Adding benchmark data wrt 14.301

Refactored the csv directory layout

docs/performance/clSPARSE-0.6.0.0/W9100/Csr2Coo.csv

@@ -0,0 +1,17 @@
+csr2coo,,
+,,
+OpenCL runtime:,1573.4 (VM),
+OpenCL Device:,w9100,
+,,Gi-Elements/s
+cant,,5.34
+consph,,5.80
+cop20k_A,,4.58
+mac_econ_fwd500,,3.37
+mc2depi,,4.22
+pdb1HYS,,5.67
+pwtk,,6.80
+rail4284,,6.15
+rma10,,4.32
+scircuit,,2.47
+shipsec1,,5.03
+webbase_1M,,3.63

docs/performance/clSPARSE-0.6.0.0/W9100/Csr2Dense.csv

@@ -0,0 +1,15 @@
+csr2dense,,
+,,
+OpenCL runtime:,1573.4 (VM),
+OpenCL Device:,w9100,
+,,Gi-Elements/s
+Dubcova1,,0.028
+hydr1c_A_11,,0.021
+hydr1c_A_72,,0.021
+hydr1c_A_76,,0.021
+Maragal_6,,0.069
+Na5,,0.224
+psse1,,0.010
+Reuters911,,0.046
+Si10H16,,0.085
+tomography,,0.196

docs/performance/clSPARSE-0.6.0.0/W9100/README.md

@@ -0,0 +1,26 @@
+# Benchmarking
+## Hardware
+w9100
+
+## Environment
+Ubuntu 14.04
+
+clSPARSE v0.6.0.0
+
+[Catalyst FirePro](http://support.amd.com/en-us/download/workstation?os=Linux%20x86_64#catalyst-pro) 14.301
+
+## Tool
+[clsparse-bench](clSPARSE\src\benchmarks\clsparse-bench)
+
+## Methodology
+For each data point, we took 20 samples. Each sample consists of 20 calls
+with a wait afterward. We benchmark with respect to the API, utilizing host timers
+(not pure kernel time with ).  
+Outlying samples beyond 1 standard deviation were removed.
+
+Conversion routines benchmarked as number of Gi-Elements/s converted
+
+SpM-dV routine calculated as Gi-Bytes/s
+```c
+( sizeof( cl_int )*( csrMtx.num_nonzeros + csrMtx.num_rows ) + sizeof( T ) * ( csrMtx.num_nonzeros + csrMtx.num_cols + csrMtx.num_rows ) ) / time_in_ns( );
+```

docs/performance/clSPARSE-0.6.0.0/W9100/SpM-dV.csv

@@ -0,0 +1,17 @@
+SpM-dV,,
+,,
+OpenCL runtime:,1573.4 (VM),
+OpenCL Device:,w9100,
+,,Gi-Bytes/s
+cant,,99.7
+consph,,118.8
+cop20k_A,,90.1
+mac_econ_fwd500,,65.9
+mc2depi,,113.4
+pdb1HYS,,102.6
+pwtk,,144.9
+rail4284,,100.7
+rma10,,78.4
+scircuit,,56.5
+shipsec1,,100.5
+webbase_1M,,120.3

docs/performance/clSPARSE-0.6.0.0/W9100/clsparse-bench.sh

@@ -0,0 +1,33 @@
+#!/bin/bash
+
+#Executable
+clsparse_bench=~/code/github/clMathLibraries/bin/clSPARSE/release/clSPARSE-build/staging/clsparse-bench
+
+#Data directories
+mtx_cant=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/cant
+mtx_consph=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/consph
+mtx_cop20k_A=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/cop20k_A
+mtx_mac_econ_fwd500=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/mac_econ_fwd500
+mtx_mc2depi=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/mc2depi
+mtx_pdb1HYS=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/pdb1HYS
+mtx_pwtk=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/pwtk
+mtx_rail4284=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/rail4284
+mtx_rma10=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/rma10
+mtx_scircuit=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/scircuit
+mtx_shipsec1=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/shipsec1
+mtx_webbase_1M=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/webbase-1M
+
+clsparse_timing_dir=timings/spm-dv/14.301
+mkdir -p ${clsparse_timing_dir}
+${clsparse_bench} -f spmdv -d ${mtx_cant} > ${clsparse_timing_dir}/cant.txt
+${clsparse_bench} -f spmdv -d ${mtx_consph} > ${clsparse_timing_dir}/consph.txt
+${clsparse_bench} -f spmdv -d ${mtx_cop20k_A} > ${clsparse_timing_dir}/cop20k_A.txt
+${clsparse_bench} -f spmdv -d ${mtx_mac_econ_fwd500} > ${clsparse_timing_dir}/mac_econ_fwd500.txt
+${clsparse_bench} -f spmdv -d ${mtx_mc2depi} > ${clsparse_timing_dir}/mc2depi.txt
+${clsparse_bench} -f spmdv -d ${mtx_pdb1HYS} > ${clsparse_timing_dir}/pdb1HYS.txt
+${clsparse_bench} -f spmdv -d ${mtx_pwtk} > ${clsparse_timing_dir}/pwtk.txt
+${clsparse_bench} -f spmdv -d ${mtx_rail4284} > ${clsparse_timing_dir}/rail4284.txt
+${clsparse_bench} -f spmdv -d ${mtx_rma10} > ${clsparse_timing_dir}/rma10.txt
+${clsparse_bench} -f spmdv -d ${mtx_scircuit} > ${clsparse_timing_dir}/scircuit.txt
+${clsparse_bench} -f spmdv -d ${mtx_shipsec1} > ${clsparse_timing_dir}/shipsec1.txt
+${clsparse_bench} -f spmdv -d ${mtx_webbase_1M} > ${clsparse_timing_dir}/webbase_1M.txt

docs/performance/cuSPARSE-7.0/Tesla_K40/Csr2Coo.csv

@@ -0,0 +1,17 @@
+Csr2Coo,,
+,,
+CUDA driver:,346.47,
+OpenCL Device:,k40c,
+,,Gi-Elements/s
+cant,,12.58
+consph,,13.22
+cop20k_A,,6.12
+mac_econ_fwd500,,1.89
+mc2depi,,1.26
+pdb1HYS,,16.58
+pwtk,,12.41
+rail4284,,14.61
+rma10,,10.85
+scircuit,,1.71
+shipsec1,,13.00
+webbase_1M,,0.99

docs/performance/cuSPARSE-7.0/Tesla_K40/Csr2Dense.csv

@@ -0,0 +1,15 @@
+Csr2Dense,,
+,,
+CUDA driver:,346.47,
+OpenCL Device:,k40c,
+,,Gi-Elements/s
+Dubcova1,,0.011
+hydr1c_A_11,,0.012
+hydr1c_A_72,,0.012
+hydr1c_A_76,,0.012
+Maragal_6,,0.026
+Na5,,0.156
+psse1,,0.005
+Reuters911,,0.022
+Si10H16,,0.031
+tomography,,0.566

docs/performance/cuSPARSE-7.0/Tesla_K40/README.md

@@ -0,0 +1,26 @@
+# Benchmarking
+## Hardware
+Tesla K40c
+
+## Environment
+OpenSUSE 13.2
+
+cuSPARSE v7.0
+
+Tesla driver 346.47
+
+## Tool
+[cusparse-bench](clSPARSE\src\benchmarks\cusparse-bench)
+
+## Methodology
+For each data point, we took 20 samples. Each sample consists of 20 calls
+with a wait afterward. We benchmark with respect to the API, utilizing host timers
+(not pure kernel time with ).  
+Outlying samples beyond 1 standard deviation were removed.
+
+Conversion routines benchmarked as number of Gi-Elements/s converted
+
+SpM-dV routine calculated as Gi-Bytes/s
+```c
+( sizeof( cl_int )*( csrMtx.num_nonzeros + csrMtx.num_rows ) + sizeof( T ) * ( csrMtx.num_nonzeros + csrMtx.num_cols + csrMtx.num_rows ) ) / time_in_ns( );
+```

docs/performance/cuSPARSE-7.0/Tesla_K40/SpM-dV.csv

@@ -0,0 +1,17 @@
+SpM-dV,,
+,,
+CUDA driver:,346.47,
+OpenCL Device:,k40c,
+,,Gi-Bytes/s
+cant,,96.5
+consph,,102.1
+cop20k_A,,72.1
+mac_econ_fwd500,,47.2
+mc2depi,,74.3
+pdb1HYS,,103.0
+pwtk,,74.2
+rail4284,,34.8
+rma10,,64.9
+scircuit,,51.2
+shipsec1,,161.2
+webbase_1M,,32.6

docs/performance/cuSPARSE-7.0/Tesla_K40/cusparse-bench.sh

@@ -0,0 +1,32 @@
+#!/bin/bash
+
+#Executable
+cusparse_bench=~/code/github/clMathLibraries/bin/clSPARSE/release/clSPARSE-build/staging/cusparse-bench
+
+#Data directories
+mtx_cant=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/cant
+mtx_consph=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/consph
+mtx_cop20k_A=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/cop20k_A
+mtx_mac_econ_fwd500=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/mac_econ_fwd500
+mtx_mc2depi=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/mc2depi
+mtx_pdb1HYS=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/pdb1HYS
+mtx_pwtk=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/pwtk
+mtx_rail4284=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/rail4284
+mtx_rma10=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/rma10
+mtx_scircuit=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/scircuit
+mtx_shipsec1=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/shipsec1
+mtx_webbase_1M=~/code/github/clMathLibraries/bin/deps/release/Externals/MTX/Bell_Garland/webbase-1M
+
+mkdir -p timings-cusparse-spmdv
+${cusparse_bench} -f spmdv -d ${mtx_cant} > timings-cusparse-spmdv/cant.txt
+${cusparse_bench} -f spmdv -d ${mtx_consph} > timings-cusparse-spmdv/consph.txt
+${cusparse_bench} -f spmdv -d ${mtx_cop20k_A} > timings-cusparse-spmdv/cop20k_A.txt
+${cusparse_bench} -f spmdv -d ${mtx_mac_econ_fwd500} > timings-cusparse-spmdv/mac_econ_fwd500.txt
+${cusparse_bench} -f spmdv -d ${mtx_mc2depi} > timings-cusparse-spmdv/mc2depi.txt
+${cusparse_bench} -f spmdv -d ${mtx_pdb1HYS} > timings-cusparse-spmdv/pdb1HYS.txt
+${cusparse_bench} -f spmdv -d ${mtx_pwtk} > timings-cusparse-spmdv/pwtk.txt
+${cusparse_bench} -f spmdv -d ${mtx_rail4284} > timings-cusparse-spmdv/rail4284.txt
+${cusparse_bench} -f spmdv -d ${mtx_rma10} > timings-cusparse-spmdv/rma10.txt
+${cusparse_bench} -f spmdv -d ${mtx_scircuit} > timings-cusparse-spmdv/scircuit.txt
+${cusparse_bench} -f spmdv -d ${mtx_shipsec1} > timings-cusparse-spmdv/shipsec1.txt
+${cusparse_bench} -f spmdv -d ${mtx_webbase_1M} > timings-cusparse-spmdv/webbase_1M.txt