": "j" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.VECTOR", - "Alias": "k" - } - ], - "Constants": [], - "Formula": "100 * ( ( ( a / ( b + c + a + d ) ) * e / f ) + ( ( g + h ) / ( ( a / ( b + c + a + d ) ) * ( i ) ) ) + ( ( j + k ) / ( ( a / ( b + c + a + d ) ) * ( i ) ) ) )", - "BaseFormula": " tma_x87_use + tma_fp_scalar + tma_fp_vector", - "Category": "TMA", - "CountDomain": "Uops", - "Threshold": { - "Formula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "HPC", - "LocateWith": "" - }, - { - "MetricName": "X87_Use", - "LegacyName": "metric_TMA_......X87_Use(%)", - "ParentCategory": "FP_Arith", - "Level": 4, - "BriefDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "UOPS_EXECUTED.X87", - "Alias": "e" - }, - { - "Name": "UOPS_EXECUTED.THREAD", - "Alias": "f" - } - ], - "Constants": [], - "Formula": "100 * ( ( a / ( b + c + a + d ) ) * e / f )", - "BaseFormula": " tma_retiring * uops_executed.x87 / uops_executed.thread", - "Category": "TMA", - "CountDomain": "Uops", - "Threshold": { - "Formula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Compute", - "LocateWith": "" - }, - { - "MetricName": "FP_Scalar", - "LegacyName": "metric_TMA_......FP_Scalar(%)", - "ParentCategory": "FP_Arith", - "Level": 4, - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "FP_ARITH_INST_RETIRED.SCALAR", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.SCALAR", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "e" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - } - ], - "Constants": [], - "Formula": "100 * ( ( a + b ) / ( ( c / ( d + e + c + f ) ) * ( g ) ) )", - "BaseFormula": " ( fp_arith_inst_retired.scalar + fp_arith_inst_retired2.scalar ) / ( tma_retiring * tma_info_thread_slots )", - "Category": "TMA", - "CountDomain": "Uops", - "Threshold": { - "Formula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Compute;Flops", - "LocateWith": "" - }, - { - "MetricName": "FP_Vector", - "LegacyName": "metric_TMA_......FP_Vector(%)", - "ParentCategory": "FP_Arith", - "Level": 4, - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "FP_ARITH_INST_RETIRED.VECTOR", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.VECTOR", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "e" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - } - ], - "Constants": [], - "Formula": "100 * ( ( a + b ) / ( ( c / ( d + e + c + f ) ) * ( g ) ) )", - "BaseFormula": " ( fp_arith_inst_retired.vector + fp_arith_inst_retired2.vector ) / ( tma_retiring * tma_info_thread_slots )", - "Category": "TMA", - "CountDomain": "Uops", - "Threshold": { - "Formula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Compute;Flops", - "LocateWith": "" - }, - { - "MetricName": "FP_Vector_128b", - "LegacyName": "metric_TMA_........FP_Vector_128b(%)", - "ParentCategory": "FP_Vector", - "Level": 5, - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE", - "Alias": "b" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.128B_PACKED_HALF", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "e" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "f" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "g" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "h" - } - ], - "Constants": [], - "Formula": "100 * ( ( a + b + c ) / ( ( d / ( e + f + d + g ) ) * ( h ) ) )", - "BaseFormula": " ( fp_arith_inst_retired.128b_packed_double + fp_arith_inst_retired.128b_packed_single + fp_arith_inst_retired2.128b_packed_half ) / ( tma_retiring * tma_info_thread_slots )", - "Category": "TMA", - "CountDomain": "Uops", - "Threshold": { - "Formula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Compute;Flops", - "LocateWith": "" - }, - { - "MetricName": "FP_Vector_256b", - "LegacyName": "metric_TMA_........FP_Vector_256b(%)", - "ParentCategory": "FP_Vector", - "Level": 5, - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE", - "Alias": "b" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.256B_PACKED_HALF", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "e" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "f" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "g" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "h" - } - ], - "Constants": [], - "Formula": "100 * ( ( a + b + c ) / ( ( d / ( e + f + d + g ) ) * ( h ) ) )", - "BaseFormula": " ( fp_arith_inst_retired.256b_packed_double + fp_arith_inst_retired.256b_packed_single + fp_arith_inst_retired2.256b_packed_half ) / ( tma_retiring * tma_info_thread_slots )", - "Category": "TMA", - "CountDomain": "Uops", - "Threshold": { - "Formula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Compute;Flops", - "LocateWith": "" - }, - { - "MetricName": "FP_Vector_512b", - "LegacyName": "metric_TMA_........FP_Vector_512b(%)", - "ParentCategory": "FP_Vector", - "Level": 5, - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE", - "Alias": "b" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.512B_PACKED_HALF", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "e" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "f" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "g" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "h" - } - ], - "Constants": [], - "Formula": "100 * ( ( a + b + c ) / ( ( d / ( e + f + d + g ) ) * ( h ) ) )", - "BaseFormula": " ( fp_arith_inst_retired.512b_packed_double + fp_arith_inst_retired.512b_packed_single + fp_arith_inst_retired2.512b_packed_half ) / ( tma_retiring * tma_info_thread_slots )", - "Category": "TMA", - "CountDomain": "Uops", - "Threshold": { - "Formula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Compute;Flops", - "LocateWith": "" - }, - { - "MetricName": "Int_Operations", - "LegacyName": "metric_TMA_....Int_Operations(%)", - "ParentCategory": "Light_Operations", - "Level": 3, - "BriefDescription": "This metric represents overall Integer (Int) select operations fraction the CPU has executed (retired). Vector/Matrix Int operations and shuffles are counted. Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "INT_VEC_RETIRED.ADD_128", - "Alias": "a" - }, - { - "Name": "INT_VEC_RETIRED.VNNI_128", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "e" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - }, - { - "Name": "INT_VEC_RETIRED.ADD_256", - "Alias": "h" - }, - { - "Name": "INT_VEC_RETIRED.MUL_256", - "Alias": "i" - }, - { - "Name": "INT_VEC_RETIRED.VNNI_256", - "Alias": "j" - } - ], - "Constants": [], - "Formula": "100 * ( ( ( a + b ) / ( ( c / ( d + e + c + f ) ) * ( g ) ) ) + ( ( h + i + j ) / ( ( c / ( d + e + c + f ) ) * ( g ) ) ) )", - "BaseFormula": " tma_int_vector_128b + tma_int_vector_256b", - "Category": "TMA", - "CountDomain": "Uops", - "Threshold": { - "Formula": "metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Pipeline", - "LocateWith": "" - }, - { - "MetricName": "Int_Vector_128b", - "LegacyName": "metric_TMA_......Int_Vector_128b(%)", - "ParentCategory": "Int_Operations", - "Level": 4, - "BriefDescription": "This metric represents 128-bit vector Integer ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "INT_VEC_RETIRED.ADD_128", - "Alias": "a" - }, - { - "Name": "INT_VEC_RETIRED.VNNI_128", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "e" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - } - ], - "Constants": [], - "Formula": "100 * ( ( a + b ) / ( ( c / ( d + e + c + f ) ) * ( g ) ) )", - "BaseFormula": " ( int_vec_retired.add_128 + int_vec_retired.vnni_128 ) / ( tma_retiring * tma_info_thread_slots )", - "Category": "TMA", - "CountDomain": "Uops", - "Threshold": { - "Formula": "metric_TMA_......Int_Vector_128b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Compute;IntVector;Pipeline", - "LocateWith": "" - }, - { - "MetricName": "Int_Vector_256b", - "LegacyName": "metric_TMA_......Int_Vector_256b(%)", - "ParentCategory": "Int_Operations", - "Level": 4, - "BriefDescription": "This metric represents 256-bit vector Integer ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fraction the CPU has retired.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "INT_VEC_RETIRED.ADD_256", - "Alias": "a" - }, - { - "Name": "INT_VEC_RETIRED.MUL_256", - "Alias": "b" - }, - { - "Name": "INT_VEC_RETIRED.VNNI_256", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "e" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "f" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "g" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "h" - } - ], - "Constants": [], - "Formula": "100 * ( ( a + b + c ) / ( ( d / ( e + f + d + g ) ) * ( h ) ) )", - "BaseFormula": " ( int_vec_retired.add_256 + int_vec_retired.mul_256 + int_vec_retired.vnni_256 ) / ( tma_retiring * tma_info_thread_slots )", - "Category": "TMA", - "CountDomain": "Uops", - "Threshold": { - "Formula": "metric_TMA_......Int_Vector_256b(%) > 10 & metric_TMA_....Int_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Compute;IntVector;Pipeline", - "LocateWith": "" - }, - { - "MetricName": "Memory_Operations", - "LegacyName": "metric_TMA_....Memory_Operations(%)", - "ParentCategory": "Light_Operations", - "Level": 3, - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.HEAVY_OPERATIONS", - "Alias": "e" - }, - { - "Name": "MEM_UOP_RETIRED.ANY", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - } - ], - "Constants": [], - "Formula": "100 * ( ( max( 0 , ( a / ( b + c + a + d ) ) - ( e / ( b + c + a + d ) ) ) ) * f / ( ( a / ( b + c + a + d ) ) * ( g ) ) )", - "BaseFormula": " tma_light_operations * mem_uop_retired.any / ( tma_retiring * tma_info_thread_slots )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Pipeline", - "LocateWith": "" - }, - { - "MetricName": "Fused_Instructions", - "LegacyName": "metric_TMA_....Fused_Instructions(%)", - "ParentCategory": "Light_Operations", - "Level": 3, - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring fused instructions , where one uop can represent multiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Other_Light_Ops in MTL!)}", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.HEAVY_OPERATIONS", - "Alias": "e" - }, - { - "Name": "INST_RETIRED.MACRO_FUSED", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - } - ], - "Constants": [], - "Formula": "100 * ( ( max( 0 , ( a / ( b + c + a + d ) ) - ( e / ( b + c + a + d ) ) ) ) * f / ( ( a / ( b + c + a + d ) ) * ( g ) ) )", - "BaseFormula": " tma_light_operations * inst_retired.macro_fused / ( tma_retiring * tma_info_thread_slots )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_....Fused_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Branches;BvBO;Pipeline", - "LocateWith": "" - }, - { - "MetricName": "Non_Fused_Branches", - "LegacyName": "metric_TMA_....Non_Fused_Branches(%)", - "ParentCategory": "Light_Operations", - "Level": 3, - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions that were not fused. Non-conditional branches like direct JMP or CALL would count here. Can be used to examine fusible conditional jumps that were not fused.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.HEAVY_OPERATIONS", - "Alias": "e" - }, - { - "Name": "BR_INST_RETIRED.ALL_BRANCHES", - "Alias": "f" - }, - { - "Name": "INST_RETIRED.MACRO_FUSED", - "Alias": "g" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "h" - } - ], - "Constants": [], - "Formula": "100 * ( ( max( 0 , ( a / ( b + c + a + d ) ) - ( e / ( b + c + a + d ) ) ) ) * ( f - g ) / ( ( a / ( b + c + a + d ) ) * ( h ) ) )", - "BaseFormula": " tma_light_operations * ( br_inst_retired.all_branches - inst_retired.macro_fused ) / ( tma_retiring * tma_info_thread_slots )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_....Non_Fused_Branches(%) > 10 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Branches;BvBO;Pipeline", - "LocateWith": "" - }, - { - "MetricName": "Other_Light_Ops", - "LegacyName": "metric_TMA_....Other_Light_Ops(%)", - "ParentCategory": "Light_Operations", - "Level": 3, - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.HEAVY_OPERATIONS", - "Alias": "e" - }, - { - "Name": "UOPS_EXECUTED.X87", - "Alias": "f" - }, - { - "Name": "UOPS_EXECUTED.THREAD", - "Alias": "g" - }, - { - "Name": "FP_ARITH_INST_RETIRED.SCALAR", - "Alias": "h" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.SCALAR", - "Alias": "i" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "j" - }, - { - "Name": "FP_ARITH_INST_RETIRED.VECTOR", - "Alias": "k" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.VECTOR", - "Alias": "l" - }, - { - "Name": "INT_VEC_RETIRED.ADD_128", - "Alias": "m" - }, - { - "Name": "INT_VEC_RETIRED.VNNI_128", - "Alias": "n" - }, - { - "Name": "INT_VEC_RETIRED.ADD_256", - "Alias": "o" - }, - { - "Name": "INT_VEC_RETIRED.MUL_256", - "Alias": "p" - }, - { - "Name": "INT_VEC_RETIRED.VNNI_256", - "Alias": "q" - }, - { - "Name": "MEM_UOP_RETIRED.ANY", - "Alias": "r" - }, - { - "Name": "INST_RETIRED.MACRO_FUSED", - "Alias": "s" - }, - { - "Name": "BR_INST_RETIRED.ALL_BRANCHES", - "Alias": "t" - } - ], - "Constants": [], - "Formula": "100 * ( max( 0 , ( max( 0 , ( a / ( b + c + a + d ) ) - ( e / ( b + c + a + d ) ) ) ) - ( ( ( ( a / ( b + c + a + d ) ) * f / g ) + ( ( h + i ) / ( ( a / ( b + c + a + d ) ) * ( j ) ) ) + ( ( k + l ) / ( ( a / ( b + c + a + d ) ) * ( j ) ) ) ) + ( ( ( m + n ) / ( ( a / ( b + c + a + d ) ) * ( j ) ) ) + ( ( o + p + q ) / ( ( a / ( b + c + a + d ) ) * ( j ) ) ) ) + ( ( max( 0 , ( a / ( b + c + a + d ) ) - ( e / ( b + c + a + d ) ) ) ) * r / ( ( a / ( b + c + a + d ) ) * ( j ) ) ) + ( ( max( 0 , ( a / ( b + c + a + d ) ) - ( e / ( b + c + a + d ) ) ) ) * s / ( ( a / ( b + c + a + d ) ) * ( j ) ) ) + ( ( max( 0 , ( a / ( b + c + a + d ) ) - ( e / ( b + c + a + d ) ) ) ) * ( t - s ) / ( ( a / ( b + c + a + d ) ) * ( j ) ) ) ) ) )", - "BaseFormula": " max( 0 , tma_light_operations - ( tma_fp_arith + tma_int_operations + tma_memory_operations + tma_fused_instructions + tma_non_fused_branches ) )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Pipeline", - "LocateWith": "" - }, - { - "MetricName": "Nop_Instructions", - "LegacyName": "metric_TMA_......Nop_Instructions(%)", - "ParentCategory": "Light_Operations", - "Level": 4, - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.HEAVY_OPERATIONS", - "Alias": "e" - }, - { - "Name": "INST_RETIRED.NOP", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - } - ], - "Constants": [], - "Formula": "100 * ( ( max( 0 , ( a / ( b + c + a + d ) ) - ( e / ( b + c + a + d ) ) ) ) * f / ( ( a / ( b + c + a + d ) ) * ( g ) ) )", - "BaseFormula": " tma_light_operations * inst_retired.nop / ( tma_retiring * tma_info_thread_slots )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvBO;Pipeline", - "LocateWith": " INST_RETIRED.NOP" - }, - { - "MetricName": "Shuffles_256b", - "LegacyName": "metric_TMA_......Shuffles_256b(%)", - "ParentCategory": "Int_Operations", - "Level": 4, - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring Shuffle operations of 256-bit vector size (FP or Integer). Shuffles may incur slow cross \"vector lane\" data transfers.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.HEAVY_OPERATIONS", - "Alias": "e" - }, - { - "Name": "INT_VEC_RETIRED.SHUFFLES", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - } - ], - "Constants": [], - "Formula": "100 * ( ( max( 0 , ( a / ( b + c + a + d ) ) - ( e / ( b + c + a + d ) ) ) ) * f / ( ( a / ( b + c + a + d ) ) * ( g ) ) )", - "BaseFormula": " tma_light_operations * int_vec_retired.shuffles / ( tma_retiring * tma_info_thread_slots )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_......Shuffles_256b(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "HPC;Pipeline", - "LocateWith": "" - }, - { - "MetricName": "Heavy_Operations", - "LegacyName": "metric_TMA_..Heavy_Operations(%)", - "ParentCategory": "Retiring", - "Level": 2, - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+])", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.HEAVY_OPERATIONS", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b + c + d + e ) )", - "BaseFormula": " perf_metrics.heavy_operations / ( perf_metrics.frontend_bound + perf_metrics.bad_speculation + perf_metrics.retiring + perf_metrics.backend_bound )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_..Heavy_Operations(%) > 10" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Retire;TmaL2", - "LocateWith": " UOPS_RETIRED.HEAVY" - }, - { - "MetricName": "Few_Uops_Instructions", - "LegacyName": "metric_TMA_....Few_Uops_Instructions(%)", - "ParentCategory": "Heavy_Operations", - "Level": 3, - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "PERF_METRICS.HEAVY_OPERATIONS", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - }, - { - "Name": "UOPS_RETIRED.MS", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - } - ], - "Constants": [], - "Formula": "100 * ( max( 0 , ( a / ( b + c + d + e ) ) - ( f / ( g ) ) ) )", - "BaseFormula": " max( 0 , tma_heavy_operations - tma_microcode_sequencer )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": "" - }, - { - "MetricName": "Microcode_Sequencer", - "LegacyName": "metric_TMA_....Microcode_Sequencer(%)", - "ParentCategory": "Heavy_Operations", - "Level": 3, - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UOPS_RETIRED.MS", - "Alias": "a" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( a / ( b ) )", - "BaseFormula": " uops_retired.ms / tma_info_thread_slots", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MicroSeq", - "LocateWith": " UOPS_RETIRED.MS" - }, - { - "MetricName": "Assists", - "LegacyName": "metric_TMA_......Assists(%)", - "ParentCategory": "Microcode_Sequencer", - "Level": 4, - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ASSISTS.ANY", - "Alias": "a" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( ( ( 99 * 3 + 63 + 30 ) / 5 ) * a / ( b ) )", - "BaseFormula": " ( ( 99 *3 + 63 + 30 ) / 5 ) * assists.any / tma_info_thread_slots", - "Category": "TMA", - "CountDomain": "Slots_Estimated", - "Threshold": { - "Formula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BvIO", - "LocateWith": " ASSISTS.ANY" - }, - { - "MetricName": "Page_Faults", - "LegacyName": "metric_TMA_........Page_Faults(%)", - "ParentCategory": "Assists", - "Level": 5, - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Page Faults. A Page Fault may apply on first application access to a memory page. Note operating system handling of page faults accounts for the majority of its cost.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ASSISTS.PAGE_FAULT", - "Alias": "a" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( 99 * a / ( b ) )", - "BaseFormula": " 99 * assists.page_fault / tma_info_thread_slots", - "Category": "TMA", - "CountDomain": "Slots_Estimated", - "Threshold": { - "Formula": "metric_TMA_........Page_Faults(%) > 5" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": "" - }, - { - "MetricName": "FP_Assists", - "LegacyName": "metric_TMA_........FP_Assists(%)", - "ParentCategory": "Assists", - "Level": 5, - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals).", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ASSISTS.FP", - "Alias": "a" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( 30 * a / ( b ) )", - "BaseFormula": " 30 * assists.fp / tma_info_thread_slots", - "Category": "TMA", - "CountDomain": "Slots_Estimated", - "Threshold": { - "Formula": "metric_TMA_........FP_Assists(%) > 10" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "HPC", - "LocateWith": "" - }, - { - "MetricName": "AVX_Assists", - "LegacyName": "metric_TMA_........AVX_Assists(%)", - "ParentCategory": "Assists", - "Level": 5, - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists. ", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "ASSISTS.SSE_AVX_MIX", - "Alias": "a" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "100 * ( 63 * a / ( b ) )", - "BaseFormula": " 63 * assists.sse_avx_mix / tma_info_thread_slots", - "Category": "TMA", - "CountDomain": "Slots_Estimated", - "Threshold": { - "Formula": "metric_TMA_........AVX_Assists(%) > 10" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "HPC", - "LocateWith": "" - }, - { - "MetricName": "CISC", - "LegacyName": "metric_TMA_......CISC(%)", - "ParentCategory": "Microcode_Sequencer", - "Level": 4, - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources.", - "UnitOfMeasure": "percent", - "Events": [ - { - "Name": "UOPS_RETIRED.MS", - "Alias": "a" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "b" - }, - { - "Name": "ASSISTS.ANY", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "100 * ( max( 0 , ( a / ( b ) ) - ( ( ( 99 * 3 + 63 + 30 ) / 5 ) * c / ( b ) ) ) )", - "BaseFormula": " max( 0 , tma_microcode_sequencer - tma_assists )", - "Category": "TMA", - "CountDomain": "Slots", - "Threshold": { - "Formula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "", - "LocateWith": " FRONTEND_RETIRED.MS_FLOWS " - }, - { - "MetricName": "Info_Botlnk_L0_Core_Bound_Likely", - "LegacyName": "metric_TMA_Info_Botlnk_L0_Core_Bound_Likely", - "Level": 1, - "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.MEMORY_BOUND", - "Alias": "e" - }, - { - "Name": "EXE_ACTIVITY.EXE_BOUND_0_PORTS", - "Alias": "f" - }, - { - "Name": "RS.EMPTY_RESOURCE", - "Alias": "g" - }, - { - "Name": "RESOURCE_STALLS.SCOREBOARD", - "Alias": "h" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "i" - }, - { - "Name": "CYCLE_ACTIVITY.STALLS_TOTAL", - "Alias": "j" - }, - { - "Name": "EXE_ACTIVITY.BOUND_ON_LOADS", - "Alias": "k" - }, - { - "Name": "EXE_ACTIVITY.1_PORTS_UTIL", - "Alias": "l" - }, - { - "Name": "EXE_ACTIVITY.2_3_PORTS_UTIL", - "Alias": "m" - }, - { - "Name": "ARITH.DIV_ACTIVE", - "Alias": "n" - }, - { - "Name": "CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE", - "Alias": "o" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_DISTRIBUTED", - "Alias": "p" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( 1 - ( max( 0 , ( a / ( b + c + d + a ) ) - ( e / ( b + c + d + a ) ) ) ) / ( ( ( ( f + max( g - h , 0 ) ) / ( i ) * ( j - k ) / ( i ) ) * ( i ) + ( l + ( d / ( b + c + d + a ) ) * m ) ) / ( i ) if ( n < ( j - k ) ) else ( l + ( d / ( b + c + d + a ) ) * m ) / ( i ) ) if ( max( 0 , ( a / ( b + c + d + a ) ) - ( e / ( b + c + d + a ) ) ) ) < ( ( ( ( f + max( g - h , 0 ) ) / ( i ) * ( j - k ) / ( i ) ) * ( i ) + ( l + ( d / ( b + c + d + a ) ) * m ) ) / ( i ) if ( n < ( j - k ) ) else ( l + ( d / ( b + c + d + a ) ) * m ) / ( i ) ) else 1 ) if ( 1 - o / p if smt_on else 0 ) > 0.5 else 0", - "BaseFormula": " 100 * ( 1 - tma_core_bound / tma_ports_utilization if tma_core_bound < tma_ports_utilization else 1 ) if tma_info_system_smt_2t_utilization > 0.5 else 0", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Botlnk_Core_Bound_Likely" - } - ], - "Formula": "a > 0.5", - "BaseFormula": "metric_TMA_Info_Botlnk_Core_Bound_Likely > 0.5", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Cor;SMT", - "LocateWith": "" - }, - { - "MetricName": "Info_Thread_IPC", - "LegacyName": "metric_TMA_Info_Thread_IPC", - "Level": 1, - "BriefDescription": "Instructions Per Cycle (per Logical Processor)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / ( b )", - "BaseFormula": " inst_retired.any / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD", - "MetricGroup": "Ret;Summary", - "LocateWith": "" - }, - { - "MetricName": "Info_Thread_UopPI", - "LegacyName": "metric_TMA_Info_Thread_UopPI", - "Level": 1, - "BriefDescription": "Uops Per Instruction", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "e" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "f" - } - ], - "Constants": [], - "Formula": "( ( a / ( b + c + a + d ) ) * ( e ) ) / f", - "BaseFormula": " ( tma_retiring * tma_info_thread_slots ) / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Thread_UopPI" - } - ], - "Formula": "a > 1.05", - "BaseFormula": "metric_TMA_Info_Thread_UopPI > 1.05", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD", - "MetricGroup": "Pipeline;Ret;Retire", - "LocateWith": "" - }, - { - "MetricName": "Info_Thread_UpTB", - "LegacyName": "metric_TMA_Info_Thread_UpTB", - "Level": 1, - "BriefDescription": "Uops per taken branch", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "e" - }, - { - "Name": "BR_INST_RETIRED.NEAR_TAKEN", - "Alias": "f" - } - ], - "Constants": [], - "Formula": "( ( a / ( b + c + a + d ) ) * ( e ) ) / f", - "BaseFormula": " ( tma_retiring * tma_info_thread_slots ) / br_inst_retired.near_taken", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Thread_UpTB" - } - ], - "Formula": "a < 6 * 1.5", - "BaseFormula": "metric_TMA_Info_Thread_UpTB < 6 * 1.5", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD", - "MetricGroup": "Branches;Fed;FetchBW", - "LocateWith": "" - }, - { - "MetricName": "Info_Thread_CPI", - "LegacyName": "metric_TMA_Info_Thread_CPI", - "Level": 1, - "BriefDescription": "Cycles Per Instruction (per Logical Processor)", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1 / ( a / ( b ) )", - "BaseFormula": " 1 / tma_info_thread_ipc", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD", - "MetricGroup": "Pipeline;Mem", - "LocateWith": "" - }, - { - "MetricName": "Info_Thread_CLKS", - "LegacyName": "metric_TMA_Info_Thread_CLKS", - "Level": 1, - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "a", - "BaseFormula": " cpu_clk_unhalted.thread", - "Category": "TMA", - "CountDomain": "Count", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD", - "MetricGroup": "Pipeline", - "LocateWith": "" - }, - { - "MetricName": "Info_Thread_SLOTS", - "LegacyName": "metric_TMA_Info_Thread_SLOTS", - "Level": 1, - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "a", - "BaseFormula": " topdown.slots:perf_metrics", - "Category": "TMA", - "CountDomain": "Count", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD", - "MetricGroup": "TmaL1", - "LocateWith": "" - }, - { - "MetricName": "Info_Thread_Slots_Utilization", - "LegacyName": "metric_TMA_Info_Thread_Slots_Utilization", - "Level": 1, - "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "a" - }, - { - "Name": "TOPDOWN.SLOTS:percore", - "Alias": "b" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "( a ) / ( b / 2 ) if smt_on else 1", - "BaseFormula": " tma_info_thread_slots / ( topdown.slots:percore / 2 ) if smt_on else 1", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD", - "MetricGroup": "SMT;TmaL1", - "LocateWith": "" - }, - { - "MetricName": "Info_Thread_Execute_per_Issue", - "LegacyName": "metric_TMA_Info_Thread_Execute_per_Issue", - "Level": 1, - "BriefDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UOPS_EXECUTED.THREAD", - "Alias": "a" - }, - { - "Name": "UOPS_ISSUED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " uops_executed.thread / uops_issued.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD", - "MetricGroup": "Cor;Pipeline", - "LocateWith": "" - }, - { - "MetricName": "Info_Core_CoreIPC", - "LegacyName": "metric_TMA_Info_Core_CoreIPC", - "Level": 1, - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "b" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "a / ( b if smt_on else ( c ) )", - "BaseFormula": " inst_retired.any / tma_info_core_core_clks", - "Category": "TMA", - "CountDomain": "Core_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE", - "MetricGroup": "Ret;SMT;TmaL1", - "LocateWith": "" - }, - { - "MetricName": "Info_Core_FLOPc", - "LegacyName": "metric_TMA_Info_Core_FLOPc", - "Level": 1, - "BriefDescription": "Floating Point Operations Per Cycle", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "FP_ARITH_INST_RETIRED.SCALAR", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.SCALAR_HALF", - "Alias": "b" - }, - { - "Name": "FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE", - "Alias": "c" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.COMPLEX_SCALAR_HALF", - "Alias": "d" - }, - { - "Name": "FP_ARITH_INST_RETIRED.4_FLOPS", - "Alias": "e" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.128B_PACKED_HALF", - "Alias": "f" - }, - { - "Name": "FP_ARITH_INST_RETIRED.8_FLOPS", - "Alias": "g" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.256B_PACKED_HALF", - "Alias": "h" - }, - { - "Name": "FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE", - "Alias": "i" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.512B_PACKED_HALF", - "Alias": "j" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "k" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "l" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "( ( 1 * ( a + b ) + 2 * ( c + d ) + 4 * e + 8 * ( f + g ) + 16 * ( h + i ) + 32 * j ) ) / ( k if smt_on else ( l ) )", - "BaseFormula": " ( ( 1 * ( fp_arith_inst_retired.scalar + fp_arith_inst_retired2.scalar_half ) + 2 * ( fp_arith_inst_retired.128b_packed_double + fp_arith_inst_retired2.complex_scalar_half ) + 4 * fp_arith_inst_retired.4_flops + 8 * ( fp_arith_inst_retired2.128b_packed_half + fp_arith_inst_retired.8_flops ) + 16 * ( fp_arith_inst_retired2.256b_packed_half + fp_arith_inst_retired.512b_packed_single ) + 32 * fp_arith_inst_retired2.512b_packed_half ) ) / tma_info_core_core_clks", - "Category": "TMA", - "CountDomain": "Core_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE", - "MetricGroup": "Ret;Flops", - "LocateWith": "" - }, - { - "MetricName": "Info_Core_FP_Arith_Utilization", - "LegacyName": "metric_TMA_Info_Core_FP_Arith_Utilization", - "Level": 1, - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common).", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "FP_ARITH_DISPATCHED.PORT_0", - "Alias": "a" - }, - { - "Name": "FP_ARITH_DISPATCHED.PORT_1", - "Alias": "b" - }, - { - "Name": "FP_ARITH_DISPATCHED.PORT_5", - "Alias": "c" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "d" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "e" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "( a + b + c ) / ( 2 * ( d if smt_on else ( e ) ) )", - "BaseFormula": " ( fp_arith_dispatched.port_0 + fp_arith_dispatched.port_1 + fp_arith_dispatched.port_5 ) / ( 2 * tma_info_core_core_clks )", - "Category": "TMA", - "CountDomain": "Core_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE", - "MetricGroup": "Cor;Flops;HPC", - "LocateWith": "" - }, - { - "MetricName": "Info_Core_ILP", - "LegacyName": "metric_TMA_Info_Core_ILP", - "Level": 1, - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UOPS_EXECUTED.THREAD", - "Alias": "a" - }, - { - "Name": "UOPS_EXECUTED.THREAD:c1", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " uops_executed.thread / uops_executed.thread:c1", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE", - "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", - "LocateWith": "" - }, - { - "MetricName": "Info_Core_EPC", - "LegacyName": "metric_TMA_Info_Core_EPC", - "Level": 1, - "BriefDescription": "uops Executed per Cycle", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UOPS_EXECUTED.THREAD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / ( b )", - "BaseFormula": " uops_executed.thread / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE", - "MetricGroup": "Power", - "LocateWith": "" - }, - { - "MetricName": "Info_Core_CORE_CLKS", - "LegacyName": "metric_TMA_Info_Core_CORE_CLKS", - "Level": 1, - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "a if smt_on else ( b )", - "BaseFormula": " cpu_clk_unhalted.distributed if smt_on else tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Count", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE", - "MetricGroup": "SMT", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpLoad", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpLoad", - "Level": 1, - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "MEM_INST_RETIRED.ALL_LOADS", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / mem_inst_retired.all_loads", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Inst_Mix_IpLoad" - } - ], - "Formula": "a < 3", - "BaseFormula": "metric_TMA_Info_Inst_Mix_IpLoad < 3", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "InsType", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpStore", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpStore", - "Level": 1, - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "MEM_INST_RETIRED.ALL_STORES", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / mem_inst_retired.all_stores", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Inst_Mix_IpStore" - } - ], - "Formula": "a < 8", - "BaseFormula": "metric_TMA_Info_Inst_Mix_IpStore < 8", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "InsType", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpBranch", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpBranch", - "Level": 1, - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "BR_INST_RETIRED.ALL_BRANCHES", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / br_inst_retired.all_branches", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Inst_Mix_IpBranch" - } - ], - "Formula": "a < 8", - "BaseFormula": "metric_TMA_Info_Inst_Mix_IpBranch < 8", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Branches;Fed;InsType", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpCall", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpCall", - "Level": 1, - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "BR_INST_RETIRED.NEAR_CALL", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / br_inst_retired.near_call", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Inst_Mix_IpCall" - } - ], - "Formula": "a < 200", - "BaseFormula": "metric_TMA_Info_Inst_Mix_IpCall < 200", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Branches;Fed;PGO", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpTB", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpTB", - "Level": 1, - "BriefDescription": "Instructions per taken branch", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "BR_INST_RETIRED.NEAR_TAKEN", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / br_inst_retired.near_taken", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Inst_Mix_IpTB" - } - ], - "Formula": "a < 6 * 2 + 1", - "BaseFormula": "metric_TMA_Info_Inst_Mix_IpTB < 6 * 2 + 1", - "ThresholdIssues": "$issueFB" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_BpTkBranch", - "LegacyName": "metric_TMA_Info_Inst_Mix_BpTkBranch", - "Level": 1, - "BriefDescription": "Branch instructions per taken branch. ", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "BR_INST_RETIRED.ALL_BRANCHES", - "Alias": "a" - }, - { - "Name": "BR_INST_RETIRED.NEAR_TAKEN", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " br_inst_retired.all_branches / br_inst_retired.near_taken", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Branches;Fed;PGO", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpFLOP", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpFLOP", - "Level": 1, - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED.SCALAR", - "Alias": "b" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.SCALAR_HALF", - "Alias": "c" - }, - { - "Name": "FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE", - "Alias": "d" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.COMPLEX_SCALAR_HALF", - "Alias": "e" - }, - { - "Name": "FP_ARITH_INST_RETIRED.4_FLOPS", - "Alias": "f" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.128B_PACKED_HALF", - "Alias": "g" - }, - { - "Name": "FP_ARITH_INST_RETIRED.8_FLOPS", - "Alias": "h" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.256B_PACKED_HALF", - "Alias": "i" - }, - { - "Name": "FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE", - "Alias": "j" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.512B_PACKED_HALF", - "Alias": "k" - } - ], - "Constants": [], - "Formula": "a / ( ( 1 * ( b + c ) + 2 * ( d + e ) + 4 * f + 8 * ( g + h ) + 16 * ( i + j ) + 32 * k ) )", - "BaseFormula": " inst_retired.any / ( ( 1 * ( fp_arith_inst_retired.scalar + fp_arith_inst_retired2.scalar_half ) + 2 * ( fp_arith_inst_retired.128b_packed_double + fp_arith_inst_retired2.complex_scalar_half ) + 4 * fp_arith_inst_retired.4_flops + 8 * ( fp_arith_inst_retired2.128b_packed_half + fp_arith_inst_retired.8_flops ) + 16 * ( fp_arith_inst_retired2.256b_packed_half + fp_arith_inst_retired.512b_packed_single ) + 32 * fp_arith_inst_retired2.512b_packed_half ) )", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Inst_Mix_IpFLOP" - } - ], - "Formula": "a < 10", - "BaseFormula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Flops;InsType", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpArith", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpArith", - "Level": 1, - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED.SCALAR", - "Alias": "b" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.SCALAR", - "Alias": "c" - }, - { - "Name": "FP_ARITH_INST_RETIRED.VECTOR", - "Alias": "d" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.VECTOR", - "Alias": "e" - } - ], - "Constants": [], - "Formula": "a / ( ( b + c ) + ( d + e ) )", - "BaseFormula": " inst_retired.any / ( ( fp_arith_inst_retired.scalar + fp_arith_inst_retired2.scalar ) + ( fp_arith_inst_retired.vector + fp_arith_inst_retired2.vector ) )", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Inst_Mix_IpArith" - } - ], - "Formula": "a < 10", - "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith < 10", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Flops;InsType", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpArith_Scalar_HP", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_HP", - "Level": 1, - "BriefDescription": "Instructions per FP Arithmetic Scalar Half-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.SCALAR", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / fp_arith_inst_retired2.scalar", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_HP" - } - ], - "Formula": "a < 10", - "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_HP < 10", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Flops;FpScalar;InsType;Server", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpArith_Scalar_SP", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP", - "Level": 1, - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED.SCALAR_SINGLE", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / fp_arith_inst_retired.scalar_single", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP" - } - ], - "Formula": "a < 10", - "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Flops;FpScalar;InsType", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpArith_Scalar_DP", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP", - "Level": 1, - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / fp_arith_inst_retired.scalar_double", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP" - } - ], - "Formula": "a < 10", - "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Flops;FpScalar;InsType", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpArith_AVX128", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpArith_AVX128", - "Level": 1, - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE", - "Alias": "b" - }, - { - "Name": "FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE", - "Alias": "c" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.128B_PACKED_HALF", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "a / ( b + c + d )", - "BaseFormula": " inst_retired.any / ( fp_arith_inst_retired.128b_packed_double + fp_arith_inst_retired.128b_packed_single + fp_arith_inst_retired2.128b_packed_half )", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX128" - } - ], - "Formula": "a < 10", - "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Flops;FpVector;InsType", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpArith_AVX256", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpArith_AVX256", - "Level": 1, - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE", - "Alias": "b" - }, - { - "Name": "FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE", - "Alias": "c" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.256B_PACKED_HALF", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "a / ( b + c + d )", - "BaseFormula": " inst_retired.any / ( fp_arith_inst_retired.256b_packed_double + fp_arith_inst_retired.256b_packed_single + fp_arith_inst_retired2.256b_packed_half )", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX256" - } - ], - "Formula": "a < 10", - "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Flops;FpVector;InsType", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpArith_AVX512", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpArith_AVX512", - "Level": 1, - "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE", - "Alias": "b" - }, - { - "Name": "FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE", - "Alias": "c" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.512B_PACKED_HALF", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "a / ( b + c + d )", - "BaseFormula": " inst_retired.any / ( fp_arith_inst_retired.512b_packed_double + fp_arith_inst_retired.512b_packed_single + fp_arith_inst_retired2.512b_packed_half )", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX512" - } - ], - "Formula": "a < 10", - "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX512 < 10", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Flops;FpVector;InsType", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpPause", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpPause", - "Level": 1, - "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.PAUSE_INST", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "( a ) / b", - "BaseFormula": " tma_info_inst_mix_instructions / cpu_clk_unhalted.pause_inst", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Flops;FpVector;InsType", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_IpSWPF", - "LegacyName": "metric_TMA_Info_Inst_Mix_IpSWPF", - "Level": 1, - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "SW_PREFETCH_ACCESS.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / sw_prefetch_access.any", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Inst_Mix_IpSWPF" - } - ], - "Formula": "a < 100", - "BaseFormula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Prefetches", - "LocateWith": "" - }, - { - "MetricName": "Info_Inst_Mix_Instructions", - "LegacyName": "metric_TMA_Info_Inst_Mix_Instructions", - "Level": 1, - "BriefDescription": "Total number of retired Instructions", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "a", - "BaseFormula": " inst_retired.any", - "Category": "TMA", - "CountDomain": "Count", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Summary;TmaL1", - "LocateWith": "INST_RETIRED.PREC_DIST" - }, - { - "MetricName": "Info_Pipeline_Retire", - "LegacyName": "metric_TMA_Info_Pipeline_Retire", - "Level": 1, - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "e" - }, - { - "Name": "UOPS_RETIRED.SLOTS:c1", - "Alias": "f" - } - ], - "Constants": [], - "Formula": "( ( a / ( b + c + a + d ) ) * ( e ) ) / f", - "BaseFormula": " ( tma_retiring * tma_info_thread_slots ) / uops_retired.slots:c1", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Pipeline;Ret", - "LocateWith": "" - }, - { - "MetricName": "Info_Pipeline_Strings_Cycles", - "LegacyName": "metric_TMA_Info_Pipeline_Strings_Cycles", - "Level": 1, - "BriefDescription": "Estimated fraction of retirement-cycles dealing with repeat instructions", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.REP_ITERATION", - "Alias": "a" - }, - { - "Name": "UOPS_RETIRED.SLOTS:c1", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.rep_iteration / uops_retired.slots:c1", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Pipeline_Strings_Cycles" - } - ], - "Formula": "a > 0.1", - "BaseFormula": "metric_TMA_Info_Pipeline_Strings_Cycles > 0.1", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MicroSeq;Pipeline;Ret", - "LocateWith": "" - }, - { - "MetricName": "Info_Pipeline_IpAssist", - "LegacyName": "metric_TMA_Info_Pipeline_IpAssist", - "Level": 1, - "BriefDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "ASSISTS.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / assists.any", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Pipeline_IpAssist" - } - ], - "Formula": "a < 100000", - "BaseFormula": "metric_TMA_Info_Pipeline_IpAssist < 100000", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", - "LocateWith": "" - }, - { - "MetricName": "Info_Pipeline_Execute", - "LegacyName": "metric_TMA_Info_Pipeline_Execute", - "Level": 1, - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per core", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UOPS_EXECUTED.THREAD", - "Alias": "a" - }, - { - "Name": "UOPS_EXECUTED.CORE_CYCLES_GE_1", - "Alias": "b" - }, - { - "Name": "UOPS_EXECUTED.THREAD:c1", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "a / ( ( b / 2 ) if smt_on else c )", - "BaseFormula": " uops_executed.thread / ( ( uops_executed.core_cycles_ge_1 / 2 ) if smt_on else uops_executed.thread:c1 )", - "Category": "TMA", - "CountDomain": "Metric", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", - "LocateWith": "" - }, - { - "MetricName": "Info_Pipeline_Fetch_DSB", - "LegacyName": "metric_TMA_Info_Pipeline_Fetch_DSB", - "Level": 1, - "BriefDescription": "Average number of uops fetched from DSB per cycle", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "IDQ.DSB_UOPS", - "Alias": "a" - }, - { - "Name": "IDQ.DSB_CYCLES_ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " idq.dsb_uops / idq.dsb_cycles_any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Fed;FetchBW", - "LocateWith": "" - }, - { - "MetricName": "Info_Pipeline_Fetch_MITE", - "LegacyName": "metric_TMA_Info_Pipeline_Fetch_MITE", - "Level": 1, - "BriefDescription": "Average number of uops fetched from MITE per cycle", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "IDQ.MITE_UOPS", - "Alias": "a" - }, - { - "Name": "IDQ.MITE_CYCLES_ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " idq.mite_uops / idq.mite_cycles_any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Fed;FetchBW", - "LocateWith": "" - }, - { - "MetricName": "Info_Frontend_Fetch_UpC", - "LegacyName": "metric_TMA_Info_Frontend_Fetch_UpC", - "Level": 1, - "BriefDescription": "Average number of Uops issued by front-end when it issued something", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UOPS_ISSUED.ANY", - "Alias": "a" - }, - { - "Name": "UOPS_ISSUED.ANY:c1", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " uops_issued.any / uops_issued.any:c1", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Fed;FetchBW", - "LocateWith": "" - }, - { - "MetricName": "Info_Frontend_DSB_Coverage", - "LegacyName": "metric_TMA_Info_Frontend_DSB_Coverage", - "Level": 1, - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "IDQ.DSB_UOPS", - "Alias": "a" - }, - { - "Name": "UOPS_ISSUED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / ( b )", - "BaseFormula": " idq.dsb_uops / ( uops_issued.any )", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Frontend_DSB_Coverage" - }, - { - "Alias": "b", - "Value": "IPC" - } - ], - "Formula": "a < 0.7 & b / 6 > 0.35", - "BaseFormula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 6 > 0.35", - "ThresholdIssues": "$issueFB" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "DSB;Fed;FetchBW", - "LocateWith": "" - }, - { - "MetricName": "Info_Frontend_Unknown_Branch_Cost", - "LegacyName": "metric_TMA_Info_Frontend_Unknown_Branch_Cost", - "Level": 1, - "BriefDescription": "Average number of cycles the front-end was delayed due to an Unknown Branch detection. See Unknown_Branches node.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INT_MISC.UNKNOWN_BRANCH_CYCLES", - "Alias": "a" - }, - { - "Name": "INT_MISC.UNKNOWN_BRANCH_CYCLES:c1:e1", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " int_misc.unknown_branch_cycles / int_misc.unknown_branch_cycles:c1:e1", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Fed", - "LocateWith": "" - }, - { - "MetricName": "Info_Frontend_DSB_Switch_Cost", - "LegacyName": "metric_TMA_Info_Frontend_DSB_Switch_Cost", - "Level": 1, - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "DSB2MITE_SWITCHES.PENALTY_CYCLES", - "Alias": "a" - }, - { - "Name": "DSB2MITE_SWITCHES.PENALTY_CYCLES:c1:e1", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " dsb2mite_switches.penalty_cycles / dsb2mite_switches.penalty_cycles:c1:e1", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "DSBmiss", - "LocateWith": "" - }, - { - "MetricName": "Info_Frontend_TBpC", - "LegacyName": "metric_TMA_Info_Frontend_TBpC", - "Level": 1, - "BriefDescription": "Taken Branches retired Per Cycle", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "BR_INST_RETIRED.NEAR_TAKEN", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / ( b )", - "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Metric", - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Branches;FetchBW", - "LocateWith": "" - }, - { - "MetricName": "Info_Frontend_ICache_Miss_Latency", - "LegacyName": "metric_TMA_Info_Frontend_ICache_Miss_Latency", - "Level": 1, - "BriefDescription": "Average Latency for L1 instruction cache misses", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "ICACHE_DATA.STALLS", - "Alias": "a" - }, - { - "Name": "ICACHE_DATA.STALLS:c1:e1", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " icache_data.stalls / icache_data.stalls:c1:e1", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Fed;FetchLat;IcMiss", - "LocateWith": "" - }, - { - "MetricName": "Info_Frontend_IpDSB_Miss_Ret", - "LegacyName": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret", - "Level": 1, - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "FRONTEND_RETIRED.ANY_DSB_MISS", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / frontend_retired.any_dsb_miss", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret" - } - ], - "Formula": "a < 50", - "BaseFormula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "DSBmiss;Fed", - "LocateWith": "" - }, - { - "MetricName": "Info_Frontend_IpUnknown_Branch", - "LegacyName": "metric_TMA_Info_Frontend_IpUnknown_Branch", - "Level": 1, - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "BACLEARS.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "( a ) / b", - "BaseFormula": " tma_info_inst_mix_instructions / baclears.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Fed", - "LocateWith": "" - }, - { - "MetricName": "Info_Frontend_L2MPKI_Code", - "LegacyName": "metric_TMA_Info_Frontend_L2MPKI_Code", - "Level": 1, - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction ", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "FRONTEND_RETIRED.L2_MISS", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * frontend_retired.l2_miss / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "IcMiss", - "LocateWith": "" - }, - { - "MetricName": "Info_Frontend_L2MPKI_Code_All", - "LegacyName": "metric_TMA_Info_Frontend_L2MPKI_Code_All", - "Level": 1, - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction ", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L2_RQSTS.CODE_RD_MISS", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * l2_rqsts.code_rd_miss / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "IcMiss", - "LocateWith": "" - }, - { - "MetricName": "Info_Botlnk_L2_DSB_Misses", - "LegacyName": "metric_TMA_Info_Botlnk_L2_DSB_Misses", - "Level": 1, - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.FETCH_LATENCY", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - }, - { - "Name": "DSB2MITE_SWITCHES.PENALTY_CYCLES", - "Alias": "h" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "i" - }, - { - "Name": "ICACHE_DATA.STALLS", - "Alias": "j" - }, - { - "Name": "ICACHE_TAG.STALLS", - "Alias": "k" - }, - { - "Name": "INT_MISC.CLEAR_RESTEER_CYCLES", - "Alias": "l" - }, - { - "Name": "INT_MISC.UNKNOWN_BRANCH_CYCLES", - "Alias": "m" - }, - { - "Name": "UOPS_RETIRED.MS:c1:e1", - "Alias": "n" - }, - { - "Name": "UOPS_RETIRED.SLOTS", - "Alias": "o" - }, - { - "Name": "UOPS_ISSUED.ANY", - "Alias": "p" - }, - { - "Name": "DECODE.LCP", - "Alias": "q" - }, - { - "Name": "IDQ.MITE_CYCLES_ANY", - "Alias": "r" - }, - { - "Name": "IDQ.MITE_CYCLES_OK", - "Alias": "s" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "t" - }, - { - "Name": "IDQ.DSB_CYCLES_ANY", - "Alias": "u" - }, - { - "Name": "IDQ.DSB_CYCLES_OK", - "Alias": "v" - }, - { - "Name": "IDQ.MS_CYCLES_ANY", - "Alias": "w" - }, - { - "Name": "UOPS_RETIRED.MS:c1", - "Alias": "x" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( h / ( i ) ) / ( ( j / ( i ) ) + ( k / ( i ) ) + ( l / ( i ) + ( m / ( i ) ) ) + ( ( 3 ) * n / ( o / p ) / ( i ) ) + ( q / ( i ) ) + ( h / ( i ) ) ) + ( max( 0 , ( b / ( b + c + d + e ) - f / ( g ) ) - ( ( a / ( b + c + d + e ) - f / ( g ) ) ) ) ) * ( ( r - s ) / ( t if smt_on else ( i ) ) / 2 ) / ( ( ( r - s ) / ( t if smt_on else ( i ) ) / 2 ) + ( ( u - v ) / ( t if smt_on else ( i ) ) / 2 ) + ( max( w , x / ( o / p ) ) / ( t if smt_on else ( i ) ) / 2 ) ) )", - "BaseFormula": " 100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb + tma_ms ) )", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Botlnk_DSB_Misses" - } - ], - "Formula": "a > 10", - "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Misses > 10", - "ThresholdIssues": "$issueFB" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "DSBmiss;Fed", - "LocateWith": "" - }, - { - "MetricName": "Info_Botlnk_L2_DSB_Bandwidth", - "LegacyName": "metric_TMA_Info_Botlnk_L2_DSB_Bandwidth", - "Level": 1, - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "d" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "e" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "f" - }, - { - "Name": "PERF_METRICS.FETCH_LATENCY", - "Alias": "g" - }, - { - "Name": "IDQ.DSB_CYCLES_ANY", - "Alias": "h" - }, - { - "Name": "IDQ.DSB_CYCLES_OK", - "Alias": "i" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "j" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "k" - }, - { - "Name": "IDQ.MITE_CYCLES_ANY", - "Alias": "l" - }, - { - "Name": "IDQ.MITE_CYCLES_OK", - "Alias": "m" - }, - { - "Name": "IDQ.MS_CYCLES_ANY", - "Alias": "n" - }, - { - "Name": "UOPS_RETIRED.MS:c1", - "Alias": "o" - }, - { - "Name": "UOPS_RETIRED.SLOTS", - "Alias": "p" - }, - { - "Name": "UOPS_ISSUED.ANY", - "Alias": "q" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "100 * ( ( a / ( a + b + c + d ) - e / ( f ) ) * ( ( max( 0 , ( a / ( a + b + c + d ) - e / ( f ) ) - ( ( g / ( a + b + c + d ) - e / ( f ) ) ) ) ) / ( ( ( g / ( a + b + c + d ) - e / ( f ) ) ) + ( max( 0 , ( a / ( a + b + c + d ) - e / ( f ) ) - ( ( g / ( a + b + c + d ) - e / ( f ) ) ) ) ) ) ) * ( ( ( h - i ) / ( j if smt_on else ( k ) ) / 2 ) / ( ( ( l - m ) / ( j if smt_on else ( k ) ) / 2 ) + ( ( h - i ) / ( j if smt_on else ( k ) ) / 2 ) + ( max( n , o / ( p / q ) ) / ( j if smt_on else ( k ) ) / 2 ) ) ) )", - "BaseFormula": " 100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb + tma_ms ) ) )", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Botlnk_DSB_Bandwidth" - } - ], - "Formula": "a > 10", - "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10", - "ThresholdIssues": "$issueFB" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "DSB;Fed;FetchBW", - "LocateWith": "" - }, - { - "MetricName": "Info_Botlnk_L2_IC_Misses", - "LegacyName": "metric_TMA_Info_Botlnk_L2_IC_Misses", - "Level": 1, - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "PERF_METRICS.FETCH_LATENCY", - "Alias": "a" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "e" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "f" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "g" - }, - { - "Name": "ICACHE_DATA.STALLS", - "Alias": "h" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "i" - }, - { - "Name": "ICACHE_TAG.STALLS", - "Alias": "j" - }, - { - "Name": "INT_MISC.CLEAR_RESTEER_CYCLES", - "Alias": "k" - }, - { - "Name": "INT_MISC.UNKNOWN_BRANCH_CYCLES", - "Alias": "l" - }, - { - "Name": "UOPS_RETIRED.MS:c1:e1", - "Alias": "m" - }, - { - "Name": "UOPS_RETIRED.SLOTS", - "Alias": "n" - }, - { - "Name": "UOPS_ISSUED.ANY", - "Alias": "o" - }, - { - "Name": "DECODE.LCP", - "Alias": "p" - }, - { - "Name": "DSB2MITE_SWITCHES.PENALTY_CYCLES", - "Alias": "q" - } - ], - "Constants": [], - "Formula": "100 * ( ( ( a / ( b + c + d + e ) - f / ( g ) ) ) * ( h / ( i ) ) / ( ( h / ( i ) ) + ( j / ( i ) ) + ( k / ( i ) + ( l / ( i ) ) ) + ( ( 3 ) * m / ( n / o ) / ( i ) ) + ( p / ( i ) ) + ( q / ( i ) ) ) )", - "BaseFormula": " 100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "Category": "TMA", - "CountDomain": "Scaled_Slots", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Botlnk_IC_Misses" - } - ], - "Formula": "a > 5", - "BaseFormula": "metric_TMA_Info_Botlnk_IC_Misses > 5", - "ThresholdIssues": "$issueFL" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Fed;FetchLat;IcMiss", - "LocateWith": "" - }, - { - "MetricName": "Info_Bad_Spec_IpMispredict", - "LegacyName": "metric_TMA_Info_Bad_Spec_IpMispredict", - "Level": 1, - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "BR_MISP_RETIRED.ALL_BRANCHES", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / br_misp_retired.all_branches", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Bad_Spec_IpMispredict" - } - ], - "Formula": "a < 200", - "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Bad;BadSpec;BrMispredicts", - "LocateWith": "" - }, - { - "MetricName": "Info_Bad_Spec_IpMisp_Cond_Ntaken", - "LegacyName": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken", - "Level": 1, - "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate).", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "BR_MISP_RETIRED.COND_NTAKEN", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / br_misp_retired.cond_ntaken", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken" - } - ], - "Formula": "a < 200", - "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Bad;BrMispredicts", - "LocateWith": "" - }, - { - "MetricName": "Info_Bad_Spec_IpMisp_Cond_Taken", - "LegacyName": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken", - "Level": 1, - "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate).", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "BR_MISP_RETIRED.COND_TAKEN", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / br_misp_retired.cond_taken", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken" - } - ], - "Formula": "a < 200", - "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken < 200", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Bad;BrMispredicts", - "LocateWith": "" - }, - { - "MetricName": "Info_Bad_Spec_IpMisp_Ret", - "LegacyName": "metric_TMA_Info_Bad_Spec_IpMisp_Ret", - "Level": 1, - "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate).", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "BR_MISP_RETIRED.RET", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / br_misp_retired.ret", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Ret" - } - ], - "Formula": "a < 500", - "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Bad;BrMispredicts", - "LocateWith": "" - }, - { - "MetricName": "Info_Bad_Spec_IpMisp_Indirect", - "LegacyName": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect", - "Level": 1, - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "BR_MISP_RETIRED.INDIRECT", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / br_misp_retired.indirect", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect" - } - ], - "Formula": "a < 1000", - "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Bad;BrMispredicts", - "LocateWith": "" - }, - { - "MetricName": "Info_Bad_Spec_Branch_Misprediction_Cost", - "LegacyName": "metric_TMA_Info_Bad_Spec_Branch_Misprediction_Cost", - "Level": 1, - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UOPS_RETIRED.MS", - "Alias": "a" - }, - { - "Name": "TOPDOWN.SLOTS:perf_metrics", - "Alias": "b" - }, - { - "Name": "PERF_METRICS.BRANCH_MISPREDICTS", - "Alias": "c" - }, - { - "Name": "PERF_METRICS.FRONTEND_BOUND", - "Alias": "d" - }, - { - "Name": "PERF_METRICS.BAD_SPECULATION", - "Alias": "e" - }, - { - "Name": "PERF_METRICS.RETIRING", - "Alias": "f" - }, - { - "Name": "PERF_METRICS.BACKEND_BOUND", - "Alias": "g" - }, - { - "Name": "BR_MISP_RETIRED.ALL_BRANCHES", - "Alias": "h" - }, - { - "Name": "INT_MISC.CLEARS_COUNT", - "Alias": "i" - }, - { - "Name": "MACHINE_CLEARS.COUNT", - "Alias": "j" - }, - { - "Name": "PERF_METRICS.FETCH_LATENCY", - "Alias": "k" - }, - { - "Name": "INT_MISC.UOP_DROPPING", - "Alias": "l" - }, - { - "Name": "INT_MISC.CLEAR_RESTEER_CYCLES", - "Alias": "m" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "n" - }, - { - "Name": "ICACHE_DATA.STALLS", - "Alias": "o" - }, - { - "Name": "ICACHE_TAG.STALLS", - "Alias": "p" - }, - { - "Name": "INT_MISC.UNKNOWN_BRANCH_CYCLES", - "Alias": "q" - }, - { - "Name": "UOPS_RETIRED.MS:c1:e1", - "Alias": "r" - }, - { - "Name": "UOPS_RETIRED.SLOTS", - "Alias": "s" - }, - { - "Name": "UOPS_ISSUED.ANY", - "Alias": "t" - }, - { - "Name": "DECODE.LCP", - "Alias": "u" - }, - { - "Name": "DSB2MITE_SWITCHES.PENALTY_CYCLES", - "Alias": "v" - } - ], - "Constants": [], - "Formula": "( 100 * ( 1 - ( 10 * ( a / ( b ) ) * ( max( ( c / ( d + e + f + g ) ) * ( 1 - h / ( i - j ) ) , 0.0001 ) ) / ( c / ( d + e + f + g ) ) ) ) * ( ( c / ( d + e + f + g ) ) + ( ( k / ( d + e + f + g ) - l / ( b ) ) ) * ( ( ( c / ( d + e + f + g ) ) / ( max( 1 - ( ( d / ( d + e + f + g ) - l / ( b ) ) + ( g / ( d + e + f + g ) ) + ( f / ( d + e + f + g ) ) ) , 0 ) ) ) * m / ( n ) ) / ( ( o / ( n ) ) + ( p / ( n ) ) + ( m / ( n ) + ( q / ( n ) ) ) + ( ( 3 ) * r / ( s / t ) / ( n ) ) + ( u / ( n ) ) + ( v / ( n ) ) ) ) ) * ( b ) / ( 6 ) / h / 100", - "BaseFormula": " tma_mispredictions * tma_info_thread_slots / ( 6 ) / br_misp_retired.all_branches / 100", - "Category": "TMA", - "CountDomain": "Core_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "$issueBM" - }, - "ResolutionLevels": "CORE, SOCKET, SYSTEM", - "MetricGroup": "Bad;BrMispredicts", - "LocateWith": "" - }, - { - "MetricName": "Info_Bad_Spec_Spec_Clears_Ratio", - "LegacyName": "metric_TMA_Info_Bad_Spec_Spec_Clears_Ratio", - "Level": 1, - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INT_MISC.CLEARS_COUNT", - "Alias": "a" - }, - { - "Name": "BR_MISP_RETIRED.ALL_BRANCHES", - "Alias": "b" - }, - { - "Name": "MACHINE_CLEARS.COUNT", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "a / ( b + c )", - "BaseFormula": " int_misc.clears_count / ( br_misp_retired.all_branches + machine_clears.count )", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "BrMispredicts", - "LocateWith": "" - }, - { - "MetricName": "Info_Branches_Cond_NT", - "LegacyName": "metric_TMA_Info_Branches_Cond_NT", - "Level": 1, - "BriefDescription": "Fraction of branches that are non-taken conditionals", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "BR_INST_RETIRED.COND_NTAKEN", - "Alias": "a" - }, - { - "Name": "BR_INST_RETIRED.ALL_BRANCHES", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " br_inst_retired.cond_ntaken / br_inst_retired.all_branches", - "Category": "TMA", - "CountDomain": "Fraction", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Bad;Branches;CodeGen;PGO", - "LocateWith": "" - }, - { - "MetricName": "Info_Branches_Cond_TK", - "LegacyName": "metric_TMA_Info_Branches_Cond_TK", - "Level": 1, - "BriefDescription": "Fraction of branches that are taken conditionals", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "BR_INST_RETIRED.COND_TAKEN", - "Alias": "a" - }, - { - "Name": "BR_INST_RETIRED.ALL_BRANCHES", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " br_inst_retired.cond_taken / br_inst_retired.all_branches", - "Category": "TMA", - "CountDomain": "Fraction", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Bad;Branches;CodeGen;PGO", - "LocateWith": "" - }, - { - "MetricName": "Info_Branches_CallRet", - "LegacyName": "metric_TMA_Info_Branches_CallRet", - "Level": 1, - "BriefDescription": "Fraction of branches that are CALL or RET", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "BR_INST_RETIRED.NEAR_CALL", - "Alias": "a" - }, - { - "Name": "BR_INST_RETIRED.NEAR_RETURN", - "Alias": "b" - }, - { - "Name": "BR_INST_RETIRED.ALL_BRANCHES", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "( a + b ) / c", - "BaseFormula": " ( br_inst_retired.near_call + br_inst_retired.near_return ) / br_inst_retired.all_branches", - "Category": "TMA", - "CountDomain": "Fraction", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Bad;Branches", - "LocateWith": "" - }, - { - "MetricName": "Info_Branches_Jump", - "LegacyName": "metric_TMA_Info_Branches_Jump", - "Level": 1, - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "BR_INST_RETIRED.NEAR_TAKEN", - "Alias": "a" - }, - { - "Name": "BR_INST_RETIRED.COND_TAKEN", - "Alias": "b" - }, - { - "Name": "BR_INST_RETIRED.NEAR_CALL", - "Alias": "c" - }, - { - "Name": "BR_INST_RETIRED.ALL_BRANCHES", - "Alias": "d" - } - ], - "Constants": [], - "Formula": "( a - b - 2 * c ) / d", - "BaseFormula": " ( br_inst_retired.near_taken - br_inst_retired.cond_taken - 2 * br_inst_retired.near_call ) / br_inst_retired.all_branches", - "Category": "TMA", - "CountDomain": "Fraction", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Bad;Branches", - "LocateWith": "" - }, - { - "MetricName": "Info_Branches_Other_Branches", - "LegacyName": "metric_TMA_Info_Branches_Other_Branches", - "Level": 1, - "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "BR_INST_RETIRED.COND_NTAKEN", - "Alias": "a" - }, - { - "Name": "BR_INST_RETIRED.ALL_BRANCHES", - "Alias": "b" - }, - { - "Name": "BR_INST_RETIRED.COND_TAKEN", - "Alias": "c" - }, - { - "Name": "BR_INST_RETIRED.NEAR_CALL", - "Alias": "d" - }, - { - "Name": "BR_INST_RETIRED.NEAR_RETURN", - "Alias": "e" - }, - { - "Name": "BR_INST_RETIRED.NEAR_TAKEN", - "Alias": "f" - } - ], - "Constants": [], - "Formula": "1 - ( ( a / b ) + ( c / b ) + ( ( d + e ) / b ) + ( ( f - c - 2 * d ) / b ) )", - "BaseFormula": " 1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", - "Category": "TMA", - "CountDomain": "Fraction", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Bad;Branches", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Load_Miss_Real_Latency", - "LegacyName": "metric_TMA_Info_Memory_Load_Miss_Real_Latency", - "Level": 1, - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L1D_PEND_MISS.PENDING", - "Alias": "a" - }, - { - "Name": "MEM_LOAD_COMPLETED.L1_MISS_ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " l1d_pend_miss.pending / mem_load_completed.l1_miss_any", - "Category": "TMA", - "CountDomain": "Clocks_Latency", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryBound;MemoryLat", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_MLP", - "LegacyName": "metric_TMA_Info_Memory_MLP", - "Level": 1, - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L1D_PEND_MISS.PENDING", - "Alias": "a" - }, - { - "Name": "L1D_PEND_MISS.PENDING_CYCLES", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " l1d_pend_miss.pending / l1d_pend_miss.pending_cycles", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryBound;MemoryBW", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_L1MPKI", - "LegacyName": "metric_TMA_Info_Memory_L1MPKI", - "Level": 1, - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "MEM_LOAD_RETIRED.L1_MISS", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * mem_load_retired.l1_miss / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "CacheHits;Mem", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_L1MPKI_Load", - "LegacyName": "metric_TMA_Info_Memory_L1MPKI_Load", - "Level": 1, - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L2_RQSTS.ALL_DEMAND_DATA_RD", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * l2_rqsts.all_demand_data_rd / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "CacheHits;Mem", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_L2MPKI", - "LegacyName": "metric_TMA_Info_Memory_L2MPKI", - "Level": 1, - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "MEM_LOAD_RETIRED.L2_MISS", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * mem_load_retired.l2_miss / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem;Backend;CacheHits", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_L2MPKI_All", - "LegacyName": "metric_TMA_Info_Memory_L2MPKI_All", - "Level": 1, - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L2_RQSTS.MISS", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * l2_rqsts.miss / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "CacheHits;Mem;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_L2MPKI_Load", - "LegacyName": "metric_TMA_Info_Memory_L2MPKI_Load", - "Level": 1, - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L2_RQSTS.DEMAND_DATA_RD_MISS", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_miss / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "CacheHits;Mem", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_L2MPKI_RFO", - "LegacyName": "metric_TMA_Info_Memory_L2MPKI_RFO", - "Level": 1, - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L2_RQSTS.RFO_MISS", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * l2_rqsts.rfo_miss / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "CacheMisses;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_L2HPKI_All", - "LegacyName": "metric_TMA_Info_Memory_L2HPKI_All", - "Level": 1, - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L2_RQSTS.REFERENCES", - "Alias": "a" - }, - { - "Name": "L2_RQSTS.MISS", - "Alias": "b" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "1000 * ( a - b ) / c", - "BaseFormula": " 1000 * ( l2_rqsts.references - l2_rqsts.miss ) / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "CacheHits;Mem", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_L2HPKI_Load", - "LegacyName": "metric_TMA_Info_Memory_L2HPKI_Load", - "Level": 1, - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L2_RQSTS.DEMAND_DATA_RD_HIT", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_hit / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "CacheHits;Mem", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_L3MPKI", - "LegacyName": "metric_TMA_Info_Memory_L3MPKI", - "Level": 1, - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "MEM_LOAD_RETIRED.L3_MISS", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * mem_load_retired.l3_miss / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_FB_HPKI", - "LegacyName": "metric_TMA_Info_Memory_FB_HPKI", - "Level": 1, - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "MEM_LOAD_RETIRED.FB_HIT", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * mem_load_retired.fb_hit / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "CacheHits;Mem", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_L1D_Cache_Fill_BW", - "LegacyName": "metric_TMA_Info_Memory_L1D_Cache_Fill_BW", - "Level": 1, - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L1D.REPLACEMENT", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "64 * a / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) )", - "BaseFormula": " 64 * l1d.replacement / ( 1000000000 ) / tma_info_system_time", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryBW", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_L2_Cache_Fill_BW", - "LegacyName": "metric_TMA_Info_Memory_L2_Cache_Fill_BW", - "Level": 1, - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L2_LINES_IN.ALL", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "64 * a / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) )", - "BaseFormula": " 64 * l2_lines_in.all / ( 1000000000 ) / tma_info_system_time", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryBW", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_L3_Cache_Fill_BW", - "LegacyName": "metric_TMA_Info_Memory_L3_Cache_Fill_BW", - "Level": 1, - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "LONGEST_LAT_CACHE.MISS", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "64 * a / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) )", - "BaseFormula": " 64 * longest_lat_cache.miss / ( 1000000000 ) / tma_info_system_time", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryBW", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_L3_Cache_Access_BW", - "LegacyName": "metric_TMA_Info_Memory_L3_Cache_Access_BW", - "Level": 1, - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "OFFCORE_REQUESTS.ALL_REQUESTS", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "64 * a / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) )", - "BaseFormula": " 64 * offcore_requests.all_requests / ( 1000000000 ) / tma_info_system_time", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryBW;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_TLB_Page_Walks_Utilization", - "LegacyName": "metric_TMA_Info_Memory_TLB_Page_Walks_Utilization", - "Level": 1, - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "ITLB_MISSES.WALK_PENDING", - "Alias": "a" - }, - { - "Name": "DTLB_LOAD_MISSES.WALK_PENDING", - "Alias": "b" - }, - { - "Name": "DTLB_STORE_MISSES.WALK_PENDING", - "Alias": "c" - }, - { - "Name": "CPU_CLK_UNHALTED.DISTRIBUTED", - "Alias": "d" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "e" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "( a + b + c ) / ( 4 * ( d if smt_on else ( e ) ) )", - "BaseFormula": " ( itlb_misses.walk_pending + dtlb_load_misses.walk_pending + dtlb_store_misses.walk_pending ) / ( 4 * tma_info_core_core_clks )", - "Category": "TMA", - "CountDomain": "Core_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_Memory_Page_Walks_Utilization" - } - ], - "Formula": "a > 0.5", - "BaseFormula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_TLB_Code_STLB_MPKI", - "LegacyName": "metric_TMA_Info_Memory_TLB_Code_STLB_MPKI", - "Level": 1, - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "ITLB_MISSES.WALK_COMPLETED", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * itlb_misses.walk_completed / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Fed;MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_TLB_Load_STLB_MPKI", - "LegacyName": "metric_TMA_Info_Memory_TLB_Load_STLB_MPKI", - "Level": 1, - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "DTLB_LOAD_MISSES.WALK_COMPLETED", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * dtlb_load_misses.walk_completed / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_TLB_Store_STLB_MPKI", - "LegacyName": "metric_TMA_Info_Memory_TLB_Store_STLB_MPKI", - "Level": 1, - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "DTLB_STORE_MISSES.WALK_COMPLETED", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * dtlb_store_misses.walk_completed / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryTLB", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Core_L1D_Cache_Fill_BW_2T", - "LegacyName": "metric_TMA_Info_Memory_Core_L1D_Cache_Fill_BW_2T", - "Level": 1, - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L1D.REPLACEMENT", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "( 64 * a / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) )", - "BaseFormula": " tma_info_memory_l1d_cache_fill_bw", - "Category": "TMA", - "CountDomain": "Core_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryBW", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Core_L2_Cache_Fill_BW_2T", - "LegacyName": "metric_TMA_Info_Memory_Core_L2_Cache_Fill_BW_2T", - "Level": 1, - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L2_LINES_IN.ALL", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "( 64 * a / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) )", - "BaseFormula": " tma_info_memory_l2_cache_fill_bw", - "Category": "TMA", - "CountDomain": "Core_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryBW", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Core_L3_Cache_Fill_BW_2T", - "LegacyName": "metric_TMA_Info_Memory_Core_L3_Cache_Fill_BW_2T", - "Level": 1, - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "LONGEST_LAT_CACHE.MISS", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "( 64 * a / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) )", - "BaseFormula": " tma_info_memory_l3_cache_fill_bw", - "Category": "TMA", - "CountDomain": "Core_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryBW", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Core_L3_Cache_Access_BW_2T", - "LegacyName": "metric_TMA_Info_Memory_Core_L3_Cache_Access_BW_2T", - "Level": 1, - "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "OFFCORE_REQUESTS.ALL_REQUESTS", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "( 64 * a / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) ) )", - "BaseFormula": " tma_info_memory_l3_cache_access_bw", - "Category": "TMA", - "CountDomain": "Core_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryBW;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Core_L2_Evictions_Silent_PKI", - "LegacyName": "metric_TMA_Info_Memory_Core_L2_Evictions_Silent_PKI", - "Level": 1, - "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L2_LINES_OUT.SILENT", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / ( b )", - "BaseFormula": " 1000 * l2_lines_out.silent / tma_info_inst_mix_instructions", - "Category": "TMA", - "CountDomain": "Core_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE, SOCKET, SYSTEM", - "MetricGroup": "L2Evicts;Mem;Server", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Core_L2_Evictions_NonSilent_PKI", - "LegacyName": "metric_TMA_Info_Memory_Core_L2_Evictions_NonSilent_PKI", - "Level": 1, - "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L2_LINES_OUT.NON_SILENT", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / ( b )", - "BaseFormula": " 1000 * l2_lines_out.non_silent / tma_info_inst_mix_instructions", - "Category": "TMA", - "CountDomain": "Core_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE, SOCKET, SYSTEM", - "MetricGroup": "L2Evicts;Mem;Server", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Prefetches_Useless_HWPF", - "LegacyName": "metric_TMA_Info_Memory_Prefetches_Useless_HWPF", - "Level": 1, - "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "L2_LINES_OUT.USELESS_HWPF", - "Alias": "a" - }, - { - "Name": "L2_LINES_OUT.SILENT", - "Alias": "b" - }, - { - "Name": "L2_LINES_OUT.NON_SILENT", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "a / ( b + c )", - "BaseFormula": " l2_lines_out.useless_hwpf / ( l2_lines_out.silent + l2_lines_out.non_silent )", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Prefetches", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Latency_Load_L2_Miss_Latency", - "LegacyName": "metric_TMA_Info_Memory_Latency_Load_L2_Miss_Latency", - "Level": 1, - "BriefDescription": "Average Latency for L2 cache miss demand Loads", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD", - "Alias": "a" - }, - { - "Name": "OFFCORE_REQUESTS.DEMAND_DATA_RD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests.demand_data_rd", - "Category": "TMA", - "CountDomain": "Clocks_Latency", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "LockCont;Memory_Lat;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Latency_Load_L3_Miss_Latency", - "LegacyName": "metric_TMA_Info_Memory_Latency_Load_L3_Miss_Latency", - "Level": 1, - "BriefDescription": "Average Latency for L3 cache miss demand Loads", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.L3_MISS_DEMAND_DATA_RD", - "Alias": "a" - }, - { - "Name": "OFFCORE_REQUESTS.L3_MISS_DEMAND_DATA_RD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " offcore_requests_outstanding.l3_miss_demand_data_rd / offcore_requests.l3_miss_demand_data_rd", - "Category": "TMA", - "CountDomain": "Clocks_Latency", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Memory_Lat;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Latency_Load_L2_MLP", - "LegacyName": "metric_TMA_Info_Memory_Latency_Load_L2_MLP", - "Level": 1, - "BriefDescription": "Average Parallel L2 cache miss demand Loads", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD", - "Alias": "a" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD:c1", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests_outstanding.demand_data_rd:c1", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Memory_BW;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Latency_Data_L2_MLP", - "LegacyName": "metric_TMA_Info_Memory_Latency_Data_L2_MLP", - "Level": 1, - "BriefDescription": "Average Parallel L2 cache miss data reads", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.DATA_RD", - "Alias": "a" - }, - { - "Name": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " offcore_requests_outstanding.data_rd / offcore_requests_outstanding.cycles_with_data_rd", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Memory_BW;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Mix_Offcore_Read_Any_PKI", - "LegacyName": "metric_TMA_Info_Memory_Mix_Offcore_Read_Any_PKI", - "Level": 1, - "BriefDescription": "Off-core accesses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "OCR.READS_TO_CORE.ANY_RESPONSE", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / ( b )", - "BaseFormula": " 1000 * ocr.reads_to_core.any_response / tma_info_inst_mix_instructions", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "CacheHits;Offcore", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Mix_Offcore_Read_L3M_PKI", - "LegacyName": "metric_TMA_Info_Memory_Mix_Offcore_Read_L3M_PKI", - "Level": 1, - "BriefDescription": "L3 cache misses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "OCR.READS_TO_CORE.L3_MISS", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / ( b )", - "BaseFormula": " 1000 * ocr.reads_to_core.l3_miss / tma_info_inst_mix_instructions", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Offcore", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Mix_Offcore_Read_HBM_PKI", - "LegacyName": "metric_TMA_Info_Memory_Mix_Offcore_Read_HBM_PKI", - "Level": 1, - "BriefDescription": "High-Bandwidth Memory (HBM) accesses per kilo instruction for reads-to-core requests (speculative; including in-core HW prefetches)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "OCR.DEMAND_DATA_RD.PMM", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "#NA if 0 > 2 else 1000 * a / ( b )", - "BaseFormula": " if 0 > 2 else 1000 * ocr.demand_data_rd.pmm / tma_info_inst_mix_instructions", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Offcore;Server", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Mix_Offcore_MWrite_Any_PKI", - "LegacyName": "metric_TMA_Info_Memory_Mix_Offcore_MWrite_Any_PKI", - "Level": 1, - "BriefDescription": "Off-core accesses per kilo instruction for modified write requests", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "OCR.MODIFIED_WRITE.ANY_RESPONSE", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / ( b )", - "BaseFormula": " 1000 * ocr.modified_write.any_response / tma_info_inst_mix_instructions", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Offcore", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Mix_UC_Load_PKI", - "LegacyName": "metric_TMA_Info_Memory_Mix_UC_Load_PKI", - "Level": 1, - "BriefDescription": "Un-cacheable retired load per kilo instruction", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "MEM_LOAD_MISC_RETIRED.UC", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * mem_load_misc_retired.uc / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_Mix_Bus_Lock_PKI", - "LegacyName": "metric_TMA_Info_Memory_Mix_Bus_Lock_PKI", - "Level": 1, - "BriefDescription": "\"Bus lock\" per kilo instruction", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "SQ_MISC.BUS_LOCK", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "1000 * a / b", - "BaseFormula": " 1000 * sq_misc.bus_lock / inst_retired.any", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Mem", - "LocateWith": "" - }, - { - "MetricName": "Info_System_CPU_Utilization", - "LegacyName": "metric_TMA_Info_System_CPU_Utilization", - "Level": 1, - "BriefDescription": "Average CPU Utilization (percentage)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "b" - }, - { - "Name": "system.sockets[0].cpus.count * system.socket_count", - "Alias": "c" - } - ], - "Formula": "( a / b ) / c", - "BaseFormula": " tma_info_system_cpus_utilized / num_cpus", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "HPC;Summary", - "LocateWith": "" - }, - { - "MetricName": "Info_System_CPUs_Utilized", - "LegacyName": "metric_TMA_Info_System_CPUs_Utilized", - "Level": 1, - "BriefDescription": "Average number of utilized CPUs", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "b" - } - ], - "Formula": "a / b", - "BaseFormula": " cpu_clk_unhalted.ref_tsc / tsc", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "SYSTEM", - "MetricGroup": "Summary", - "LocateWith": "" - }, - { - "MetricName": "Info_System_Core_Frequency", - "LegacyName": "metric_TMA_Info_System_Core_Frequency", - "Level": 1, - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "b" - } - ], - "Constants": [ - { - "Name": "SYSTEM_TSC_FREQ", - "Alias": "c" - }, - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "( ( a ) / b ) * c / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) )", - "BaseFormula": " tma_info_system_turbo_utilization * tsc / ( 1000000000 ) / tma_info_system_time", - "Category": "TMA", - "CountDomain": "System_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "SOCKET, SYSTEM", - "MetricGroup": "Summary;Power", - "LocateWith": "" - }, - { - "MetricName": "Info_System_Uncore_Frequency", - "LegacyName": "metric_TMA_Info_System_Uncore_Frequency", - "Level": 1, - "BriefDescription": "Measured Average Uncore Frequency for the SoC [GHz]", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UNC_CHA_CLOCKTICKS:one_unit", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "( a ) / 1e9 / ( ( durationtimeinmilliseconds / 1000 ) )", - "BaseFormula": " tma_info_system_socket_clks / 1e9 / tma_info_system_time", - "Category": "TMA", - "CountDomain": "System_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "SOCKET, SYSTEM", - "MetricGroup": "SoC", - "LocateWith": "" - }, - { - "MetricName": "Info_System_GFLOPs", - "LegacyName": "metric_TMA_Info_System_GFLOPs", - "Level": 1, - "BriefDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "FP_ARITH_INST_RETIRED.SCALAR", - "Alias": "a" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.SCALAR_HALF", - "Alias": "b" - }, - { - "Name": "FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE", - "Alias": "c" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.COMPLEX_SCALAR_HALF", - "Alias": "d" - }, - { - "Name": "FP_ARITH_INST_RETIRED.4_FLOPS", - "Alias": "e" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.128B_PACKED_HALF", - "Alias": "f" - }, - { - "Name": "FP_ARITH_INST_RETIRED.8_FLOPS", - "Alias": "g" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.256B_PACKED_HALF", - "Alias": "h" - }, - { - "Name": "FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE", - "Alias": "i" - }, - { - "Name": "FP_ARITH_INST_RETIRED2.512B_PACKED_HALF", - "Alias": "j" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "( ( ( 1 * ( a + b ) + 2 * ( c + d ) + 4 * e + 8 * ( f + g ) + 16 * ( h + i ) + 32 * j ) ) / ( 1000000000 ) ) / ( ( durationtimeinmilliseconds / 1000 ) )", - "BaseFormula": " ( ( ( 1 * ( fp_arith_inst_retired.scalar + fp_arith_inst_retired2.scalar_half ) + 2 * ( fp_arith_inst_retired.128b_packed_double + fp_arith_inst_retired2.complex_scalar_half ) + 4 * fp_arith_inst_retired.4_flops + 8 * ( fp_arith_inst_retired2.128b_packed_half + fp_arith_inst_retired.8_flops ) + 16 * ( fp_arith_inst_retired2.256b_packed_half + fp_arith_inst_retired.512b_packed_single ) + 32 * fp_arith_inst_retired2.512b_packed_half ) ) / ( 1000000000 ) ) / tma_info_system_time", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Cor;Flops;HPC", - "LocateWith": "" - }, - { - "MetricName": "Info_System_Turbo_Utilization", - "LegacyName": "metric_TMA_Info_System_Turbo_Utilization", - "Level": 1, - "BriefDescription": "Average Frequency Utilization relative nominal frequency", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_TSC", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "( a ) / b", - "BaseFormula": " tma_info_thread_clks / cpu_clk_unhalted.ref_tsc", - "Category": "TMA", - "CountDomain": "Core_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE, SOCKET, SYSTEM", - "MetricGroup": "Power", - "LocateWith": "" - }, - { - "MetricName": "Info_System_SMT_2T_Utilization", - "LegacyName": "metric_TMA_Info_System_SMT_2T_Utilization", - "Level": 1, - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.REF_DISTRIBUTED", - "Alias": "b" - } - ], - "Constants": [ - { - "Name": "HYPERTHREADING_ON", - "Alias": "smt_on" - }, - { - "Name": "THREADS_PER_CORE", - "Alias": "threads" - } - ], - "Formula": "1 - a / b if smt_on else 0", - "BaseFormula": " 1 - cpu_clk_unhalted.one_thread_active / cpu_clk_unhalted.ref_distributed if smt_on else 0", - "Category": "TMA", - "CountDomain": "Core_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "CORE, SOCKET, SYSTEM", - "MetricGroup": "SMT", - "LocateWith": "" - }, - { - "MetricName": "Info_System_Kernel_Utilization", - "LegacyName": "metric_TMA_Info_System_Kernel_Utilization", - "Level": 1, - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD_P:SUP", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " cpu_clk_unhalted.thread_p:sup / cpu_clk_unhalted.thread", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_System_Kernel_Utilization" - } - ], - "Formula": "a > 0.05", - "BaseFormula": "metric_TMA_Info_System_Kernel_Utilization > 0.05", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "OS", - "LocateWith": "" - }, - { - "MetricName": "Info_System_Kernel_CPI", - "LegacyName": "metric_TMA_Info_System_Kernel_CPI", - "Level": 1, - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", - "UnitOfMeasure": "per instruction", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD_P:SUP", - "Alias": "a" - }, - { - "Name": "INST_RETIRED.ANY_P:SUP", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " cpu_clk_unhalted.thread_p:sup / inst_retired.any_p:sup", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "OS", - "LocateWith": "" - }, - { - "MetricName": "Info_System_C0_Wait", - "LegacyName": "metric_TMA_Info_System_C0_Wait", - "Level": 1, - "BriefDescription": "Fraction of cycles the processor is waiting yet unhalted; covering legacy PAUSE instruction, as well as C0.1 / C0.2 power-performance optimized states", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.C0_WAIT", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / ( b )", - "BaseFormula": " cpu_clk_unhalted.c0_wait / tma_info_thread_clks", - "Category": "TMA", - "CountDomain": "Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_System_C0_Wait" - } - ], - "Formula": "a > 0.05", - "BaseFormula": "metric_TMA_Info_System_C0_Wait > 0.05", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "C0Wait", - "LocateWith": "" - }, - { - "MetricName": "Info_System_DRAM_BW_Use", - "LegacyName": "metric_TMA_Info_System_DRAM_BW_Use", - "Level": 1, - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UNC_M_CAS_COUNT.RD", - "Alias": "a" - }, - { - "Name": "UNC_M_CAS_COUNT.WR", - "Alias": "b" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "( 64 * ( a + b ) / ( 1000000000 ) ) / ( ( durationtimeinmilliseconds / 1000 ) )", - "BaseFormula": " ( 64 * ( unc_m_cas_count.rd + unc_m_cas_count.wr ) / ( 1000000000 ) ) / tma_info_system_time", - "Category": "TMA", - "CountDomain": "GB/sec", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "$issueBW" - }, - "ResolutionLevels": "CHANNEL, IMC, SOCKET, SYSTEM", - "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC", - "LocateWith": "" - }, - { - "MetricName": "Info_System_MEM_Read_Latency", - "LegacyName": "metric_TMA_Info_System_MEM_Read_Latency", - "Level": 1, - "BriefDescription": "Average latency of data read request to external memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetches. ([RKL+]memory-controller only)", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD", - "Alias": "b" - }, - { - "Name": "UNC_CHA_CLOCKTICKS:one_unit", - "Alias": "c" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "( 1000000000 ) * ( a / b ) / ( ( c ) / ( ( durationtimeinmilliseconds / 1000 ) ) )", - "BaseFormula": " ( 1000000000 ) * ( unc_cha_tor_occupancy.ia_miss_drd / unc_cha_tor_inserts.ia_miss_drd ) / ( tma_info_system_socket_clks / tma_info_system_time )", - "Category": "TMA", - "CountDomain": "NanoSeconds", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryLat;SoC", - "LocateWith": "" - }, - { - "MetricName": "Info_System_MEM_Parallel_Reads", - "LegacyName": "metric_TMA_Info_System_MEM_Parallel_Reads", - "Level": 1, - "BriefDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD:c1", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " unc_cha_tor_occupancy.ia_miss_drd / unc_cha_tor_occupancy.ia_miss_drd:c1", - "Category": "TMA", - "CountDomain": "System_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "SOCKET, SYSTEM", - "MetricGroup": "Mem;MemoryBW;SoC", - "LocateWith": "" - }, - { - "MetricName": "Info_System_MEM_DRAM_Read_Latency", - "LegacyName": "metric_TMA_Info_System_MEM_DRAM_Read_Latency", - "Level": 1, - "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]. Accounts for demand loads and L1/L2 data-read prefetches", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR", - "Alias": "b" - }, - { - "Name": "UNC_CHA_CLOCKTICKS:one_unit", - "Alias": "c" - } - ], - "Constants": [], - "Formula": "( 1000000000 ) * ( a / b ) / c", - "BaseFormula": " ( 1000000000 ) * ( unc_cha_tor_occupancy.ia_miss_drd_ddr / unc_cha_tor_inserts.ia_miss_drd_ddr ) / unc_cha_clockticks:one_unit", - "Category": "TMA", - "CountDomain": "NanoSeconds", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "SOCKET, SYSTEM", - "MetricGroup": "MemOffcore;MemoryLat;SoC;Server", - "LocateWith": "" - }, - { - "MetricName": "Info_System_IO_Read_BW", - "LegacyName": "metric_TMA_Info_System_IO_Read_BW", - "Level": 1, - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]. Bandwidth of IO reads that are initiated by end device controllers that are requesting memory from the CPU", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "a * 64 / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) )", - "BaseFormula": " unc_cha_tor_inserts.io_pcirdcur * 64 / ( 1000000000 ) / tma_info_system_time", - "Category": "TMA", - "CountDomain": "GB/sec", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "SOCKET, SYSTEM", - "MetricGroup": "IoBW;MemOffcore;SoC;Server", - "LocateWith": "" - }, - { - "MetricName": "Info_System_IO_Write_BW", - "LegacyName": "metric_TMA_Info_System_IO_Write_BW", - "Level": 1, - "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]. Bandwidth of IO writes that are initiated by end device controllers that are writing memory to the CPU", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UNC_CHA_TOR_INSERTS.IO_ITOM", - "Alias": "a" - }, - { - "Name": "UNC_CHA_TOR_INSERTS.IO_ITOMCACHENEAR", - "Alias": "b" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "( a + b ) * 64 / ( 1000000000 ) / ( ( durationtimeinmilliseconds / 1000 ) )", - "BaseFormula": " ( unc_cha_tor_inserts.io_itom + unc_cha_tor_inserts.io_itomcachenear ) * 64 / ( 1000000000 ) / tma_info_system_time", - "Category": "TMA", - "CountDomain": "GB/sec", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "SOCKET, SYSTEM", - "MetricGroup": "IoBW;MemOffcore;SoC;Server", - "LocateWith": "" - }, - { - "MetricName": "Info_System_UPI_Data_Transmit_BW", - "LegacyName": "metric_TMA_Info_System_UPI_Data_Transmit_BW", - "Level": 1, - "BriefDescription": "Cross-socket Ultra Path Interconnect (UPI) data transmit bandwidth for data only [MB / sec]", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UNC_UPI_TxL_FLITS.ALL_DATA", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "a * 64 / 9 / 1000000", - "BaseFormula": " unc_upi_txl_flits.all_data * 64 / 9 / 1000000", - "Category": "TMA", - "CountDomain": "MB/sec", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "UPI, SOCKET, SYSTEM", - "MetricGroup": "SoC;Server", - "LocateWith": "" - }, - { - "MetricName": "Info_System_Power", - "LegacyName": "metric_TMA_Info_System_Power", - "Level": 1, - "BriefDescription": "Total package Power in Watts", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "FREERUN_PKG_ENERGY_STATUS", - "Alias": "a" - }, - { - "Name": "FREERUN_DRAM_ENERGY_STATUS", - "Alias": "b" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "( a * ( 61 ) + 15.6 * b ) / ( ( durationtimeinmilliseconds / 1000 ) * ( 1000000 ) )", - "BaseFormula": " ( freerun_pkg_energy_status * ( 61 ) + 15.6 * freerun_dram_energy_status ) / ( ( duration_time ) * ( 1000000 ) )", - "Category": "TMA", - "CountDomain": "System_Metric", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "SOCKET, SYSTEM", - "MetricGroup": "Power;SoC", - "LocateWith": "" - }, - { - "MetricName": "Info_System_Time", - "LegacyName": "metric_TMA_Info_System_Time", - "Level": 1, - "BriefDescription": "Run duration time in seconds", - "UnitOfMeasure": "", - "Events": [], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "( durationtimeinmilliseconds / 1000 )", - "BaseFormula": " duration_time", - "Category": "TMA", - "CountDomain": "Seconds", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_System_Time" - } - ], - "Formula": "a < 1", - "BaseFormula": "metric_TMA_Info_System_Time < 1", - "ThresholdIssues": "" - }, - "ResolutionLevels": "SOCKET, SYSTEM", - "MetricGroup": "Summary", - "LocateWith": "" - }, - { - "MetricName": "Info_System_MUX", - "LegacyName": "metric_TMA_Info_System_MUX", - "Level": 1, - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "CPU_CLK_UNHALTED.THREAD_P", - "Alias": "a" - }, - { - "Name": "CPU_CLK_UNHALTED.THREAD", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " cpu_clk_unhalted.thread_p / cpu_clk_unhalted.thread", - "Category": "TMA", - "CountDomain": "Clocks", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_System_MUX" - } - ], - "Formula": "a > 1.1 | a < 0.9", - "BaseFormula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Summary", - "LocateWith": "" - }, - { - "MetricName": "Info_System_Socket_CLKS", - "LegacyName": "metric_TMA_Info_System_Socket_CLKS", - "Level": 1, - "BriefDescription": "Socket actual clocks when any core is active on that socket", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "UNC_CHA_CLOCKTICKS:one_unit", - "Alias": "a" - } - ], - "Constants": [], - "Formula": "a", - "BaseFormula": " unc_cha_clockticks:one_unit", - "Category": "TMA", - "CountDomain": "Count", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "SOCKET, SYSTEM", - "MetricGroup": "SoC", - "LocateWith": "" - }, - { - "MetricName": "Info_System_IpFarBranch", - "LegacyName": "metric_TMA_Info_System_IpFarBranch", - "Level": 1, - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "INST_RETIRED.ANY", - "Alias": "a" - }, - { - "Name": "BR_INST_RETIRED.FAR_BRANCH:USER", - "Alias": "b" - } - ], - "Constants": [], - "Formula": "a / b", - "BaseFormula": " inst_retired.any / br_inst_retired.far_branch:user", - "Category": "TMA", - "CountDomain": "Inst_Metric", - "Threshold": { - "ThresholdMetrics": [ - { - "Alias": "a", - "Value": "metric_TMA_Info_System_IpFarBranch" - } - ], - "Formula": "a < 1000000", - "BaseFormula": "metric_TMA_Info_System_IpFarBranch < 1000000", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "Branches;OS", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_SoC_R2C_Offcore_BW", - "LegacyName": "metric_TMA_Info_Memory_SoC_R2C_Offcore_BW", - "Level": 1, - "BriefDescription": "Average Off-core access BW for Reads-to-Core (R2C). R2C account for demand or prefetch load/RFO/code access that fill data into the Core caches.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "OCR.READS_TO_CORE.ANY_RESPONSE", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "64 * a / 1e9 / ( ( durationtimeinmilliseconds / 1000 ) )", - "BaseFormula": " 64 * ocr.reads_to_core.any_response / 1e9 / tma_info_system_time", - "Category": "TMA", - "CountDomain": "GB/sec", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "HPC;Mem;MemoryBW;SoC", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_SoC_R2C_L3M_BW", - "LegacyName": "metric_TMA_Info_Memory_SoC_R2C_L3M_BW", - "Level": 1, - "BriefDescription": "Average L3-cache miss BW for Reads-to-Core (R2C). This covering going to DRAM or other memory off-chip memory tears. See R2C_Offcore_BW.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "OCR.READS_TO_CORE.L3_MISS", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "64 * a / 1e9 / ( ( durationtimeinmilliseconds / 1000 ) )", - "BaseFormula": " 64 * ocr.reads_to_core.l3_miss / 1e9 / tma_info_system_time", - "Category": "TMA", - "CountDomain": "GB/sec", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "HPC;Mem;MemoryBW;SoC", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_SoC_R2C_DRAM_BW", - "LegacyName": "metric_TMA_Info_Memory_SoC_R2C_DRAM_BW", - "Level": 1, - "BriefDescription": "Average DRAM BW for Reads-to-Core (R2C) covering for memory attached to local- and remote-socket. See R2C_Offcore_BW.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "OCR.READS_TO_CORE.DRAM", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "64 * a / 1e9 / ( ( durationtimeinmilliseconds / 1000 ) )", - "BaseFormula": " 64 * ocr.reads_to_core.dram / 1e9 / tma_info_system_time", - "Category": "TMA", - "CountDomain": "GB/sec", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "HPC;Mem;MemoryBW;SoC", - "LocateWith": "" - }, - { - "MetricName": "Info_Memory_SoC_R2C_HBM_BW", - "LegacyName": "metric_TMA_Info_Memory_SoC_R2C_HBM_BW", - "Level": 1, - "BriefDescription": "Average HBM BW for Reads-to-Core. See R2C_Offcore_BW.", - "UnitOfMeasure": "", - "Events": [ - { - "Name": "OCR.DEMAND_DATA_RD.PMM", - "Alias": "a" - } - ], - "Constants": [ - { - "Name": "DURATIONTIMEINMILLISECONDS", - "Alias": "durationtimeinmilliseconds" - } - ], - "Formula": "#NA if 0 > 2 else 64 * a / 1e9 / ( ( durationtimeinmilliseconds / 1000 ) )", - "BaseFormula": " if 0 > 2 else 64 * ocr.demand_data_rd.pmm / 1e9 / tma_info_system_time", - "Category": "TMA", - "CountDomain": "GB/sec", - "Threshold": { - "Formula": "", - "BaseFormula": "", - "ThresholdIssues": "" - }, - "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", - "MetricGroup": "HPC;Mem;MemoryBW;Server;SoC", - "LocateWith": "" - } - ] -} \ No newline at end of file diff --git a/TGL/metrics/perf/tigerlake_metrics_perf.json b/TGL/metrics/perf/tigerlake_metrics_perf.json index 1d7ecc0b..6943aef2 100644 --- a/TGL/metrics/perf/tigerlake_metrics_perf.json +++ b/TGL/metrics/perf/tigerlake_metrics_perf.json @@ -1,1586 +1,1750 @@ [ { - "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks.", + "BriefDescription": "Total pipeline cost of Branch Misprediction related bottlenecks", "MetricExpr": "100 * ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * ( tma_branch_mispredicts + tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", + "MetricThreshold": "tma_bottleneck_mispredictions > 20", + "PublicDescription": "Total pipeline cost of Branch Misprediction related bottlenecks. Related metrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses).", + "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", "MetricExpr": "100 * tma_fetch_latency * ( tma_itlb_misses + tma_icache_misses + tma_unknown_branches ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches )", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_big_code", + "MetricThreshold": "tma_bottleneck_big_code > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end).", + "BriefDescription": "Total pipeline cost of instruction fetch bandwidth related bottlenecks (when the front-end could not sustain operations delivery to the back-end)", "MetricExpr": "100 * ( tma_frontend_bound - ( 1 - ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) ) * tma_fetch_latency * tma_mispredicts_resteers / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_mispredicts_resteers ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) ) ) - tma_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_instruction_fetch_bw", + "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_bandwidth / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_sq_full / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_fb_full / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) )", - "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", + "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks.", + "BriefDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks", "MetricExpr": "100 * ( ( tma_memory_bound * ( tma_dram_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_mem_latency / ( tma_mem_bandwidth + tma_mem_latency ) ) ) + ( tma_memory_bound * ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l3_hit_latency / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) ) ) + ( tma_memory_bound * tma_l2_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_l1_latency_dependency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_lock_latency / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_l1_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_loads / ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_split_stores / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_store_latency / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvML;Mem;MemoryLat;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", + "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", + "PublicDescription": "Total pipeline cost of external Memory- or Cache-Latency related bottlenecks. Related metrics: tma_l3_hit_latency, tma_mem_latency.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs).", + "BriefDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs)", "MetricExpr": "100 * ( tma_memory_bound * ( tma_l1_bound / max( tma_memory_bound , ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) ) * ( tma_dtlb_load / max( tma_l1_bound , ( tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_full ) ) ) + ( tma_memory_bound * ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_dtlb_store / ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) ) ) )", - "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", + "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", + "PublicDescription": "Total pipeline cost of Memory Address Translation related bottlenecks (data-side TLBs). Related metrics: tma_dtlb_load, tma_dtlb_store.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors).", + "BriefDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors)", "MetricExpr": "100 * ( tma_memory_bound * ( ( tma_l3_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * ( tma_contested_accesses + tma_data_sharing ) / ( tma_contested_accesses + tma_data_sharing + tma_l3_hit_latency + tma_sq_full ) + ( tma_store_bound / ( tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound ) ) * tma_false_sharing / ( ( tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store ) - tma_store_latency ) ) + tma_machine_clears * ( 1 - tma_other_nukes / ( tma_other_nukes ) ) )", - "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvMS;LockCont;Mem;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", + "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", + "PublicDescription": "Total pipeline cost of Memory Synchronization related bottlenecks (data transfers and coherency updates across processors). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation.", + "BriefDescription": "Total pipeline cost when the execution is compute-bound - an estimation", "MetricExpr": "100 * ( ( tma_core_bound * tma_divider / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( tma_core_bound * ( tma_ports_utilization / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) * ( tma_ports_utilized_3m / ( tma_ports_utilized_0 + tma_ports_utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m ) ) ) )", - "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "BvCB;Cor;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", + "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", "PublicDescription": "Total pipeline cost when the execution is compute-bound - an estimation. Covers Core Bound when High ILP as well as when long-latency execution units are busy.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of irregular execution (e.g.", + "BriefDescription": "Total pipeline cost of irregular execution (e.g", "MetricExpr": "100 * ( ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * ( tma_fetch_latency * ( tma_ms_switches + tma_branch_resteers * ( tma_clears_resteers + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_mispredicts_resteers ) / ( tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches ) ) / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_ms / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) ) + ( 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts ) * tma_branch_mispredicts + ( tma_machine_clears * tma_other_nukes / ( tma_other_nukes ) ) + ( tma_core_bound * ( tma_serializing_operation + tma_core_bound * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0 ) / ( tma_divider + tma_serializing_operation + tma_ports_utilization ) ) + ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", - "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", + "MetricGroup": "Bad;BvIO;Cor;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", - "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments).", + "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", + "PublicDescription": "Total pipeline cost of irregular execution (e.g. FP-assists in HPC, Wait time with work imbalance multithreaded workloads, overhead in system services or virtualized environments). Related metrics: tma_microcode_sequencer, tma_ms_switches.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end.", + "BriefDescription": "Total pipeline cost of remaining bottlenecks in the back-end", "MetricExpr": "100 - ( tma_big_code + tma_instruction_fetch_bw + tma_mispredictions + tma_cache_memory_bandwidth + tma_cache_memory_latency + tma_memory_data_tlbs + tma_memory_synchronization + tma_compute_bound_est + tma_irregular_overhead + tma_branching_overhead + tma_useful_work )", "MetricGroup": "BvOB;Cor;Offcore;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_other_bottlenecks", + "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", "PublicDescription": "Total pipeline cost of remaining bottlenecks in the back-end. Examples include data-dependencies (Core Bound when Low ILP) and other unlisted memory-related stalls.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA.", + "BriefDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA", "MetricExpr": "100 * ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots )", "MetricGroup": "BvBO;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_branching_overhead", + "MetricThreshold": "tma_bottleneck_branching_overhead > 5", "PublicDescription": "Total pipeline cost of instructions used for program control-flow - a subset of the Retiring category in TMA. Examples include function calls; loops and alignments. (A lower bound).", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead.", + "BriefDescription": "Total pipeline cost of \"useful operations\" - the portion of Retiring category not covered by Branching_Overhead nor Irregular_Overhead", "MetricExpr": "100 * ( tma_retiring - ( ( BR_INST_RETIRED.ALL_BRANCHES + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP ) / tma_info_thread_slots ) - ( ( ( tma_microcode_sequencer / ( tma_few_uops_instructions + tma_microcode_sequencer ) ) * ( tma_assists / tma_microcode_sequencer ) ) * tma_heavy_operations ) )", "MetricGroup": "BvUW;Ret;TopdownL1;tma_L1_group;Default;Scaled_Slots", "MetricName": "tma_bottleneck_useful_work", + "MetricThreshold": "tma_bottleneck_useful_work > 20", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend.", + "BriefDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend", "MetricExpr": "topdown\\-fe\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) - INT_MISC.UOP_DROPPING / tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;TmaL1;PGO;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_frontend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_frontend_bound > 0.15", "PublicDescription": "This category represents fraction of slots where the processor's Frontend undersupplies its Backend. Frontend denotes the first part of the processor core responsible to fetch operations that are executed later on by the Backend part. Within the Frontend; a branch predictor predicts the next address to fetch; cache-lines are fetched from the memory subsystem; parsed into instructions; and lastly decoded into micro-operations (uops). Ideally the Frontend can issue Pipeline_Width uops every cycle to the Backend. Frontend Bound denotes unutilized issue-slots when there is no Backend stall; i.e. bubbles where Frontend delivered no uops while Backend could have accepted them. For example; stalls due to instruction-cache misses would be categorized under Frontend Bound. Sample with: FRONTEND_RETIRED.LATENCY_GE_4.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues", "MetricExpr": "( ( 5 ) * IDQ_UOPS_NOT_DELIVERED.CYCLES_0_UOPS_DELIV.CORE - INT_MISC.UOP_DROPPING ) / tma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", "MetricName": "tma_fetch_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend latency issues. For example; instruction-cache misses; iTLB misses or fetch stalls after a branch misprediction are categorized under Frontend Latency. In such cases; the Frontend eventually delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_16, FRONTEND_RETIRED.LATENCY_GE_8.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses", "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_icache_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RETIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that hit in the L2 cache", "MetricExpr": "max( 0 , tma_icache_misses - tma_code_l2_miss )", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache.", + "BriefDescription": "This metric estimates fraction of cycles the CPU was stalled due to instruction cache misses that miss in the L2 cache", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_CODE_RD / tma_info_thread_clks", "MetricGroup": "IcMiss;FetchLat;Offcore;TopdownL4;tma_L4_group;tma_icache_misses_group;Clocks_Retired", "MetricName": "tma_code_l2_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", "MetricName": "tma_itlb_misses", "ScaleUnit": "100%", + "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTEND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) ITLB was missed by instructions fetches, that later on hit in second-level TLB (STLB)", "MetricExpr": "max( 0 , tma_itlb_misses - tma_code_stlb_miss )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by instruction fetches, performing a hardware page walk", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb_misses_group;Clocks_Retired", "MetricName": "tma_code_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for (instruction) code accesses", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_4M / ( ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M )", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_code_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_miss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers", "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks + ( 10 ) * BACLEARS.ANY / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_overlap", "MetricName": "tma_branch_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES." + "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_l3_hit_latency, tma_store_latency." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueBM", "MetricName": "tma_mispredicts_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Branch Misprediction at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears", "MetricExpr": "( 1 - ( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) ) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", - "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", + "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks;tma_issueMC", "MetricName": "tma_clears_resteers", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES." + "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers as a result of Machine Clears. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears", "MetricExpr": "( 10 ) * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;tma_branch_resteers_group;Clocks", "MetricName": "tma_unknown_branches", "ScaleUnit": "100%", + "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_resteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to new branch address clears. These are fetched branches the Branch Prediction Unit was unable to recognize (e.g. first time the branch is fetched or hitting BTB capacity limit) hence called Unknown Branches. Sample with: BACLEARS.ANY." }, { - "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS).", + "BriefDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS)", "MetricExpr": "( 3 ) * IDQ.MS_SWITCHES / tma_info_thread_clks", - "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated", + "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks_Estimated;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES." + "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric estimates the fraction of cycles when the CPU was stalled due to switches of uop delivery to the Microcode Sequencer (MS). Commonly used instructions are optimized for delivery by the DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Certain operations cannot be handled natively by the execution pipeline; and must be performed by microcode (small programs injected into the execution stream). Switching to the MS too often can negatively impact performance. The MS is designated to deliver long uop flows required by CISC instructions like CPUID; or uncommon conditions like Floating Point Assists when dealing with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation." }, { - "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs).", + "BriefDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs)", "MetricExpr": "DECODE.LCP / tma_info_thread_clks", - "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_lcp", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this." + "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles CPU was stalled due to Length Changing Prefixes (LCPs). Using proper compiler flags or Intel Compiler by default will certainly avoid this. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_clks", - "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks", + "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_group;Clocks;tma_issueFB", "MetricName": "tma_dsb_switches", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS." + "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to switches from DSB to MITE pipelines. The DSB (decoded i-cache) is a Uop Cache where the front-end directly delivers Uops (micro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter latency and delivered higher bandwidth than the MITE (legacy instruction decode pipeline). Switching between the two pipelines can cause penalties hence this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DSB_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues.", + "BriefDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues", "MetricExpr": "max( 0 , tma_frontend_bound - tma_fetch_latency )", - "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots", + "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend_bound_group;Slots;tma_issueFB", "MetricName": "tma_fetch_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2.", + "MetricThreshold": "tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of slots the CPU was stalled due to Frontend bandwidth issues. For example; inefficiencies at the instruction decoders; or restrictions for caching in the DSB (decoded uops cache) are categorized under Fetch Bandwidth. In such cases; the Frontend typically delivers suboptimal amount of uops to the Backend. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATENCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline).", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline)", "MetricExpr": "( IDQ.MITE_CYCLES_ANY - IDQ.MITE_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_mite", "ScaleUnit": "100%", + "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the MITE pipeline (the legacy decode pipeline). This pipeline is used for code that was not pre-cached in the DSB or LSD. For example; inefficiencies due to asymmetric decoders; use of long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sample with: FRONTEND_RETIRED.ANY_DSB_MISS." }, { - "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder.", + "BriefDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder", "MetricExpr": "( cpu@INST_DECODED.DECODERS\\,cmask\\=0x1@ - cpu@INST_DECODED.DECODERS\\,cmask\\=0x2@ ) / tma_info_core_core_clks / 2", - "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated", + "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Slots_Estimated;tma_issueD0", "MetricName": "tma_decoder0_alone", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", + "PublicDescription": "This metric represents fraction of cycles where decoder-0 was the only active decoder. Related metrics: tma_few_uops_instructions." }, { - "BriefDescription": "This metric represents fraction of cycles where (only) 4 uops were delivered by the MITE pipeline.", + "BriefDescription": "This metric represents fraction of cycles where (only) 4 uops were delivered by the MITE pipeline", "MetricExpr": "( cpu@IDQ.MITE_UOPS\\,cmask\\=0x4@ - cpu@IDQ.MITE_UOPS\\,cmask\\=0x5@ ) / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_group;Core_Clocks", "MetricName": "tma_mite_4wide", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_mite_4wide > 0.05 & tma_mite > 0.1 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline", "MetricExpr": "( IDQ.DSB_CYCLES_ANY - IDQ.DSB_CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_dsb", "ScaleUnit": "100%", + "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to DSB (decoded uop cache) fetch pipeline. For example; inefficient utilization of the DSB cache structure or bank conflict when reading from it; are categorized here." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit", "MetricExpr": "( LSD.CYCLES_ACTIVE - LSD.CYCLES_OK ) / tma_info_core_core_clks / 2", "MetricGroup": "FetchBW;LSD;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_lsd", "ScaleUnit": "100%", + "MetricThreshold": "tma_lsd > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to LSD (Loop Stream Detector) unit. LSD typically does well sustaining Uop supply. However; in some rare cases; optimal uop-delivery could not be reached for small loops whose size (in terms of number of uops) does not suit well the LSD structure." }, { - "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details.", + "BriefDescription": "This metric represents Core fraction of cycles in which CPU was likely limited due to the Microcode Sequencer (MS) unit - see Microcode_Sequencer node for details", "MetricExpr": "cpu@IDQ.MS_UOPS\\,cmask\\=0x1@ / tma_info_core_core_clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidth_group;Slots_Estimated", "MetricName": "tma_ms", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2" }, { - "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations.", + "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations", "MetricExpr": "max( 1 - ( tma_frontend_bound + tma_backend_bound + tma_retiring ) , 0 )", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_bad_speculation", "ScaleUnit": "100%", + "MetricThreshold": "tma_bad_speculation > 0.15", "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction", "MetricExpr": "( BR_MISP_RETIRED.ALL_BRANCHES / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT ) ) * tma_bad_speculation", - "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", + "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueBM", "MetricName": "tma_branch_mispredicts", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES.", + "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction. These slots are either wasted by uops fetched from an incorrectly speculated program path; or stalls when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types).", + "BriefDescription": "This metric estimates fraction of slots the CPU was stalled due to other cases of misprediction (non-retired x86 branches or other types)", "MetricExpr": "max( tma_branch_mispredicts * ( 1 - BR_MISP_RETIRED.ALL_BRANCHES / ( INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT ) ) , 0.0001 )", "MetricGroup": "BvIO;BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mispredicts_group;Slots", "MetricName": "tma_other_mispredicts", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_mispredicts > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears", "MetricExpr": "max( 0 , tma_bad_speculation - tma_branch_mispredicts )", - "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots", + "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_group;tma_bad_speculation_group;Slots;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT.", + "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation > 0.15", + "PublicDescription": "This metric represents fraction of slots the CPU has wasted due to Machine Clears. These slots are either wasted by uops fetched prior to the clear; or stalls the out-of-order portion of the machine needs to recover its state after the clear. For example; this can happen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modifying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering.", + "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Nukes (Machine Clears) not related to memory ordering", "MetricExpr": "max( tma_machine_clears * ( 1 - MACHINE_CLEARS.MEMORY_ORDERING / MACHINE_CLEARS.COUNT ) , 0.0001 )", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_machine_clears_group;Slots", "MetricName": "tma_other_nukes", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > 0.1 & tma_bad_speculation > 0.15" }, { - "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend.", + "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend", "MetricExpr": "topdown\\-be\\-bound / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) + ( ( 5 ) * INT_MISC.CLEARS_COUNT ) / tma_info_thread_slots", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_backend_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_backend_bound > 0.2", "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck.", + "BriefDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck", "MetricExpr": "( ( CYCLE_ACTIVITY.STALLS_MEM_ANY + EXE_ACTIVITY.BOUND_ON_STORES ) / ( CYCLE_ACTIVITY.STALLS_TOTAL + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) + EXE_ACTIVITY.BOUND_ON_STORES ) ) * tma_backend_bound", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_memory_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots the Memory subsystem within the Backend was a bottleneck. Memory Bound estimates fraction of slots where pipeline is likely stalled due to demand load or store instructions. This accounts mainly for (1) non-completed in-flight memory demand loads which coincides with execution units starvation; in addition to (2) cases where stores could impose backpressure on the pipeline when many of them get buffered at the same time (less common out of the two).", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache.", + "BriefDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max( ( CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY.STALLS_L1D_MISS ) / tma_info_thread_clks , 0 )", - "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", + "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls;tma_issueL1;tma_issueMC", "MetricName": "tma_l1_bound", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT." + "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often the CPU was stalled without loads missing the L1 Data (L1D) cache. The L1D cache typically has the shortest latency. However; in certain cases like loads blocked on older stores; a load might suffer due to high latency even though it is being satisfied by the L1D. Another example is loads who miss in the TLB. These cases are characterized by execution unit stalls; while some non-completed demand load lives in the machine without having that demand load missing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses", "MetricExpr": "min( ( 7 ) * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_load", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS." + "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Translation Look-aside Buffers) are processor caches for recently used entries out of the Page Tables that are used to map virtual- to physical-addresses by the operating system. This metric approximates the potential delay of demand loads missing the first-level data TLB (assuming worst case scenario with back to back misses to different pages). This includes hitting in the second-level TLB (STLB) as well as performing a hardware page walk on an STLB miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_store." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the (first level) DTLB was missed by load accesses, that later on hit in second-level TLB (STLB)", "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Estimated", "MetricName": "tma_load_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the Second-level TLB (STLB) was missed by load accesses, performing a hardware page walk", "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_group;Clocks_Calculated", "MetricName": "tma_load_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_4K / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data load accesses", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETED_1G / ( DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_load_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores.", + "BriefDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores", "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_store_fwd_blk", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric roughly estimates fraction of cycles when the memory subsystem had loads blocked since they could not forward data from earlier (in program order) overlapping stores. To streamline memory operations in the pipeline; a load can avoid waiting for memory if a prior in-flight store is writing the data that the load wants to read (store forwarding process). However; in some cases the load may be blocked for a significant time pending the store forward. For example; when the prior store is writing a smaller region than the load is reading." }, { - "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache.", + "BriefDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache", "MetricExpr": "min( 2 * ( MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETIRED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS ) * 20 / 100 , max( CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS , 0 ) ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_l1_latency_dependency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates fraction of cycles with demand load accesses that hit the L1D cache. The short latency of the L1D cache may be exposed in pointer-chasing memory access patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations", "MetricExpr": "( 16 * max( 0 , MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS.ALL_RFO ) + ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) * ( ( 10 ) * L2_RQSTS.RFO_HIT + ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) ) / tma_info_thread_clks", - "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks", + "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks;tma_issueRFO", "MetricName": "tma_lock_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS." + "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU spent handling cache misses due to lock operations. Due to the microarchitecture handling of locks; they are classified as L1_Bound regardless of what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOADS. Related metrics: tma_store_latency." }, { - "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary.", + "BriefDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary", "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", "MetricName": "tma_split_loads", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles handling memory load split accesses - load that cross 64-byte cache line boundary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS." }, { - "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset.", + "BriefDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset", "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Estimated", "MetricName": "tma_4k_aliasing", "ScaleUnit": "100%", + "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often memory load accesses were aliased by preceding stores (in program order) with a 4K address offset. False match is possible; which incur a few cycles load re-issue. However; the short re-issue duration is often hidden by the out-of-order core and HW optimizations; hence a user may safely ignore a high value of this metric unless it manages to propagate up into parent nodes of the hierarchy (e.g. to L1_Bound)." }, { - "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed.", + "BriefDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed", "MetricExpr": "L1D_PEND_MISS.FB_FULL / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated", + "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_l1_bound_group;Clocks_Calculated;tma_issueBW;tma_issueSL;tma_issueSmSt", "MetricName": "tma_fb_full", "ScaleUnit": "100%", - "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory)." + "MetricThreshold": "tma_fb_full > 0.3", + "PublicDescription": "This metric does a *rough estimation* of how often L1D Fill Buffer unavailability limited additional L1D miss memory access requests to proceed. The higher the metric value; the deeper the memory hierarchy level the misses are satisfied from (metric values >1 are valid). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads", "MetricExpr": "( ( MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) ) ) / ( ( MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) ) ) + L1D_PEND_MISS.FB_FULL_PERIODS ) ) * ( ( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks )", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l2_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 misses/L2 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited).", + "BriefDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited)", "MetricExpr": "( 5 * tma_info_system_core_frequency ) * MEM_LOAD_RETIRED.L2_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_group;Clocks_Retired", "MetricName": "tma_l2_hit_latency", "ScaleUnit": "100%", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles with demand load accesses that hit the L2 cache under unloaded scenarios (possibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hits) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core.", + "BriefDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L2_MISS - CYCLE_ACTIVITY.STALLS_L3_MISS ) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_l3_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled due to loads accesses to L3 cache or contended with a sibling Core. Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses", "MetricExpr": "( ( ( 54 * tma_info_system_core_frequency ) - ( 5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) + ( ( 53 * tma_info_system_core_frequency ) - ( 5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_contested_accesses", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS." + "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to contested accesses. Contested accesses occur when data written by one Logical Processor are read by another Logical Processor on a different Physical Core. Examples of contested accesses include synchronizations such as locks; true data sharing such as modified locked variables; and false sharing. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses.", + "BriefDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses", "MetricExpr": "( ( 53 * tma_info_system_core_frequency ) - ( 5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD + MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * ( 1 - ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / ( OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD ) ) ) ) * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) / tma_info_thread_clks", - "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_data_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD." + "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles while the memory subsystem was handling synchronizations due to data-sharing accesses. Data shared by multiple Logical Processors (even just read shared) may cause increased access latency due to cache coherency. Excessive data sharing can drastically harm multithreaded performance. Sample with: MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited).", + "BriefDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited)", "MetricExpr": "( ( 22.5 * tma_info_system_core_frequency ) - ( 5 * tma_info_system_core_frequency ) ) * ( MEM_LOAD_RETIRED.L3_HIT * ( 1 + ( MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS ) / 2 ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks_Estimated;tma_issueLat;tma_overlap", "MetricName": "tma_l3_hit_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT." + "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles with demand load accesses that hit the L3 cache under unloaded scenarios (possibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3 hits) will improve the latency; reduce contention with sibling physical cores and increase performance. Note the value of this node may overlap with its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_bottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_store_latency." }, { - "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors).", + "BriefDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors)", "MetricExpr": "L1D_PEND_MISS.L2_STALL / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_l3_bound_group;Clocks;tma_issueBW", "MetricName": "tma_sq_full", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric measures fraction of cycles where the Super Queue (SQ) was full taking into account all request-types and both hardware SMT threads (Logical Processors). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_mem_bandwidth." }, { - "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads.", + "BriefDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads", "MetricExpr": "( CYCLE_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clks + ( ( CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS ) / tma_info_thread_clks ) - tma_l2_bound )", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_dram_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often the CPU was stalled on accesses to external memory (DRAM) by loads. Better caching can improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED.L3_MISS." }, { - "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , cpu@OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=0x4@ ) ) / tma_info_thread_clks", - "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueBW", "MetricName": "tma_mem_bandwidth", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that)." + "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the core's performance was likely hurt due to approaching bandwidth limits of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuristic assumes that a similar off-core traffic is generated by all IA cores. This metric does not aggregate non-data-read requests by this logical processor; requests from other IA Logical Processors/Physical Cores/sockets; or other non-IA devices like GPU; hence the maximum external memory bandwidth limits may or may not be approached when this metric is flagged (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full." }, { - "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM).", + "BriefDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD ) ) / tma_info_thread_clks - tma_mem_bandwidth", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks", + "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_dram_bound_group;Clocks;tma_issueLat", "MetricName": "tma_mem_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that)." + "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles where the performance was likely hurt due to latency from external memory - DRAM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from other Logical Processors/Physical Cores/sockets (see Uncore counters for that). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write.", + "BriefDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write", "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_memory_bound_group;Stalls", "MetricName": "tma_store_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates how often CPU was stalled due to RFO store memory accesses; RFO store issue a read-for-ownership request before the write. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should RFO stores be a bottleneck. Sample with: MEM_INST_RETIRED.ALL_STORES." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses.", + "BriefDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses", "MetricExpr": "( ( L2_RQSTS.RFO_HIT * ( 10 ) * ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) ) + ( 1 - ( MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES ) ) * ( min( CPU_CLK_UNHALTED.THREAD , OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO ) ) ) / tma_info_thread_clks", - "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueRFO;tma_issueSL;tma_overlap", "MetricName": "tma_store_latency", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full)." + "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates fraction of cycles the CPU spent handling L1D store misses. Store accesses usually less impact out-of-order core performance; however; holding resources for longer time can lead into undesired implications (e.g. contention on L1D fill-buffer entries - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l3_hit_latency, tma_lock_latency." }, { - "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing.", + "BriefDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing", "MetricExpr": "( 54 * tma_info_system_core_frequency ) * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSyncxn", "MetricName": "tma_false_sharing", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM." + "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates how often CPU was handling synchronizations due to False Sharing. False Sharing is a multithreading hiccup; where multiple Logical Processors contend on different data-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sharing, tma_machine_clears." }, { - "BriefDescription": "This metric represents rate of split store accesses.", + "BriefDescription": "This metric represents rate of split store accesses", "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_clks", - "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization", + "MetricGroup": "TopdownL4;tma_L4_group;tma_store_bound_group;Core_Utilization;tma_issueSpSt", "MetricName": "tma_split_stores", "ScaleUnit": "100%", + "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents rate of split store accesses. Consider aligning your data to the 64-byte cache line granularity. Sample with: MEM_INST_RETIRED.SPLIT_STORES." }, { - "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores.", + "BriefDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores", "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread_clks", - "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueSmSt", "MetricName": "tma_streaming_stores", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE." + "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric estimates how often CPU was stalled due to Streaming store memory accesses; Streaming store optimize out a read request required by RFO stores. Even though store accesses do not typically stall out-of-order CPUs; there are few cases where stores can lead to actual stalls. This metric will be flagged should Streaming stores be a bottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma_fb_full." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses.", + "BriefDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses", "MetricExpr": "( ( 7 ) * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=0x1@ + DTLB_STORE_MISSES.WALK_ACTIVE ) / tma_info_core_core_clks", - "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated", + "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_store_bound_group;Clocks_Estimated;tma_issueTLB", "MetricName": "tma_dtlb_store", "ScaleUnit": "100%", - "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES." + "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "PublicDescription": "This metric roughly estimates the fraction of cycles spent handling first-level data TLB store misses. As with ordinary data caching; focus on improving data locality and reducing working-set size to reduce DTLB overhead. Additionally; consider using profile-guided optimization (PGO) to collocate frequently-used data on the same page. Try using larger page sizes for large amounts of frequently-used data. Sample with: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_memory_data_tlbs, tma_dtlb_load." }, { - "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB).", + "BriefDescription": "This metric roughly estimates the fraction of cycles where the TLB was missed by store accesses, hitting in the second-level TLB (STLB)", "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Estimated", "MetricName": "tma_store_stlb_hit", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk.", + "BriefDescription": "This metric estimates the fraction of cycles where the STLB was missed by store accesses, performing a hardware page walk", "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_group;Clocks_Calculated", "MetricName": "tma_store_stlb_miss", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 4 KB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_4K / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_4k", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 2 or 4 MB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_2m", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses.", + "BriefDescription": "This metric estimates the fraction of cycles to walk the memory paging structures to cache translation of 1 GB pages for data store accesses", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLETED_1G / ( DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G )", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_miss_group;Clocks_Estimated", "MetricName": "tma_store_stlb_miss_1g", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb_miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck.", + "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck", "MetricExpr": "max( 0 , tma_backend_bound - tma_memory_bound )", "MetricGroup": "Backend;TmaL2;Compute;TopdownL2;tma_L2_group;tma_backend_bound_group;Slots", "MetricName": "tma_core_bound", "ScaleUnit": "100%", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck. Shortage in hardware compute resources; or dependencies in software's instructions are both categorized under Core Bound. Hence it may indicate the machine ran out of an out-of-order resource; certain execution units are overloaded or dependencies in program's data- or instruction-flow are limiting the performance (e.g. FP-chained long-latency arithmetic operations).", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Divider unit was active", "MetricExpr": "ARITH.DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_divider", "ScaleUnit": "100%", + "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles where the Divider unit was active. Divide and square root instructions are performed by the Divider unit and can take considerably longer latency than integer or Floating Point addition; subtraction; or multiplication. Sample with: ARITH.DIVIDER_ACTIVE." }, { - "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Floating-Point Divider unit was active", "MetricExpr": "ARITH.FP_DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_fp_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active.", + "BriefDescription": "This metric represents fraction of cycles where the Integer Divider unit was active", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group;Clocks", "MetricName": "tma_int_divider", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tma_core_bound > 0.1 & tma_backend_bound > 0.2" }, { - "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations.", + "BriefDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations", "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks", - "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", + "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks;tma_issueSO", "MetricName": "tma_serializing_operation", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD." + "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles the CPU issue-pipeline was stalled due to serializing operations. Instructions like CPUID; WRMSR or LFENCE serialize the out-of-order execution which may limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions.", + "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions", "MetricExpr": "140 * MISC_RETIRED.PAUSE_INST / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_group;Clocks", "MetricName": "tma_slow_pause", "ScaleUnit": "100%", + "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to PAUSE Instructions. Sample with: MISC_RETIRED.PAUSE_INST." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related).", + "BriefDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related)", "MetricExpr": "( tma_ports_utilized_0 * tma_info_thread_clks + ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) ) / tma_info_thread_clks if ( ARITH.DIVIDER_ACTIVE < ( CYCLE_ACTIVITY.STALLS_TOTAL - CYCLE_ACTIVITY.STALLS_MEM_ANY ) ) else ( EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL ) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_group;Clocks", "MetricName": "tma_ports_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric estimates fraction of cycles the CPU performance was potentially limited due to Core computation issues (non divider-related). Two distinct categories can be attributed into this metric: (1) heavy data-dependency among contiguous instructions would manifest in this metric - such cases are often referred to as low Instruction Level Parallelism (ILP). (2) Contention on some hardware execution unit other than Divider. For example; when there are too many multiply operations." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.EXE_BOUND_0_PORTS / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_0", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed no uops on any execution port (Logical Processor cycles since ICL, Physical Core cycles otherwise). Long-latency instructions like divides may contribute to this metric." }, { - "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles).", + "BriefDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles)", "MetricExpr": "UOPS_ISSUED.VECTOR_WIDTH_MISMATCH / UOPS_ISSUED.ANY", - "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks", + "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_0_group;Clocks;tma_issueMV", "MetricName": "tma_mixing_vectors", "ScaleUnit": "100%", - "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic." + "MetricThreshold": "tma_mixing_vectors > 0.05", + "PublicDescription": "This metric estimates penalty in terms of percentage of([SKL+] injected blend uops out of all Uops Issued , the Count Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigating. Read more in Appendix B1 of the Optimizations Guide for this topic. Related metrics: tma_ms_switches." }, { - "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issueL1", "MetricName": "tma_ports_utilized_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles where the CPU executed total of 1 uop per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). This can be due to heavy data-dependency among software instructions; or over oversubscribing a particular hardware resource. In some other cases with high 1_Port_Utilized and L1_Bound; this metric can point to L1 data-cache latency bottleneck that may not necessarily manifest with complete execution starvation (due to the short L1 latency e.g. walking a linked list) - looking at the assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related metrics: tma_l1_bound." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", - "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", + "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks;tma_issue2P", "MetricName": "tma_ports_utilized_2", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL." + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "PublicDescription": "This metric represents fraction of cycles CPU executed total of 2 uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization -most compilers feature auto-Vectorization options today- reduces pressure on the execution ports as multiple elements are calculated with same uop. Sample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6." }, { - "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise).", + "BriefDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise)", "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_utilization_group;Clocks", "MetricName": "tma_ports_utilized_3m", "ScaleUnit": "100%", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utilization > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", "PublicDescription": "This metric represents fraction of cycles CPU executed total of 3 or more uops per cycle on all execution ports (Logical Processor cycles since ICL, Physical Core cycles otherwise). Sample with: UOPS_EXECUTED.CYCLES_GE_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + UOPS_DISPATCHED.PORT_5 + UOPS_DISPATCHED.PORT_6 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_alu_op_utilization", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_alu_op_utilization > 0.4" }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch)", "MetricExpr": "UOPS_DISPATCHED.PORT_0 / tma_info_core_core_clks", - "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_0", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0." + "MetricThreshold": "tma_port_0 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_1 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_1", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1." + "MetricThreshold": "tma_port_1 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_5 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_5", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED.PORT_5." + "MetricThreshold": "tma_port_5 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+] ALU). Sample with: UOPS_DISPATCHED.PORT_5. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU).", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU)", "MetricExpr": "UOPS_DISPATCHED.PORT_6 / tma_info_core_core_clks", - "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks", + "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_group;Core_Clocks;tma_issue2P", "MetricName": "tma_port_6", "ScaleUnit": "100%", - "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1." + "MetricThreshold": "tma_port_6 > 0.6", + "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simple ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations", "MetricExpr": "UOPS_DISPATCHED.PORT_2_3 / ( 2 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_load_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_load_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Load operations. Sample with: UOPS_DISPATCHED.PORT_2_3." }, { - "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations.", + "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations", "MetricExpr": "( UOPS_DISPATCHED.PORT_4_9 + UOPS_DISPATCHED.PORT_7_8 ) / ( 4 * tma_info_core_core_clks )", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group;Core_Execution", "MetricName": "tma_store_op_utilization", "ScaleUnit": "100%", + "MetricThreshold": "tma_store_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution port for Store operations. Sample with: UOPS_DISPATCHED.PORT_7_8." }, { - "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired.", + "BriefDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired", "MetricExpr": "( topdown\\-retiring / ( topdown\\-fe\\-bound + topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound ) ) + ( 0 * slots )", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group;Default;Slots", "MetricName": "tma_retiring", "ScaleUnit": "100%", + "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.1", "PublicDescription": "This category represents fraction of slots utilized by useful work i.e. issued uops that eventually get retired. Ideally; all pipeline slots would be attributed to the Retiring category. Retiring of 100% would indicate the maximum Pipeline_Width throughput was achieved. Maximizing Retiring typically increases the Instructions-per-cycle (see IPC metric). Note that a high Retiring value does not necessary mean there is no room for more performance. For example; Heavy-operations or Microcode Assists are categorized under Retiring. They often indicate suboptimal performance and can often be optimized or avoided. Sample with: UOPS_RETIRED.SLOTS.", "DefaultMetricgroupName": "TopdownL1", "MetricGroupnoGroup": "TopdownL1;Default" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation).", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation)", "MetricExpr": "max( 0 , tma_retiring - tma_heavy_operations )", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_light_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring light-weight operations , instructions that require no more than one uop (micro-operation). This correlates with total number of instructions used by the program. A uops-per-instruction (see UopPI metric) ratio of 1 or less should be expected for decently optimized code running on Intel Core/Xeon products. While this often indicates efficient X86 instructions were executed; high value does not necessarily mean better performance cannot be achieved. ([ICL+] Note this may undercount due to approximation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST.", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired).", + "BriefDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired)", "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_group;Uops", "MetricName": "tma_fp_arith", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric represents overall arithmetic floating-point (FP) operations fraction the CPU has executed (retired). Note this metric's value may exceed its parent due to use of \"Uops\" CountDomain and FMA double-counting." }, { - "BriefDescription": "This metric serves as an approximation of legacy x87 usage.", + "BriefDescription": "This metric serves as an approximation of legacy x87 usage", "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.THREAD", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", "MetricName": "tma_x87_use", "ScaleUnit": "100%", + "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", "PublicDescription": "This metric serves as an approximation of legacy x87 usage. It accounts for instructions beyond X87 FP arithmetic operations; hence may be used as a thermometer to avoid X87 high usage and preferably upgrade to modern ISA. See Tip under Tuning Hint." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_scalar", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) scalar uops fraction the CPU has retired. May overcount due to FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths.", + "BriefDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths", "MetricExpr": "( FP_ARITH_INST_RETIRED.VECTOR ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic floating-point (FP) vector uops fraction the CPU has retired aggregated across all vector widths. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_128b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 128-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_256b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL." + "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 256-bit wide vectors. May overcount due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors.", + "BriefDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors", "MetricExpr": "( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / ( tma_retiring * tma_info_thread_slots )", - "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops", + "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector_group;Uops;tma_issue2P", "MetricName": "tma_fp_vector_512b", "ScaleUnit": "100%", - "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting." + "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "PublicDescription": "This metric approximates arithmetic FP vector uops fraction the CPU has retired for 512-bit wide vectors. May overcount due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring memory operations , uops for memory load or store accesses", "MetricExpr": "tma_light_operations * MEM_INST_RETIRED.ANY / INST_RETIRED.ANY", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_memory_operations", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_memory_operations > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions", "MetricExpr": "tma_light_operations * BR_INST_RETIRED.ALL_BRANCHES / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_branch_instructions", - "ScaleUnit": "100%" + "ScaleUnit": "100%", + "MetricThreshold": "tma_branch_instructions > 0.1 & tma_light_operations > 0.6" }, { - "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes.", + "BriefDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes", "MetricExpr": "max( 0 , tma_light_operations - ( tma_fp_arith + tma_memory_operations + tma_branch_instructions ) )", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operations_group;Slots", "MetricName": "tma_other_light_ops", "ScaleUnit": "100%", + "MetricThreshold": "tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents the remaining light uops fraction the CPU has executed - remaining means not covered by other sibling nodes. May undercount due to FMA double counting." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / ( tma_retiring * tma_info_thread_slots )", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_light_ops_group;Slots", "MetricName": "tma_nop_instructions", "ScaleUnit": "100%", + "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_ops > 0.3 & tma_light_operations > 0.6", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring NOP (no op) instructions. Compilers often use NOPs for certain address alignments - e.g. start address of a function or loop body. Sample with: INST_RETIRED.NOP." }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences", "MetricExpr": "tma_microcode_sequencer + tma_retiring * ( UOPS_DECODED.DEC0 - cpu@UOPS_DECODED.DEC0\\,cmask\\=0x1@ ) / IDQ.MITE_UOPS", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_group;Slots", "MetricName": "tma_heavy_operations", "ScaleUnit": "100%", + "MetricThreshold": "tma_heavy_operations > 0.1", "PublicDescription": "This metric represents fraction of slots where the CPU was retiring heavy-weight operations , instructions that require two or more uops or micro-coded sequences. This highly-correlates with the uop length of these instructions/sequences.([ICL+] Note this may overcount due to approximation using indirect events; [ADL+]).", "MetricgroupNoGroup": "TopdownL2" }, { - "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops.", + "BriefDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops", "MetricExpr": "tma_heavy_operations - tma_microcode_sequencer", - "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueD0", "MetricName": "tma_few_uops_instructions", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions." + "MetricThreshold": "tma_few_uops_instructions > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots where the CPU was retiring instructions that that are decoder into two or more uops. This highly-correlates with the number of uops in such instructions. Related metrics: tma_decoder0_alone." }, { - "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit.", + "BriefDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit", "MetricExpr": "( UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY ) * IDQ.MS_UOPS / tma_info_thread_slots", - "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots", + "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operations_group;Slots;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "ScaleUnit": "100%", - "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS." + "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", + "PublicDescription": "This metric represents fraction of slots the CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The MS is used for CISC instructions not supported by the default decoders (like repeat move strings; or CPUID); or by microcode assists used to address some operation modes (like in Floating Point assists). These cases can often be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clears, tma_ms_switches." }, { - "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists.", + "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists", "MetricExpr": "( 34 ) * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots_Estimated", "MetricName": "tma_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY." }, { - "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists.", + "BriefDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists", "MetricExpr": "34 * ASSISTS.FP / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group;Slots_Estimated", "MetricName": "tma_fp_assists", "ScaleUnit": "100%", + "MetricThreshold": "tma_fp_assists > 0.1", "PublicDescription": "This metric roughly estimates fraction of slots the CPU retired uops as a result of handing Floating Point (FP) Assists. FP Assist may apply when working with very small floating point values (so-called Denormals)." }, { - "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction.", + "BriefDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction", "MetricExpr": "max( 0 , tma_microcode_sequencer - tma_assists )", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_group;Slots", "MetricName": "tma_cisc", "ScaleUnit": "100%", + "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.05 & tma_heavy_operations > 0.1", "PublicDescription": "This metric estimates fraction of cycles the CPU retired uops originated from CISC (complex instruction set computer) instruction. A CISC instruction has multiple uops that are required to perform the instruction's functionality as in the case of read-modify-write as an example. Since these instructions require multiple uops they may or may not imply sub-optimal use of machine resources." }, { - "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts.", + "BriefDescription": "Probability of Core Bound bottleneck hidden by SMT-profiling artifacts", "MetricExpr": "100 * ( 1 - tma_core_bound / tma_ports_utilization if tma_core_bound < tma_ports_utilization else 1 ) if tma_info_system_smt_2t_utilization > 0.5 else 0", "MetricGroup": "Cor;SMT;Metric", - "MetricName": "tma_info_botlnk_l0_core_bound_likely" + "MetricName": "tma_info_botlnk_l0_core_bound_likely", + "MetricThreshold": "tma_info_botlnk_core_bound_likely > 0.5" }, { - "BriefDescription": "Instructions Per Cycle (per Logical Processor).", + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", "MetricExpr": "INST_RETIRED.ANY / tma_info_thread_clks", "MetricGroup": "Ret;Summary;Metric", "MetricName": "tma_info_thread_ipc" }, { - "BriefDescription": "Uops Per Instruction.", + "BriefDescription": "Uops Per Instruction", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire;Metric", - "MetricName": "tma_info_thread_uoppi" + "MetricName": "tma_info_thread_uoppi", + "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { - "BriefDescription": "Uops per taken branch.", + "BriefDescription": "Uops per taken branch", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Metric", - "MetricName": "tma_info_thread_uptb" + "MetricName": "tma_info_thread_uptb", + "MetricThreshold": "tma_info_thread_uptb < 5 * 1.5" }, { - "BriefDescription": "Cycles Per Instruction (per Logical Processor).", + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Pipeline;Mem;Metric", "MetricName": "tma_info_thread_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline;Count", "MetricName": "tma_info_thread_clks" }, { - "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward).", + "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", "MetricExpr": "slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_thread_slots" }, { - "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor.", + "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor", "MetricExpr": "tma_info_thread_slots / ( TOPDOWN.SLOTS / 2 ) if #SMT_on else 1", "MetricGroup": "SMT;TmaL1;TopdownL1;tma_L1_group;Metric", "MetricName": "tma_info_thread_slots_utilization" }, { - "BriefDescription": "The ratio of Executed- by Issued-Uops.", + "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline;Metric", "MetricName": "tma_info_thread_execute_per_issue", "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." }, { - "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core).", + "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", "MetricExpr": "INST_RETIRED.ANY / tma_info_core_core_clks", "MetricGroup": "Ret;SMT;TmaL1;TopdownL1;tma_L1_group;Core_Metric", "MetricName": "tma_info_core_coreipc" }, { - "BriefDescription": "Floating Point Operations Per Cycle.", + "BriefDescription": "Floating Point Operations Per Cycle", "MetricExpr": "( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / tma_info_core_core_clks", "MetricGroup": "Ret;Flops;Core_Metric", "MetricName": "tma_info_core_flopc" }, { - "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width).", + "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Cor;Flops;HPC;Core_Metric", "MetricName": "tma_info_core_fp_arith_utilization", "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." }, { - "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor).", + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per thread (logical-processor)", "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil;Metric", "MetricName": "tma_info_core_ilp" }, { - "BriefDescription": "uops Executed per Cycle.", + "BriefDescription": "uops Executed per Cycle", "MetricExpr": "UOPS_EXECUTED.THREAD / tma_info_thread_clks", "MetricGroup": "Power;Metric", "MetricName": "tma_info_core_epc" }, { - "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core.", + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED if #SMT_on else tma_info_thread_clks", "MetricGroup": "SMT;Count", "MetricName": "tma_info_core_core_clks" }, { - "BriefDescription": "Instructions per Load (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipload" + "MetricName": "tma_info_inst_mix_ipload", + "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { - "BriefDescription": "Instructions per Store (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", "MetricGroup": "InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipstore" + "MetricName": "tma_info_inst_mix_ipstore", + "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { - "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipbranch" + "MetricName": "tma_info_inst_mix_ipbranch", + "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { - "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipcall" + "MetricName": "tma_info_inst_mix_ipcall", + "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { - "BriefDescription": "Instructions per taken branch.", + "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", - "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric", - "MetricName": "tma_info_inst_mix_iptb" + "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;Inst_Metric;tma_issueFB", + "MetricName": "tma_info_inst_mix_iptb", + "MetricThreshold": "tma_info_inst_mix_iptb < 5 * 2 + 1", + "PublicDescription": "Instructions per taken branch. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp." }, { - "BriefDescription": "Branch instructions per taken branch.", + "BriefDescription": "Branch instructions per taken branch", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;PGO;Metric", "MetricName": "tma_info_inst_mix_bptkbranch" }, { - "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", "MetricGroup": "Flops;InsType;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipflop" + "MetricName": "tma_info_inst_mix_ipflop", + "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { - "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( ( FP_ARITH_INST_RETIRED.SCALAR ) + ( FP_ARITH_INST_RETIRED.VECTOR ) )", "MetricGroup": "Flops;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith", + "MetricThreshold": "tma_info_inst_mix_iparith < 10", "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting. Approximated prior to BDW." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", "MetricGroup": "Flops;FpScalar;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx128", + "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx256", + "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_iparith_avx512", + "MetricThreshold": "tma_info_inst_mix_iparith_avx512 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate). Values < 1 are possible due to intentional FMA double counting." }, { - "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per PAUSE (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / MISC_RETIRED.PAUSE_INST", "MetricGroup": "Flops;FpVector;InsType;Inst_Metric", "MetricName": "tma_info_inst_mix_ippause" }, { - "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / SW_PREFETCH_ACCESS.ANY", "MetricGroup": "Prefetches;Inst_Metric", - "MetricName": "tma_info_inst_mix_ipswpf" + "MetricName": "tma_info_inst_mix_ipswpf", + "MetricThreshold": "tma_info_inst_mix_ipswpf < 100" }, { - "BriefDescription": "Total number of retired Instructions.", + "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", "MetricGroup": "Summary;TmaL1;TopdownL1;tma_L1_group;Count", "MetricName": "tma_info_inst_mix_instructions", "PublicDescription": "Total number of retired Instructions. Sample with: INST_RETIRED.PREC_DIST." }, { - "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired.", + "BriefDescription": "Average number of Uops retired in cycles where at least one uop has retired", "MetricExpr": "( tma_retiring * tma_info_thread_slots ) / cpu@UOPS_RETIRED.SLOTS\\,cmask\\=0x1@", "MetricGroup": "Pipeline;Ret;Metric", "MetricName": "tma_info_pipeline_retire" }, { - "BriefDescription": "Instructions per a microcode Assist invocation.", + "BriefDescription": "Instructions per a microcode Assist invocation", "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire;Inst_Metric", "MetricName": "tma_info_pipeline_ipassist", + "MetricThreshold": "tma_info_pipeline_ipassist < 100000", "PublicDescription": "Instructions per a microcode Assist invocation. See Assists tree node for details (lower number means higher occurrence rate)." }, { - "BriefDescription": ".", + "BriefDescription": "", "MetricExpr": "UOPS_EXECUTED.THREAD / ( ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=0x1@ )", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT;Metric", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Average number of uops fetched from LSD per cycle.", + "BriefDescription": "Average number of uops fetched from LSD per cycle", "MetricExpr": "LSD.UOPS / LSD.CYCLES_ACTIVE", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_lsd" }, { - "BriefDescription": "Average number of uops fetched from DSB per cycle.", + "BriefDescription": "Average number of uops fetched from DSB per cycle", "MetricExpr": "IDQ.DSB_UOPS / IDQ.DSB_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_dsb" }, { - "BriefDescription": "Average number of uops fetched from MITE per cycle.", + "BriefDescription": "Average number of uops fetched from MITE per cycle", "MetricExpr": "IDQ.MITE_UOPS / IDQ.MITE_CYCLES_ANY", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_pipeline_fetch_mite" }, { - "BriefDescription": "Average number of Uops issued by front-end when it issued something.", + "BriefDescription": "Average number of Uops issued by front-end when it issued something", "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=0x1@", "MetricGroup": "Fed;FetchBW;Metric", "MetricName": "tma_info_frontend_fetch_upc" }, { - "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache).", + "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache)", "MetricExpr": "LSD.UOPS / ( UOPS_ISSUED.ANY )", "MetricGroup": "Fed;LSD;Metric", "MetricName": "tma_info_frontend_lsd_coverage" }, { - "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache).", + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", "MetricExpr": "IDQ.DSB_UOPS / ( UOPS_ISSUED.ANY )", - "MetricGroup": "DSB;Fed;FetchBW;Metric", - "MetricName": "tma_info_frontend_dsb_coverage" + "MetricGroup": "DSB;Fed;FetchBW;Metric;tma_issueFB", + "MetricName": "tma_info_frontend_dsb_coverage", + "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & ipc / 5 > 0.35", + "PublicDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", + "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "DSBmiss;Metric", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { - "BriefDescription": "Taken Branches retired Per Cycle.", + "BriefDescription": "Taken Branches retired Per Cycle", "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", "MetricGroup": "Branches;FetchBW;Metric", "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Average Latency for L1 instruction cache misses.", + "BriefDescription": "Average Latency for L1 instruction cache misses", "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STALL\\,cmask\\=0x1\\,edge\\=0x1@", "MetricGroup": "Fed;FetchLat;IcMiss;Metric", "MetricName": "tma_info_frontend_icache_miss_latency" }, { - "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", "MetricGroup": "DSBmiss;Fed;Inst_Metric", - "MetricName": "tma_info_frontend_ipdsb_miss_ret" + "MetricName": "tma_info_frontend_ipdsb_miss_ret", + "MetricThreshold": "tma_info_frontend_ipdsb_miss_ret < 50" }, { - "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per speculative Unknown Branch Misprediction (BAClear) (lower number means higher occurrence rate)", "MetricExpr": "tma_info_inst_mix_instructions / BACLEARS.ANY", "MetricGroup": "Fed;Metric", "MetricName": "tma_info_frontend_ipunknown_branch" }, { - "BriefDescription": "L2 cache true code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache true code cacheline misses per kilo instruction", "MetricExpr": "1000 * FRONTEND_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code" }, { - "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction.", + "BriefDescription": "L2 cache speculative code cacheline misses per kilo instruction", "MetricExpr": "1000 * L2_RQSTS.CODE_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "IcMiss;Metric", "MetricName": "tma_info_frontend_l2mpki_code_all" }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_dsb_switches / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) + tma_fetch_bandwidth * tma_mite / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) )", - "MetricGroup": "DSBmiss;Fed;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_misses" + "MetricGroup": "DSBmiss;Fed;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_misses", + "MetricThreshold": "tma_info_botlnk_dsb_misses > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) misses - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck.", + "BriefDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck", "MetricExpr": "100 * ( tma_frontend_bound * ( tma_fetch_bandwidth / ( tma_fetch_latency + tma_fetch_bandwidth ) ) * ( tma_dsb / ( tma_mite + tma_dsb + tma_lsd + tma_ms ) ) )", - "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_dsb_bandwidth" + "MetricGroup": "DSB;Fed;FetchBW;Scaled_Slots;tma_issueFB", + "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", + "MetricThreshold": "tma_info_botlnk_dsb_bandwidth > 10", + "PublicDescription": "Total pipeline cost of DSB (uop cache) hits - subset of the Instruction_Fetch_BW Bottleneck. Related metrics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp." }, { - "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck.", + "BriefDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck", "MetricExpr": "100 * ( tma_fetch_latency * tma_icache_misses / ( tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches ) )", - "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots", - "MetricName": "tma_info_botlnk_l2_ic_misses" + "MetricGroup": "Fed;FetchLat;IcMiss;Scaled_Slots;tma_issueFL", + "MetricName": "tma_info_botlnk_l2_ic_misses", + "MetricThreshold": "tma_info_botlnk_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache misses - subset of the Big_Code Bottleneck." }, { - "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate).", + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmispredict" + "MetricName": "tma_info_bad_spec_ipmispredict", + "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional non-taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken" + "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", + "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for conditional taken branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_cond_taken" + "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", + "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for return branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_ret" + "MetricName": "tma_info_bad_spec_ipmisp_ret", + "MetricThreshold": "tma_info_bad_spec_ipmisp_ret < 500" }, { - "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indirect CALL or JMP branches (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts;Inst_Metric", - "MetricName": "tma_info_bad_spec_ipmisp_indirect" + "MetricName": "tma_info_bad_spec_ipmisp_indirect", + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" }, { - "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear).", + "BriefDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear)", "MetricExpr": "tma_mispredictions * tma_info_thread_slots / ( 5 ) / BR_MISP_RETIRED.ALL_BRANCHES / 100", - "MetricGroup": "Bad;BrMispredicts;Core_Metric", - "MetricName": "tma_info_bad_spec_branch_misprediction_cost" + "MetricGroup": "Bad;BrMispredicts;Core_Metric;tma_issueBM", + "MetricName": "tma_info_bad_spec_branch_misprediction_cost", + "PublicDescription": "Branch Misprediction Cost: Cycles representing fraction of TMA slots wasted per non-speculative branch misprediction (retired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_mispredicts_resteers." }, { - "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes).", + "BriefDescription": "Speculative to Retired ratio of all clears (covering Mispredicts and nukes)", "MetricExpr": "INT_MISC.CLEARS_COUNT / ( BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT )", "MetricGroup": "BrMispredicts;Metric", "MetricName": "tma_info_bad_spec_spec_clears_ratio" }, { - "BriefDescription": "Fraction of branches that are non-taken conditionals.", + "BriefDescription": "Fraction of branches that are non-taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_nt" }, { - "BriefDescription": "Fraction of branches that are taken conditionals.", + "BriefDescription": "Fraction of branches that are taken conditionals", "MetricExpr": "BR_INST_RETIRED.COND_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;CodeGen;PGO;Fraction", "MetricName": "tma_info_branches_cond_tk" }, { - "BriefDescription": "Fraction of branches that are CALL or RET.", + "BriefDescription": "Fraction of branches that are CALL or RET", "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_callret" }, { - "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps.", + "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", "MetricExpr": "( BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN - 2 * BR_INST_RETIRED.NEAR_CALL ) / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_jump" }, { - "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group).", + "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group)", "MetricExpr": "1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", "MetricGroup": "Bad;Branches;Fraction", "MetricName": "tma_info_branches_other_branches" }, { - "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles).", + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_RETIRED.L1_MISS + MEM_LOAD_RETIRED.FB_HIT )", "MetricGroup": "Mem;MemoryBound;MemoryLat;Clocks_Latency", "MetricName": "tma_info_memory_load_miss_real_latency" }, { - "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss.", + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss", "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", "MetricGroup": "Mem;MemoryBound;MemoryBW;Metric", "MetricName": "tma_info_memory_mlp", "PublicDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)." }, { - "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki" }, { - "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l1mpki_load" }, { - "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Backend;CacheHits;Metric", "MetricName": "tma_info_memory_l2mpki" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_all" }, { - "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2mpki_load" }, { - "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs.", + "BriefDescription": "Offcore requests (L2 cache miss) per kilo instruction for demand RFOs", "MetricExpr": "1000 * L2_RQSTS.RFO_MISS / INST_RETIRED.ANY", "MetricGroup": "CacheMisses;Offcore;Metric", "MetricName": "tma_info_memory_l2mpki_rfo" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_all" }, { - "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative).", + "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_l2hpki_load" }, { - "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads.", + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_l3mpki" }, { - "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries).", + "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", "MetricGroup": "CacheHits;Mem;Metric", "MetricName": "tma_info_memory_fb_hpki" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "64 * L1D.REPLACEMENT / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "64 * L2_LINES_IN.ALL / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Metric", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, { - "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Mem;MemoryBW;Offcore;Metric", "MetricName": "tma_info_memory_l3_cache_access_bw" }, { - "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses.", + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 2 * tma_info_core_core_clks )", "MetricGroup": "Mem;MemoryTLB;Core_Metric", - "MetricName": "tma_info_memory_tlb_page_walks_utilization" + "MetricName": "tma_info_memory_tlb_page_walks_utilization", + "MetricThreshold": "tma_info_memory_page_walks_utilization > 0.5" }, { - "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) code speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * ITLB_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Fed;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_code_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data load speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_LOAD_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_load_stlb_mpki" }, { - "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk).", + "BriefDescription": "STLB (2nd level TLB) data store speculative misses per kilo instruction (misses of any page-size that complete the page walk)", "MetricExpr": "1000 * DTLB_STORE_MISSES.WALK_COMPLETED / INST_RETIRED.ANY", "MetricGroup": "Mem;MemoryTLB;Metric", "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", "MetricExpr": "tma_info_memory_l1d_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l1d_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", "MetricExpr": "tma_info_memory_l2_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l2_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_fill_bw", "MetricGroup": "Mem;MemoryBW;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_fill_bw_2t" }, { - "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec].", + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", "MetricExpr": "tma_info_memory_l3_cache_access_bw", "MetricGroup": "Mem;MemoryBW;Offcore;Core_Metric", "MetricName": "tma_info_memory_core_l3_cache_access_bw_2t" }, { - "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses.", + "BriefDescription": "Rate of L2 HW prefetched lines that were not used by demand accesses", "MetricExpr": "L2_LINES_OUT.USELESS_HWPF / ( L2_LINES_OUT.SILENT + L2_LINES_OUT.NON_SILENT )", "MetricGroup": "Prefetches;Metric", - "MetricName": "tma_info_memory_prefetches_useless_hwpf" + "MetricName": "tma_info_memory_prefetches_useless_hwpf", + "MetricThreshold": "tma_info_memory_useless_hwpf > 0.15" }, { - "BriefDescription": "Average Latency for L2 cache miss demand Loads.", + "BriefDescription": "Average Latency for L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCORE_REQUESTS.DEMAND_DATA_RD", "MetricGroup": "LockCont;Memory_Lat;Offcore;Clocks_Latency", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { - "BriefDescription": "Average Parallel L2 cache miss demand Loads.", + "BriefDescription": "Average Parallel L2 cache miss demand Loads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=0x1@", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, { - "BriefDescription": "Average Parallel L2 cache miss data reads.", + "BriefDescription": "Average Parallel L2 cache miss data reads", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD / OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD", "MetricGroup": "Memory_BW;Offcore;Metric", "MetricName": "tma_info_memory_latency_data_l2_mlp" }, { - "BriefDescription": "Un-cacheable retired load per kilo instruction.", + "BriefDescription": "Un-cacheable retired load per kilo instruction", "MetricExpr": "1000 * MEM_LOAD_MISC_RETIRED.UC / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_uc_load_pki" }, { - "BriefDescription": "\"Bus lock\" per kilo instruction.", + "BriefDescription": "\"Bus lock\" per kilo instruction", "MetricExpr": "1000 * SQ_MISC.BUS_LOCK / INST_RETIRED.ANY", "MetricGroup": "Mem;Metric", "MetricName": "tma_info_memory_mix_bus_lock_pki" }, { - "BriefDescription": "Average CPU Utilization (percentage).", + "BriefDescription": "Average CPU Utilization (percentage)", "MetricExpr": "tma_info_system_cpus_utilized / #num_cpus_online", "MetricGroup": "HPC;Summary;Metric", "MetricName": "tma_info_system_cpu_utilization" }, { - "BriefDescription": "Average number of utilized CPUs.", + "BriefDescription": "Average number of utilized CPUs", "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricGroup": "Summary;Metric", "MetricName": "tma_info_system_cpus_utilized" }, { - "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz].", + "BriefDescription": "Measured Average Core Frequency for unhalted processors [GHz]", "MetricExpr": "tma_info_system_turbo_utilization * TSC / ( 1000000000 ) / tma_info_system_time", "MetricGroup": "Summary;Power;System_Metric", "MetricName": "tma_info_system_core_frequency" }, { - "BriefDescription": "Giga Floating Point Operations Per Second.", + "BriefDescription": "Giga Floating Point Operations Per Second", "MetricExpr": "( ( FP_ARITH_INST_RETIRED.SCALAR + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_FLOPS + 8 * FP_ARITH_INST_RETIRED.8_FLOPS + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / ( 1000000000 ) ) / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC;Metric", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width." }, { - "BriefDescription": "Average Frequency Utilization relative nominal frequency.", + "BriefDescription": "Average Frequency Utilization relative nominal frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0.", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0", "MetricExpr": "CORE_POWER.LVL0_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license0_utilization", "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0. This includes non-AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes." }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1.", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1", "MetricExpr": "CORE_POWER.LVL1_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license1_utilization", + "MetricThreshold": "tma_info_system_power_license1_utilization > 0.5", "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1. This includes high current AVX 256-bit instructions as well as low current AVX 512-bit instructions." }, { - "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX).", + "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX)", "MetricExpr": "CORE_POWER.LVL2_TURBO_LICENSE / tma_info_core_core_clks", "MetricGroup": "Power;Core_Metric", "MetricName": "tma_info_system_power_license2_utilization", + "MetricThreshold": "tma_info_system_power_license2_utilization > 0.5", "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX). This includes high current AVX 512-bit instructions." }, { - "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active.", + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED if #SMT_on else 0", "MetricGroup": "SMT;Core_Metric", "MetricName": "tma_info_system_smt_2t_utilization" }, { - "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode.", + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "OS;Metric", - "MetricName": "tma_info_system_kernel_utilization" + "MetricName": "tma_info_system_kernel_utilization", + "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { - "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode.", + "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS;Metric", "MetricName": "tma_info_system_kernel_cpi", "ScaleUnit": "1per_instr" }, { - "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec].", + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", "MetricExpr": "64 * ( UNC_ARB_TRK_REQUESTS.ALL + UNC_ARB_COH_TRK_REQUESTS.ALL ) / ( 1000000 ) / tma_info_system_time / 1000", - "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec", - "MetricName": "tma_info_system_dram_bw_use" + "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;GB/sec;tma_issueBW", + "MetricName": "tma_info_system_dram_bw_use", + "PublicDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]. Related metrics: tma_bottleneck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth, tma_sq_full." }, { - "BriefDescription": "Total package Power in Watts.", + "BriefDescription": "Total package Power in Watts", "MetricExpr": "power@energy\\-pkg@ * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "MetricGroup": "Power;SoC;System_Metric", "MetricName": "tma_info_system_power" }, { - "BriefDescription": "Run duration time in seconds.", + "BriefDescription": "Run duration time in seconds", "MetricExpr": "duration_time", "MetricGroup": "Summary;Seconds", - "MetricName": "tma_info_system_time" + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" }, { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator.", + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Summary;Clocks", - "MetricName": "tma_info_system_mux" + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mux < 0.9" }, { - "BriefDescription": "Socket actual clocks when any core is active on that socket.", + "BriefDescription": "Socket actual clocks when any core is active on that socket", "MetricExpr": "UNC_CLOCK.SOCKET", "MetricGroup": "SoC;Count", "MetricName": "tma_info_system_socket_clks" }, { - "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate].", + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS;Inst_Metric", - "MetricName": "tma_info_system_ipfarbranch" + "MetricName": "tma_info_system_ipfarbranch", + "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" } ] \ No newline at end of file diff --git a/TGL/metrics/tigerlake_metrics.json b/TGL/metrics/tigerlake_metrics.json index 71483349..7227dca9 100644 --- a/TGL/metrics/tigerlake_metrics.json +++ b/TGL/metrics/tigerlake_metrics.json @@ -2,8 +2,8 @@ "Header": { "Copyright": "Copyright (c) 2001 - 2024 Intel Corporation. All rights reserved.", "Info": "Performance Monitoring Metrics for 11th Generation Intel(R) Core(TM) Processor0", - "DatePublished": "11/12/2024", - "Version": "0", + "DatePublished": "11/15/2024", + "Version": "1.0", "Legend": "", "TmaVersion": "5.01", "TmaFlavor": "Full" @@ -107,7 +107,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Mispredictions > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Mispredictions" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Mispredictions > 20", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP", @@ -171,7 +179,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Big_Code > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Big_Code" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Big_Code > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBC;BigFootprint;Fed;Frontend;IcMiss;MemoryTLB", @@ -332,7 +348,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Instruction_Fetch_BW" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Instruction_Fetch_BW > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", @@ -549,7 +573,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Bandwidth > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore", @@ -798,7 +830,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Cache_Memory_Latency" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Cache_Memory_Latency > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;Mem;MemoryLat;Offcore", @@ -1011,7 +1051,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Data_TLBs" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Data_TLBs > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore", @@ -1212,7 +1260,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Memory_Synchronization > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Memory_Synchronization" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Memory_Synchronization > 10", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;LockCont;Mem;Offcore", @@ -1296,7 +1352,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Compute_Bound_Est" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Compute_Bound_Est > 20", + "ThresholdIssues": "$issueComp" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;Cor", @@ -1497,7 +1561,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Irregular_Overhead > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Irregular_Overhead" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Bottleneck_Irregular_Overhead > 10", + "ThresholdIssues": "$issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BvIO;Cor;Ret", @@ -1882,7 +1954,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Other_Bottlenecks" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Other_Bottlenecks > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;Cor;Offcore", @@ -1918,7 +1998,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Branching_Overhead > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Branching_Overhead" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Bottleneck_Branching_Overhead > 5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Ret", @@ -1998,7 +2086,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Bottleneck_Useful_Work > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bottleneck_Useful_Work" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Bottleneck_Useful_Work > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;Ret", @@ -2042,7 +2138,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvFB;BvIO;TmaL1;PGO", @@ -2075,7 +2179,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Frontend;TmaL2", @@ -2104,7 +2220,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss", @@ -2137,7 +2269,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Hit(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -2166,7 +2318,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_L2_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ICache_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_L2_Miss(%) > 5 & metric_TMA_....ICache_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss;FetchLat;Offcore", @@ -2195,7 +2367,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB", @@ -2228,7 +2416,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Hit(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2257,7 +2465,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2294,7 +2522,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_4K(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2331,7 +2583,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Code_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......Code_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....ITLB_Misses(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 10 & e > 15", + "BaseFormula": "metric_TMA_........Code_STLB_Miss_2M(%) > 5 & metric_TMA_......Code_STLB_Miss(%) > 5 & metric_TMA_....ITLB_Misses(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MemoryTLB", @@ -2364,7 +2640,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -2401,7 +2693,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Mispredicts_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Mispredicts_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP", @@ -2438,7 +2750,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Clears_Resteers(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Clears_Resteers(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;MachineClears", @@ -2467,7 +2799,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Unknown_Branches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Branch_Resteers(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 15", + "BaseFormula": "metric_TMA_......Unknown_Branches(%) > 5 & metric_TMA_....Branch_Resteers(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BigFootprint;BvBC;FetchLat", @@ -2496,7 +2848,23 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....MS_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueMC, $issueMS, $issueMV, $issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat;MicroSeq", @@ -2525,7 +2893,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LCP(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....LCP(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchLat", @@ -2554,7 +2938,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB_Switches(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Latency(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Frontend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....DSB_Switches(%) > 5 & metric_TMA_..Fetch_Latency(%) > 10 & metric_TMA_Frontend_Bound(%) > 15", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchLat", @@ -2603,7 +3003,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;Frontend;TmaL2", @@ -2649,7 +3057,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -2695,7 +3115,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Decoder0_Alone(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_......Decoder0_Alone(%) > 10 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -2728,7 +3164,23 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_......MITE_4wide(%) > 5 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MITE_4wide(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....MITE(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_......MITE_4wide(%) > 5 & metric_TMA_....MITE(%) > 10 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;FetchBW", @@ -2774,7 +3226,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DSB(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....DSB(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;FetchBW", @@ -2820,7 +3284,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....LSD(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....LSD(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 15 & b > 20", + "BaseFormula": "metric_TMA_....LSD(%) > 15 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "FetchBW;LSD", @@ -2862,7 +3338,19 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....MS(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Fetch_Bandwidth(%)" + } + ], + "Formula": "a > 5 & b > 20", + "BaseFormula": "metric_TMA_....MS(%) > 5 & metric_TMA_..Fetch_Bandwidth(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -2910,7 +3398,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 15", + "BaseFormula": "metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", @@ -2967,7 +3463,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueBM" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2", @@ -3024,7 +3532,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Mispredicts(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Branch_Mispredicts(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Mispredicts(%) > 5 & metric_TMA_..Branch_Mispredicts(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;BrMispredicts", @@ -3081,7 +3605,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 10 & b > 15", + "BaseFormula": "metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "$issueMC, $issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2", @@ -3142,7 +3678,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Nukes(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Machine_Clears(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Bad_Speculation(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 15", + "BaseFormula": "metric_TMA_....Other_Nukes(%) > 5 & metric_TMA_..Machine_Clears(%) > 10 & metric_TMA_Bad_Speculation(%) > 15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;Machine_Clears", @@ -3186,7 +3738,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20", + "BaseFormula": "metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvOB;TmaL1", @@ -3251,7 +3811,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20", + "BaseFormula": "metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2", @@ -3284,7 +3856,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1, $issueMC" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -3325,7 +3913,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -3366,7 +3974,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3395,7 +4027,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3436,7 +4092,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_4K(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3477,7 +4161,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_2M(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3518,7 +4230,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Load_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Load_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Load(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 10 & d > 10 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Load_STLB_Miss_1G(%) > 5 & metric_TMA_........Load_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Load(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -3547,7 +4287,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Fwd_Blk(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Fwd_Blk(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3597,7 +4357,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L1_Latency_Dependency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L1_Latency_Dependency(%) > 10 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -3642,7 +4422,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Lock_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Lock_Latency(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Offcore", @@ -3683,7 +4483,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......Split_Loads(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Loads(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......Split_Loads(%) > 30", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3712,7 +4520,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......4K_Aliasing(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......4K_Aliasing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L1_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......4K_Aliasing(%) > 20 & metric_TMA_....L1_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -3741,7 +4569,15 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_......FB_Full(%) > 30" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FB_Full(%)" + } + ], + "Formula": "a > 30", + "BaseFormula": "metric_TMA_......FB_Full(%) > 30", + "ThresholdIssues": "$issueBW, $issueSL, $issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW", @@ -3790,7 +4626,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem", @@ -3840,7 +4692,27 @@ "Category": "TMA", "CountDomain": "Clocks_Retired", "Threshold": { - "Formula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L2_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L2_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L2_Hit_Latency(%) > 5 & metric_TMA_....L2_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryLat", @@ -3873,7 +4745,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem", @@ -3935,7 +4823,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Contested_Accesses(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Contested_Accesses(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -3997,7 +4905,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Data_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Data_Sharing(%) > 5 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;Offcore;Snoop", @@ -4047,7 +4975,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......L3_Hit_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......L3_Hit_Latency(%) > 10 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat", @@ -4076,7 +5024,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......SQ_Full(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....L3_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 30 & b > 5 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......SQ_Full(%) > 30 & metric_TMA_....L3_Bound(%) > 5 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -4129,7 +5097,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -4158,7 +5142,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Bandwidth(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Bandwidth(%) > 20 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueBW" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMB;MemoryBW;Offcore", @@ -4191,7 +5195,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......MEM_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....DRAM_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......MEM_Latency(%) > 10 & metric_TMA_....DRAM_Bound(%) > 10 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueLat" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;MemoryLat;Offcore", @@ -4220,7 +5244,23 @@ "Category": "TMA", "CountDomain": "Stalls", "Threshold": { - "Formula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20", + "BaseFormula": "metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBound;TmaL3mem", @@ -4261,7 +5301,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Store_Latency(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Store_Latency(%) > 10 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueRFO, $issueSL, ~overlap" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore", @@ -4303,7 +5363,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......False_Sharing(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......False_Sharing(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSyncxn" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop", @@ -4345,7 +5425,27 @@ "Category": "TMA", "CountDomain": "Core_Utilization", "Threshold": { - "Formula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Split_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Split_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSpSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4374,7 +5474,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Streaming_Stores(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......Streaming_Stores(%) > 20 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSmSt" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryBW;Offcore", @@ -4420,7 +5540,27 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 20 & c > 20 & d > 20", + "BaseFormula": "metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueTLB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvMT;MemoryTLB", @@ -4466,7 +5606,31 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Hit(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Hit(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4508,7 +5672,31 @@ "Category": "TMA", "CountDomain": "Clocks_Calculated", "Threshold": { - "Formula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 20 & d > 20 & e > 20", + "BaseFormula": "metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4562,7 +5750,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_4K(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_4K(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4616,7 +5832,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_2M(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_2M(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4670,7 +5914,35 @@ "Category": "TMA", "CountDomain": "Clocks_Estimated", "Threshold": { - "Formula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Store_STLB_Miss_1G(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_........Store_STLB_Miss(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_......DTLB_Store(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_....Store_Bound(%)" + }, + { + "Alias": "e", + "Value": "metric_TMA_..Memory_Bound(%)" + }, + { + "Alias": "f", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 5 & c > 5 & d > 20 & e > 20 & f > 20", + "BaseFormula": "metric_TMA_..........Store_STLB_Miss_1G(%) > 5 & metric_TMA_........Store_STLB_Miss(%) > 5 & metric_TMA_......DTLB_Store(%) > 5 & metric_TMA_....Store_Bound(%) > 20 & metric_TMA_..Memory_Bound(%) > 20 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MemoryTLB", @@ -4735,7 +6007,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 20", + "BaseFormula": "metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;TmaL2;Compute", @@ -4764,7 +6048,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB", @@ -4793,7 +6093,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......FP_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4826,7 +6146,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......INT_Divider(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Divider(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 20 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......INT_Divider(%) > 20 & metric_TMA_....Divider(%) > 20 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4855,7 +6195,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueSO" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO;PortsUtil", @@ -4884,7 +6240,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Slow_Pause(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Serializing_Operation(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 5 & b > 10 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Slow_Pause(%) > 5 & metric_TMA_....Serializing_Operation(%) > 10 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -4949,7 +6325,23 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 10 & c > 20", + "BaseFormula": "metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -4978,7 +6370,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_0(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_0(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -5007,7 +6419,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_........Mixing_Vectors(%) > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Mixing_Vectors(%)" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_........Mixing_Vectors(%) > 5", + "ThresholdIssues": "$issueMV" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5036,7 +6456,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_1(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 20 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_1(%) > 20 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issueL1" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -5065,7 +6505,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_2(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 15 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_2(%) > 15 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "PortsUtil", @@ -5094,7 +6554,27 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Ports_Utilized_3m(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Ports_Utilization(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Core_Bound(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_Backend_Bound(%)" + } + ], + "Formula": "a > 40 & b > 15 & c > 10 & d > 20", + "BaseFormula": "metric_TMA_......Ports_Utilized_3m(%) > 40 & metric_TMA_....Ports_Utilization(%) > 15 & metric_TMA_..Core_Bound(%) > 10 & metric_TMA_Backend_Bound(%) > 20", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvCB;PortsUtil", @@ -5148,7 +6628,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........ALU_Op_Utilization(%) > 40" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........ALU_Op_Utilization(%)" + } + ], + "Formula": "a > 40", + "BaseFormula": "metric_TMA_........ALU_Op_Utilization(%) > 40", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5190,7 +6678,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_0(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_0(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_0(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -5232,7 +6728,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_1(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_1(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_1(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5274,7 +6778,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_5(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_5(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_5(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5316,7 +6828,15 @@ "Category": "TMA", "CountDomain": "Core_Clocks", "Threshold": { - "Formula": "metric_TMA_..........Port_6(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..........Port_6(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..........Port_6(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5358,7 +6878,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Load_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Load_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Load_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5404,7 +6932,15 @@ "Category": "TMA", "CountDomain": "Core_Execution", "Threshold": { - "Formula": "metric_TMA_........Store_Op_Utilization(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........Store_Op_Utilization(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_........Store_Op_Utilization(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -5440,7 +6976,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Retiring(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 70 | b > 10", + "BaseFormula": "metric_TMA_Retiring(%) > 70 | metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvUW;TmaL1", @@ -5505,7 +7053,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 60", + "BaseFormula": "metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -5562,7 +7118,19 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 20 & b > 60", + "BaseFormula": "metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -5607,7 +7175,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......X87_Use(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......X87_Use(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute", @@ -5652,7 +7236,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Scalar(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Scalar(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5697,7 +7297,23 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 20 & c > 60", + "BaseFormula": "metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5746,7 +7362,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_128b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_128b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5795,7 +7431,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_256b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_256b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5844,7 +7500,27 @@ "Category": "TMA", "CountDomain": "Uops", "Threshold": { - "Formula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Vector_512b(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_......FP_Vector(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_....FP_Arith(%)" + }, + { + "Alias": "d", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 10 & c > 20 & d > 60", + "BaseFormula": "metric_TMA_........FP_Vector_512b(%) > 10 & metric_TMA_......FP_Vector(%) > 10 & metric_TMA_....FP_Arith(%) > 20 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "$issue2P" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Compute;Flops", @@ -5917,7 +7593,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Memory_Operations(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Memory_Operations(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -5986,7 +7674,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Branch_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Branch_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 60", + "BaseFormula": "metric_TMA_....Branch_Instructions(%) > 10 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;BvBO;Pipeline", @@ -6079,7 +7779,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 30 & b > 60", + "BaseFormula": "metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", @@ -6148,7 +7860,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Nop_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Other_Light_Ops(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Light_Operations(%)" + } + ], + "Formula": "a > 10 & b > 30 & c > 60", + "BaseFormula": "metric_TMA_......Nop_Instructions(%) > 10 & metric_TMA_....Other_Light_Ops(%) > 30 & metric_TMA_..Light_Operations(%) > 60", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvBO;Pipeline", @@ -6213,7 +7941,15 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Retire;TmaL2", @@ -6278,7 +8014,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Few_Uops_Instructions(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Few_Uops_Instructions(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueD0" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6315,7 +8063,19 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 5 & b > 10", + "BaseFormula": "metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "$issueMC, $issueMS" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq", @@ -6344,7 +8104,23 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......Assists(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......Assists(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BvIO", @@ -6373,7 +8149,15 @@ "Category": "TMA", "CountDomain": "Slots_Estimated", "Threshold": { - "Formula": "metric_TMA_........FP_Assists(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_........FP_Assists(%)" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_........FP_Assists(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC", @@ -6414,7 +8198,23 @@ "Category": "TMA", "CountDomain": "Slots", "Threshold": { - "Formula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_......CISC(%)" + }, + { + "Alias": "b", + "Value": "metric_TMA_....Microcode_Sequencer(%)" + }, + { + "Alias": "c", + "Value": "metric_TMA_..Heavy_Operations(%)" + } + ], + "Formula": "a > 10 & b > 5 & c > 10", + "BaseFormula": "metric_TMA_......CISC(%) > 10 & metric_TMA_....Microcode_Sequencer(%) > 5 & metric_TMA_..Heavy_Operations(%) > 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "", @@ -6507,7 +8307,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_Core_Bound_Likely > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_Core_Bound_Likely" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Botlnk_Core_Bound_Likely > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;SMT", @@ -6534,6 +8342,11 @@ "BaseFormula": " inst_retired.any / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Summary", "LocateWith": "" @@ -6576,7 +8389,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UopPI > 1.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UopPI" + } + ], + "Formula": "a > 1.05", + "BaseFormula": "metric_TMA_Info_Thread_UopPI > 1.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret;Retire", @@ -6620,7 +8441,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Thread_UpTB < 5 * 1.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Thread_UpTB" + } + ], + "Formula": "a < 5 * 1.5", + "BaseFormula": "metric_TMA_Info_Thread_UpTB < 5 * 1.5", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW", @@ -6647,6 +8476,11 @@ "BaseFormula": " 1 / tma_info_thread_ipc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Mem", "LocateWith": "" @@ -6668,6 +8502,11 @@ "BaseFormula": " cpu_clk_unhalted.thread", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline", "LocateWith": "" @@ -6689,6 +8528,11 @@ "BaseFormula": " topdown.slots:perf_metrics", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "TmaL1", "LocateWith": "" @@ -6723,6 +8567,11 @@ "BaseFormula": " tma_info_thread_slots / ( topdown.slots:percore / 2 ) if smt_on else 1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "SMT;TmaL1", "LocateWith": "" @@ -6748,6 +8597,11 @@ "BaseFormula": " uops_executed.thread / uops_issued.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline", "LocateWith": "" @@ -6786,6 +8640,11 @@ "BaseFormula": " inst_retired.any / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;SMT;TmaL1", "LocateWith": "" @@ -6840,6 +8699,11 @@ "BaseFormula": " ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.8_flops + 16 * fp_arith_inst_retired.512b_packed_single ) / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Ret;Flops", "LocateWith": "" @@ -6882,6 +8746,11 @@ "BaseFormula": " ( ( fp_arith_inst_retired.scalar ) + ( fp_arith_inst_retired.vector ) ) / ( 2 * tma_info_core_core_clks )", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -6907,6 +8776,11 @@ "BaseFormula": " uops_executed.thread / uops_executed.thread:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "LocateWith": "" @@ -6932,6 +8806,11 @@ "BaseFormula": " uops_executed.thread / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -6966,6 +8845,11 @@ "BaseFormula": " cpu_clk_unhalted.distributed if smt_on else tma_info_thread_clks", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "SMT", "LocateWith": "" @@ -6992,7 +8876,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpLoad < 3" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpLoad" + } + ], + "Formula": "a < 3", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpLoad < 3", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -7020,7 +8912,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpStore < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpStore" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpStore < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "InsType", @@ -7048,7 +8948,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpBranch < 8" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpBranch" + } + ], + "Formula": "a < 8", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpBranch < 8", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;InsType", @@ -7076,7 +8984,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpCall < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpCall" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpCall < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", @@ -7104,7 +9020,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpTB < 5 * 2 + 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpTB" + } + ], + "Formula": "a < 5 * 2 + 1", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpTB < 5 * 2 + 1", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO", @@ -7131,6 +9055,11 @@ "BaseFormula": " br_inst_retired.all_branches / br_inst_retired.near_taken", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;Fed;PGO", "LocateWith": "" @@ -7173,7 +9102,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpFLOP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpFLOP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -7205,7 +9142,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;InsType", @@ -7233,7 +9178,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_SP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -7261,7 +9214,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_Scalar_DP < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpScalar;InsType", @@ -7293,7 +9254,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX128" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX128 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -7325,7 +9294,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX256" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX256 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -7357,7 +9334,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpArith_AVX512 < 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpArith_AVX512" + } + ], + "Formula": "a < 10", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpArith_AVX512 < 10", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", @@ -7384,6 +9369,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / misc_retired.pause_inst", "Category": "TMA", "CountDomain": "Inst_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Flops;FpVector;InsType", "LocateWith": "" @@ -7410,7 +9400,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Inst_Mix_IpSWPF" + } + ], + "Formula": "a < 100", + "BaseFormula": "metric_TMA_Info_Inst_Mix_IpSWPF < 100", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -7433,6 +9431,11 @@ "BaseFormula": " inst_retired.any", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;TmaL1", "LocateWith": "INST_RETIRED.PREC_DIST" @@ -7474,6 +9477,11 @@ "BaseFormula": " ( tma_retiring * tma_info_thread_slots ) / uops_retired.slots:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Pipeline;Ret", "LocateWith": "" @@ -7500,7 +9508,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Pipeline_IpAssist < 100000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Pipeline_IpAssist" + } + ], + "Formula": "a < 100000", + "BaseFormula": "metric_TMA_Info_Pipeline_IpAssist < 100000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", @@ -7540,6 +9556,11 @@ "BaseFormula": " uops_executed.thread / ( ( uops_executed.core_cycles_ge_1 / 2 ) if smt_on else uops_executed.thread:c1 )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "LocateWith": "" @@ -7565,6 +9586,11 @@ "BaseFormula": " lsd.uops / lsd.cycles_active", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7590,6 +9616,11 @@ "BaseFormula": " idq.dsb_uops / idq.dsb_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7615,6 +9646,11 @@ "BaseFormula": " idq.mite_uops / idq.mite_cycles_any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7640,6 +9676,11 @@ "BaseFormula": " uops_issued.any / uops_issued.any:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchBW", "LocateWith": "" @@ -7665,6 +9706,11 @@ "BaseFormula": " lsd.uops / ( uops_issued.any )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;LSD", "LocateWith": "" @@ -7691,7 +9737,19 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 5 > 0.35" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_DSB_Coverage" + }, + { + "Alias": "b", + "Value": "IPC" + } + ], + "Formula": "a < 0.7 & b / 5 > 0.35", + "BaseFormula": "metric_TMA_Info_Frontend_DSB_Coverage < 0.7 & IPC / 5 > 0.35", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -7718,6 +9776,11 @@ "BaseFormula": " dsb2mite_switches.penalty_cycles / dsb2mite_switches.penalty_cycles:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss", "LocateWith": "" @@ -7743,6 +9806,11 @@ "BaseFormula": " br_inst_retired.near_taken / tma_info_thread_clks", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;FetchBW", "LocateWith": "" @@ -7768,6 +9836,11 @@ "BaseFormula": " icache_16b.ifdata_stall / icache_16b.ifdata_stall:c1:e1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", "LocateWith": "" @@ -7794,7 +9867,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret" + } + ], + "Formula": "a < 50", + "BaseFormula": "metric_TMA_Info_Frontend_IpDSB_Miss_Ret < 50", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -7821,6 +9902,11 @@ "BaseFormula": " tma_info_inst_mix_instructions / baclears.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed", "LocateWith": "" @@ -7846,6 +9932,11 @@ "BaseFormula": " 1000 * frontend_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -7871,6 +9962,11 @@ "BaseFormula": " 1000 * l2_rqsts.code_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "IcMiss", "LocateWith": "" @@ -7990,7 +10086,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Misses > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Misses" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Misses > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSBmiss;Fed", @@ -8083,7 +10187,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_DSB_Bandwidth" + } + ], + "Formula": "a > 10", + "BaseFormula": "metric_TMA_Info_Botlnk_DSB_Bandwidth > 10", + "ThresholdIssues": "$issueFB" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "DSB;Fed;FetchBW", @@ -8147,7 +10259,15 @@ "Category": "TMA", "CountDomain": "Scaled_Slots", "Threshold": { - "Formula": "metric_TMA_Info_Botlnk_IC_Misses > 5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Botlnk_IC_Misses" + } + ], + "Formula": "a > 5", + "BaseFormula": "metric_TMA_Info_Botlnk_IC_Misses > 5", + "ThresholdIssues": "$issueFL" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;FetchLat;IcMiss", @@ -8175,7 +10295,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMispredict" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMispredict < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BadSpec;BrMispredicts", @@ -8203,7 +10331,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Ntaken < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8231,7 +10367,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken < 200" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken" + } + ], + "Formula": "a < 200", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Cond_Taken < 200", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8259,7 +10403,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Ret" + } + ], + "Formula": "a < 500", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Ret < 500", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8287,7 +10439,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect" + } + ], + "Formula": "a < 1000", + "BaseFormula": "metric_TMA_Info_Bad_Spec_IpMisp_Indirect < 1000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", @@ -8390,6 +10550,11 @@ "BaseFormula": " tma_mispredictions * tma_info_thread_slots / ( 5 ) / br_misp_retired.all_branches / 100", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "$issueBM" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;BrMispredicts", "LocateWith": "" @@ -8419,6 +10584,11 @@ "BaseFormula": " int_misc.clears_count / ( br_misp_retired.all_branches + machine_clears.count )", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "BrMispredicts", "LocateWith": "" @@ -8444,6 +10614,11 @@ "BaseFormula": " br_inst_retired.cond_ntaken / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", "LocateWith": "" @@ -8469,6 +10644,11 @@ "BaseFormula": " br_inst_retired.cond_taken / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches;CodeGen;PGO", "LocateWith": "" @@ -8498,6 +10678,11 @@ "BaseFormula": " ( br_inst_retired.near_call + br_inst_retired.near_return ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8531,6 +10716,11 @@ "BaseFormula": " ( br_inst_retired.near_taken - br_inst_retired.cond_taken - 2 * br_inst_retired.near_call ) / br_inst_retired.all_branches", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8572,6 +10762,11 @@ "BaseFormula": " 1 - ( tma_info_branches_cond_nt + tma_info_branches_cond_tk + tma_info_branches_callret + tma_info_branches_jump )", "Category": "TMA", "CountDomain": "Fraction", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Bad;Branches", "LocateWith": "" @@ -8601,6 +10796,11 @@ "BaseFormula": " l1d_pend_miss.pending / ( mem_load_retired.l1_miss + mem_load_retired.fb_hit )", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryLat", "LocateWith": "" @@ -8626,6 +10826,11 @@ "BaseFormula": " l1d_pend_miss.pending / l1d_pend_miss.pending_cycles", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBound;MemoryBW", "LocateWith": "" @@ -8651,6 +10856,11 @@ "BaseFormula": " 1000 * mem_load_retired.l1_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8676,6 +10886,11 @@ "BaseFormula": " 1000 * l2_rqsts.all_demand_data_rd / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8701,6 +10916,11 @@ "BaseFormula": " 1000 * mem_load_retired.l2_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;Backend;CacheHits", "LocateWith": "" @@ -8726,6 +10946,11 @@ "BaseFormula": " 1000 * l2_rqsts.miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem;Offcore", "LocateWith": "" @@ -8751,6 +10976,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8776,6 +11006,11 @@ "BaseFormula": " 1000 * l2_rqsts.rfo_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheMisses;Offcore", "LocateWith": "" @@ -8805,6 +11040,11 @@ "BaseFormula": " 1000 * ( l2_rqsts.references - l2_rqsts.miss ) / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8830,6 +11070,11 @@ "BaseFormula": " 1000 * l2_rqsts.demand_data_rd_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8855,6 +11100,11 @@ "BaseFormula": " 1000 * mem_load_retired.l3_miss / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -8880,6 +11130,11 @@ "BaseFormula": " 1000 * mem_load_retired.fb_hit / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "CacheHits;Mem", "LocateWith": "" @@ -8906,6 +11161,11 @@ "BaseFormula": " 64 * l1d.replacement / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8932,6 +11192,11 @@ "BaseFormula": " 64 * l2_lines_in.all / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8958,6 +11223,11 @@ "BaseFormula": " 64 * longest_lat_cache.miss / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -8984,6 +11254,11 @@ "BaseFormula": " 64 * offcore_requests.all_requests / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW;Offcore", "LocateWith": "" @@ -9031,7 +11306,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Page_Walks_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_Memory_Page_Walks_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", @@ -9058,6 +11341,11 @@ "BaseFormula": " 1000 * itlb_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Fed;MemoryTLB", "LocateWith": "" @@ -9083,6 +11371,11 @@ "BaseFormula": " 1000 * dtlb_load_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -9108,6 +11401,11 @@ "BaseFormula": " 1000 * dtlb_store_misses.walk_completed / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryTLB", "LocateWith": "" @@ -9134,6 +11432,11 @@ "BaseFormula": " tma_info_memory_l1d_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9160,6 +11463,11 @@ "BaseFormula": " tma_info_memory_l2_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9186,6 +11494,11 @@ "BaseFormula": " tma_info_memory_l3_cache_fill_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW", "LocateWith": "" @@ -9212,6 +11525,11 @@ "BaseFormula": " tma_info_memory_l3_cache_access_bw", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Mem;MemoryBW;Offcore", "LocateWith": "" @@ -9242,7 +11560,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_Memory_Useless_HWPF" + } + ], + "Formula": "a > 0.15", + "BaseFormula": "metric_TMA_Info_Memory_Useless_HWPF > 0.15", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Prefetches", @@ -9269,6 +11595,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests.demand_data_rd", "Category": "TMA", "CountDomain": "Clocks_Latency", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "LockCont;Memory_Lat;Offcore", "LocateWith": "" @@ -9294,6 +11625,11 @@ "BaseFormula": " offcore_requests_outstanding.demand_data_rd / offcore_requests_outstanding.demand_data_rd:c1", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -9319,6 +11655,11 @@ "BaseFormula": " offcore_requests_outstanding.all_data_rd / offcore_requests_outstanding.cycles_with_data_rd", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Memory_BW;Offcore", "LocateWith": "" @@ -9344,6 +11685,11 @@ "BaseFormula": " 1000 * mem_load_misc_retired.uc / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -9369,6 +11715,11 @@ "BaseFormula": " 1000 * sq_misc.bus_lock / inst_retired.any", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Mem", "LocateWith": "" @@ -9399,6 +11750,11 @@ "BaseFormula": " tma_info_system_cpus_utilized / num_cpus", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "HPC;Summary", "LocateWith": "" @@ -9425,6 +11781,11 @@ "BaseFormula": " cpu_clk_unhalted.ref_tsc / tsc", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", "LocateWith": "" @@ -9459,6 +11820,11 @@ "BaseFormula": " tma_info_system_turbo_utilization * tsc / ( 1000000000 ) / tma_info_system_time", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary;Power", "LocateWith": "" @@ -9501,6 +11867,11 @@ "BaseFormula": " ( ( fp_arith_inst_retired.scalar + 2 * fp_arith_inst_retired.128b_packed_double + 4 * fp_arith_inst_retired.4_flops + 8 * fp_arith_inst_retired.8_flops + 16 * fp_arith_inst_retired.512b_packed_single ) / ( 1000000000 ) ) / tma_info_system_time", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Cor;Flops;HPC", "LocateWith": "" @@ -9526,6 +11897,11 @@ "BaseFormula": " tma_info_thread_clks / cpu_clk_unhalted.ref_tsc", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -9564,6 +11940,11 @@ "BaseFormula": " core_power.lvl0_turbo_license / tma_info_core_core_clks", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Power", "LocateWith": "" @@ -9603,7 +11984,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Power_License1_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Power_License1_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_System_Power_License1_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Power", @@ -9644,7 +12033,15 @@ "Category": "TMA", "CountDomain": "Core_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Power_License2_Utilization > 0.5" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Power_License2_Utilization" + } + ], + "Formula": "a > 0.5", + "BaseFormula": "metric_TMA_Info_System_Power_License2_Utilization > 0.5", + "ThresholdIssues": "" }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "Power", @@ -9680,6 +12077,11 @@ "BaseFormula": " 1 - cpu_clk_unhalted.one_thread_active / cpu_clk_unhalted.ref_distributed if smt_on else 0", "Category": "TMA", "CountDomain": "Core_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CORE, SOCKET, SYSTEM", "MetricGroup": "SMT", "LocateWith": "" @@ -9706,7 +12108,15 @@ "Category": "TMA", "CountDomain": "Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_Kernel_Utilization > 0.05" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Kernel_Utilization" + } + ], + "Formula": "a > 0.05", + "BaseFormula": "metric_TMA_Info_System_Kernel_Utilization > 0.05", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "OS", @@ -9733,6 +12143,11 @@ "BaseFormula": " cpu_clk_unhalted.thread_p:sup / inst_retired.any_p:sup", "Category": "TMA", "CountDomain": "Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "OS", "LocateWith": "" @@ -9763,6 +12178,11 @@ "BaseFormula": " 64 * ( unc_arb_trk_requests.all + unc_arb_coh_trk_requests.all ) / ( 1000000 ) / tma_info_system_time / 1000", "Category": "TMA", "CountDomain": "GB/sec", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "$issueBW" + }, "ResolutionLevels": "ARB", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC", "LocateWith": "" @@ -9789,6 +12209,11 @@ "BaseFormula": " unc_pkg_energy_status * ( 61 ) / ( tma_info_system_time * ( 1000000 ) )", "Category": "TMA", "CountDomain": "System_Metric", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "PKG", "MetricGroup": "Power;SoC", "LocateWith": "" @@ -9811,7 +12236,15 @@ "Category": "TMA", "CountDomain": "Seconds", "Threshold": { - "Formula": "metric_TMA_Info_System_Time < 1" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_Time" + } + ], + "Formula": "a < 1", + "BaseFormula": "metric_TMA_Info_System_Time < 1", + "ThresholdIssues": "" }, "ResolutionLevels": "SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -9839,7 +12272,15 @@ "Category": "TMA", "CountDomain": "Clocks", "Threshold": { - "Formula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_MUX" + } + ], + "Formula": "a > 1.1 | a < 0.9", + "BaseFormula": "metric_TMA_Info_System_MUX > 1.1 | metric_TMA_Info_System_MUX < 0.9", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Summary", @@ -9862,6 +12303,11 @@ "BaseFormula": " unc_clock.socket", "Category": "TMA", "CountDomain": "Count", + "Threshold": { + "Formula": "", + "BaseFormula": "", + "ThresholdIssues": "" + }, "ResolutionLevels": "CLOCK", "MetricGroup": "SoC", "LocateWith": "" @@ -9888,7 +12334,15 @@ "Category": "TMA", "CountDomain": "Inst_Metric", "Threshold": { - "Formula": "metric_TMA_Info_System_IpFarBranch < 1000000" + "ThresholdMetrics": [ + { + "Alias": "a", + "Value": "metric_TMA_Info_System_IpFarBranch" + } + ], + "Formula": "a < 1000000", + "BaseFormula": "metric_TMA_Info_System_IpFarBranch < 1000000", + "ThresholdIssues": "" }, "ResolutionLevels": "THREAD, CORE, SOCKET, SYSTEM", "MetricGroup": "Branches;OS",