benchdnn: graph: use plain copy and paste if reorder fails

oneapi-src · Dec 17, 2024 · 141de68 · 141de68
1 parent f4fcbb6
commit 141de68
Show file tree

Hide file tree

Showing 2 changed files with 38 additions and 33 deletions.
diff --git a/tests/benchdnn/dnnl_memory.cpp b/tests/benchdnn/dnnl_memory.cpp
@@ -137,26 +137,46 @@ int execute_reorder(const dnn_mem_t &src, dnn_mem_t &dst,
     }
 #endif
 
-    if (!r_pd_) {
-        DNN_SAFE(dnnl_reorder_primitive_desc_create(&r_pd_, src.md_,
-                         src.engine(), dst.md_, dst.engine(), attr),
-                WARN);
+    while (!r_pd_) {
+        // Fallback to GPU reorder.
+        auto status = dnnl_reorder_primitive_desc_create(
+                &r_pd_, src.md_, src.engine(), dst.md_, dst.engine(), attr);
+        if (status != dnnl_success) break;
+
+        auto r_pd = make_benchdnn_dnnl_wrapper(r_pd_);
+        const auto &scratchpad_md = query_md(r_pd, DNNL_ARG_SCRATCHPAD);
+        const auto &scratchpad_engine
+                = dst.engine_kind() == dnnl_gpu ? dst.engine() : src.engine();
+        dnn_mem_t scratchpad(scratchpad_md, scratchpad_engine);
+
+        DNN_SAFE(dnnl_primitive_create(&prim_, r_pd), CRIT);
+        auto prim = make_benchdnn_dnnl_wrapper(prim_);
+
+        args_t args;
+        args.set(DNNL_ARG_FROM, *r_src);
+        args.set(DNNL_ARG_TO, *r_dst);
+        args.set(DNNL_ARG_SCRATCHPAD, scratchpad);
+
+        return execute_and_wait(prim, args);
     }
-    auto r_pd = make_benchdnn_dnnl_wrapper(r_pd_);
-    const auto &scratchpad_md = query_md(r_pd, DNNL_ARG_SCRATCHPAD);
-    const auto &scratchpad_engine
-            = dst.engine_kind() == dnnl_gpu ? dst.engine() : src.engine();
-    dnn_mem_t scratchpad(scratchpad_md, scratchpad_engine);
 
-    DNN_SAFE(dnnl_primitive_create(&prim_, r_pd), CRIT);
-    auto prim = make_benchdnn_dnnl_wrapper(prim_);
-
-    args_t args;
-    args.set(DNNL_ARG_FROM, *r_src);
-    args.set(DNNL_ARG_TO, *r_dst);
-    args.set(DNNL_ARG_SCRATCHPAD, scratchpad);
+    if (dnnl_memory_desc_equal(src.md_, dst.md_)) {
+        // If fail to create reorder pd, use plain data copy for identical mds.
+        BENCHDNN_PRINT(2, "%s\n", "[REORDER] Fallback to plain copy.");
+        const int64_t chunk_size = 64;
+        const int64_t n_chunks = div_up(src.nelems(), chunk_size);
+        benchdnn_parallel_nd(n_chunks, [&](int64_t idx_chunk) {
+            int64_t idx_start = idx_chunk * chunk_size;
+            int64_t idx_end = MIN2(idx_start + chunk_size, src.nelems());
+            for (int64_t idx = idx_start; idx < idx_end; ++idx) {
+                float e = src.get_elem(idx);
+                dst.set_elem(idx, e);
+            }
+        });
+        return OK;
+    }
 
-    return execute_and_wait(prim, args);
+    return FAIL;
 }
 
 // `swap_dt` changes `this` data type which may be needed for

diff --git a/tests/benchdnn/graph/input_displacer.cpp b/tests/benchdnn/graph/input_displacer.cpp
@@ -297,22 +297,7 @@ int partition_data_displacer_t::displace_input_data(
     }
     dnnl_memory_desc_destroy(mem_replace.md_);
     dnnl_memory_desc_clone(&mem_replace.md_, md);
-
-    // As int4->int4 reorder is not supported, use naive data copy and paste.
-    if (md->data_type == dnnl_s4 || md->data_type == dnnl_u4) {
-        const int64_t chunk_size = 64;
-        const int64_t n_chunks = div_up(mem.nelems(), chunk_size);
-        benchdnn_parallel_nd(n_chunks, [&](int64_t idx_chunk) {
-            int64_t idx_start = idx_chunk * chunk_size;
-            int64_t idx_end = MIN2(idx_start + chunk_size, mem.nelems());
-
-            for (int64_t idx = idx_start; idx < idx_end; ++idx) {
-                float e = mem_replace.get_elem(idx);
-                mem.set_elem(idx, e);
-            }
-        });
-    } else
-        SAFE(mem.reorder(mem_replace), WARN);
+    SAFE(mem.reorder(mem_replace), WARN);
 
     if (is_reshaped_dims) dnnl_memory_desc_destroy(md);
     return OK;