diff --git a/benchmarks/config/base/base.json b/benchmarks/config/base/base.json index f8495339d..15e2f3d06 100644 --- a/benchmarks/config/base/base.json +++ b/benchmarks/config/base/base.json @@ -5,28 +5,28 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": {}, - "flags": [ "100", "256", "0", "F", "32", "32", "32", "0", "0", "1024", "1024", "1024", "1024" ], + "flags": [ "100", "256", "0", "F", "32", "32", "32", "0", "1", "1024", "1024", "1024", "1024" ], "extensions": [] }, "gemm_bf16_dnn_target": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": {}, - "flags": [ "100", "256", "0", "F", "32", "32", "32", "1", "1", "1024", "1024", "1024", "1024" ], + "flags": [ "100", "256", "0", "F", "32", "32", "32", "1", "2", "1024", "1024", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "mlp_fp32_dnn_target": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": {}, - "flags": [ "100", "256", "5", "F", "32", "32", "32", "0", "0", "1024", "1024", "1024", "1024" ], + "flags": [ "100", "256", "5", "F", "32", "32", "32", "0", "1", "1024", "1024", "1024", "1024" ], "extensions": [] }, "mlp_bf16_dnn_target": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": {}, - "flags": [ "100", "256", "5", "F", "32", "32", "32", "1", "1", "1024", "1024", "1024", "1024" ], + "flags": [ "100", "256", "5", "F", "32", "32", "32", "1", "2", "1024", "1024", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "gemm_fp32_mlir": { diff --git a/benchmarks/config/fc/1024x1024x512.json b/benchmarks/config/fc/1024x1024x512.json index a19df9cdb..d98fd17c3 100644 --- a/benchmarks/config/fc/1024x1024x512.json +++ b/benchmarks/config/fc/1024x1024x512.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "3", "F", "64", "64", "64", "0", "0", "512", "1024" ], + "flags": [ "100", "1024", "3", "F", "64", "64", "64", "0", "1", "512", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "3", "F", "64", "64", "64", "0", "0", "512", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "3", "F", "64", "64", "64", "0", "1", "512", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "3", "F", "64", "64", "64", "1", "1", "512", "1024" ], + "flags": [ "100", "1024", "3", "F", "64", "64", "64", "1", "2", "512", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "3", "F", "64", "64", "64", "1", "1", "512", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "3", "F", "64", "64", "64", "1", "2", "512", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/1024x2560x1024.json b/benchmarks/config/fc/1024x2560x1024.json index 8db88e17a..67b71ca95 100644 --- a/benchmarks/config/fc/1024x2560x1024.json +++ b/benchmarks/config/fc/1024x2560x1024.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "3", "F", "64", "64", "64", "0", "0", "1024", "2560" ], + "flags": [ "100", "1024", "3", "F", "64", "64", "64", "0", "1", "1024", "2560" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "3", "F", "64", "64", "64", "0", "0", "1024", "2560" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "3", "F", "64", "64", "64", "0", "1", "1024", "2560" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "3", "F", "64", "64", "64", "1", "1", "1024", "2560" ], + "flags": [ "100", "1024", "3", "F", "64", "64", "64", "1", "2", "1024", "2560" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "3", "F", "64", "64", "64", "1", "1", "1024", "2560" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "3", "F", "64", "64", "64", "1", "2", "1024", "2560" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/1024x352x512.json b/benchmarks/config/fc/1024x352x512.json index 3502704b1..2afda78e2 100644 --- a/benchmarks/config/fc/1024x352x512.json +++ b/benchmarks/config/fc/1024x352x512.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "3", "F", "32", "32", "32", "0", "0", "512", "352" ], + "flags": [ "100", "1024", "3", "F", "32", "32", "32", "0", "1", "512", "352" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "3", "F", "32", "32", "32", "0", "0", "512", "352" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "3", "F", "32", "32", "32", "0", "1", "512", "352" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "3", "F", "32", "32", "32", "1", "1", "512", "352" ], + "flags": [ "100", "1024", "3", "F", "32", "32", "32", "1", "2", "512", "352" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "3", "F", "32", "32", "32", "1", "1", "512", "352" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "3", "F", "32", "32", "32", "1", "2", "512", "352" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/1024x512x256.json b/benchmarks/config/fc/1024x512x256.json index 02526aec2..d6c08a3eb 100644 --- a/benchmarks/config/fc/1024x512x256.json +++ b/benchmarks/config/fc/1024x512x256.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "3", "F", "64", "64", "64", "0", "0", "256", "512" ], + "flags": [ "100", "1024", "3", "F", "64", "64", "64", "0", "1", "256", "512" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "3", "F", "64", "64", "64", "0", "0", "256", "512" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "3", "F", "64", "64", "64", "0", "1", "256", "512" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "3", "F", "64", "64", "64", "1", "1", "256", "512" ], + "flags": [ "100", "1024", "3", "F", "64", "64", "64", "1", "2", "256", "512" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "3", "F", "64", "64", "64", "1", "1", "256", "512" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "3", "F", "64", "64", "64", "1", "2", "256", "512" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/128x1024x1024.json b/benchmarks/config/fc/128x1024x1024.json index a39c766e4..32ee065b5 100644 --- a/benchmarks/config/fc/128x1024x1024.json +++ b/benchmarks/config/fc/128x1024x1024.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "0", "1024", "1024" ], + "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "1", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "0", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "1", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "1", "1024", "1024" ], + "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "2", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "1", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "2", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/128x1024x4096.json b/benchmarks/config/fc/128x1024x4096.json index 91d3d0d3a..5bb7d2fb5 100644 --- a/benchmarks/config/fc/128x1024x4096.json +++ b/benchmarks/config/fc/128x1024x4096.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "0", "4096", "1024" ], + "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "1", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "0", "4096", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "1", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "1", "4096", "1024" ], + "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "2", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "1", "4096", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "2", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/128x3072x768.json b/benchmarks/config/fc/128x3072x768.json index 74ef3c736..89f412c03 100644 --- a/benchmarks/config/fc/128x3072x768.json +++ b/benchmarks/config/fc/128x3072x768.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "0", "768", "3072" ], + "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "1", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "0", "768", "3072" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "1", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "1", "768", "3072" ], + "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "2", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "1", "768", "3072" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "2", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/128x4096x1024.json b/benchmarks/config/fc/128x4096x1024.json index 78f909abc..4bd8707e0 100644 --- a/benchmarks/config/fc/128x4096x1024.json +++ b/benchmarks/config/fc/128x4096x1024.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "0", "1024", "4096" ], + "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "1", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "0", "1024", "4096" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "3", "F", "64", "64", "64", "0", "1", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "1", "1024", "4096" ], + "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "2", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "1", "1024", "4096" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "3", "F", "64", "64", "64", "1", "2", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/128x768x2304.json b/benchmarks/config/fc/128x768x2304.json index b136d59cb..2735bba1f 100644 --- a/benchmarks/config/fc/128x768x2304.json +++ b/benchmarks/config/fc/128x768x2304.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "3", "F", "64", "64", "48", "0", "0", "2304", "768" ], + "flags": [ "100", "128", "3", "F", "64", "64", "48", "0", "1", "2304", "768" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "3", "F", "64", "64", "48", "0", "0", "2304", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "3", "F", "64", "64", "48", "0", "1", "2304", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "3", "F", "64", "64", "48", "1", "1", "2304", "768" ], + "flags": [ "100", "128", "3", "F", "64", "64", "48", "1", "2", "2304", "768" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "3", "F", "64", "64", "48", "1", "1", "2304", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "3", "F", "64", "64", "48", "1", "2", "2304", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/128x768x3072.json b/benchmarks/config/fc/128x768x3072.json index 187f5b9ab..04228c324 100644 --- a/benchmarks/config/fc/128x768x3072.json +++ b/benchmarks/config/fc/128x768x3072.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "3", "F", "32", "32", "48", "0", "0", "3072", "768" ], + "flags": [ "100", "128", "3", "F", "32", "32", "48", "0", "1", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "3", "F", "32", "32", "48", "0", "0", "3072", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "3", "F", "32", "32", "48", "0", "1", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "3", "F", "32", "32", "48", "1", "1", "3072", "768" ], + "flags": [ "100", "128", "3", "F", "32", "32", "48", "1", "2", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "3", "F", "32", "32", "48", "1", "1", "3072", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "3", "F", "32", "32", "48", "1", "2", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/256x1024x1024.json b/benchmarks/config/fc/256x1024x1024.json index 9cae42455..9d86d7d35 100644 --- a/benchmarks/config/fc/256x1024x1024.json +++ b/benchmarks/config/fc/256x1024x1024.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "0", "1024", "1024" ], + "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "1", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "0", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "1", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "1", "1024", "1024" ], + "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "2", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "1", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "2", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/256x1024x4096.json b/benchmarks/config/fc/256x1024x4096.json index 6f3d16eab..47ec92120 100644 --- a/benchmarks/config/fc/256x1024x4096.json +++ b/benchmarks/config/fc/256x1024x4096.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "0", "4096", "1024" ], + "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "1", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "0", "4096", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "1", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "1", "4096", "1024" ], + "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "2", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "1", "4096", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "2", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/256x3072x768.json b/benchmarks/config/fc/256x3072x768.json index 66dfbc06d..725e9b756 100644 --- a/benchmarks/config/fc/256x3072x768.json +++ b/benchmarks/config/fc/256x3072x768.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "0", "768", "3072" ], + "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "1", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "0", "768", "3072" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "1", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "1", "768", "3072" ], + "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "2", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "1", "768", "3072" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "2", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/256x4096x1024.json b/benchmarks/config/fc/256x4096x1024.json index 5df4f8dde..72e4a6d93 100644 --- a/benchmarks/config/fc/256x4096x1024.json +++ b/benchmarks/config/fc/256x4096x1024.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "0", "1024", "4096" ], + "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "1", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "0", "1024", "4096" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "1", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "1", "1024", "4096" ], + "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "2", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "1", "1024", "4096" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "2", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/256x768x3072.json b/benchmarks/config/fc/256x768x3072.json index a5bde928a..795b5cee2 100644 --- a/benchmarks/config/fc/256x768x3072.json +++ b/benchmarks/config/fc/256x768x3072.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "0", "3072", "768" ], + "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "1", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "0", "3072", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "1", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "1", "3072", "768" ], + "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "2", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "1", "3072", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "2", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/fc/256x768x768.json b/benchmarks/config/fc/256x768x768.json index eea447d19..681ac7b7b 100644 --- a/benchmarks/config/fc/256x768x768.json +++ b/benchmarks/config/fc/256x768x768.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "0", "768", "768" ], + "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "1", "768", "768" ], "extensions": [ "(avx2|asimd)" ] }, "fc_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "0", "768", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "3", "F", "64", "64", "64", "0", "1", "768", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "1", "768", "768" ], + "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "2", "768", "768" ], "extensions": [ "(avx2|asimd)" ] }, "fc_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "1", "768", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "3", "F", "64", "64", "64", "1", "2", "768", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/1024x1024x512.json b/benchmarks/config/matmul/1024x1024x512.json index d042d9932..eda58e8bc 100644 --- a/benchmarks/config/matmul/1024x1024x512.json +++ b/benchmarks/config/matmul/1024x1024x512.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "0", "F", "64", "64", "64", "0", "512", "1024" ], + "flags": [ "100", "1024", "0", "F", "64", "64", "64", "0", "1", "512", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "0", "F", "64", "64", "64", "0", "512", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "0", "F", "64", "64", "64", "0", "1", "512", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "0", "F", "64", "64", "64", "1", "512", "1024" ], + "flags": [ "100", "1024", "0", "F", "64", "64", "64", "1", "2", "512", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "0", "F", "64", "64", "64", "1", "512", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "0", "F", "64", "64", "64", "1", "2", "512", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/1024x2560x1024.json b/benchmarks/config/matmul/1024x2560x1024.json index 7d3c79e31..e5a540135 100644 --- a/benchmarks/config/matmul/1024x2560x1024.json +++ b/benchmarks/config/matmul/1024x2560x1024.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "0", "F", "64", "64", "64", "0", "1024", "2560" ], + "flags": [ "100", "1024", "0", "F", "64", "64", "64", "0", "1", "1024", "2560" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "0", "F", "64", "64", "64", "0", "1024", "2560" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "0", "F", "64", "64", "64", "0", "1", "1024", "2560" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "0", "F", "64", "64", "64", "1", "1024", "2560" ], + "flags": [ "100", "1024", "0", "F", "64", "64", "64", "1", "2", "1024", "2560" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "0", "F", "64", "64", "64", "1", "1024", "2560" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "0", "F", "64", "64", "64", "1", "2", "1024", "2560" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/1024x352x512.json b/benchmarks/config/matmul/1024x352x512.json index 43825729a..0c4371c63 100644 --- a/benchmarks/config/matmul/1024x352x512.json +++ b/benchmarks/config/matmul/1024x352x512.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "0", "F", "32", "32", "32", "0", "512", "352" ], + "flags": [ "100", "1024", "0", "F", "32", "32", "32", "0", "1", "512", "352" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "0", "F", "32", "32", "32", "0", "512", "352" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "0", "F", "32", "32", "32", "0", "1", "512", "352" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "0", "F", "32", "32", "32", "1", "512", "352" ], + "flags": [ "100", "1024", "0", "F", "32", "32", "32", "1", "2", "512", "352" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "0", "F", "32", "32", "32", "1", "512", "352" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "0", "F", "32", "32", "32", "1", "2", "512", "352" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/1024x512x256.json b/benchmarks/config/matmul/1024x512x256.json index 541d492eb..1a5d45737 100644 --- a/benchmarks/config/matmul/1024x512x256.json +++ b/benchmarks/config/matmul/1024x512x256.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "0", "F", "64", "64", "64", "0", "256", "512" ], + "flags": [ "100", "1024", "0", "F", "64", "64", "64", "0", "1", "256", "512" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "0", "F", "64", "64", "64", "0", "256", "512" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "0", "F", "64", "64", "64", "0", "1", "256", "512" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "1024", "0", "F", "64", "64", "64", "1", "256", "512" ], + "flags": [ "100", "1024", "0", "F", "64", "64", "64", "1", "2", "256", "512" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "1024", "0", "F", "64", "64", "64", "1", "256", "512" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "1024", "0", "F", "64", "64", "64", "1", "2", "256", "512" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/128x1024x1024.json b/benchmarks/config/matmul/128x1024x1024.json index 4b0e38450..9feb47305 100644 --- a/benchmarks/config/matmul/128x1024x1024.json +++ b/benchmarks/config/matmul/128x1024x1024.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "1024", "1024" ], + "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "1", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "1", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "1024", "1024" ], + "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "2", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "2", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/128x1024x4096.json b/benchmarks/config/matmul/128x1024x4096.json index 57fff550d..f18769ada 100644 --- a/benchmarks/config/matmul/128x1024x4096.json +++ b/benchmarks/config/matmul/128x1024x4096.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "4096", "1024" ], + "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "1", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "4096", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "1", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "4096", "1024" ], + "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "2", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "4096", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "2", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/128x3072x768.json b/benchmarks/config/matmul/128x3072x768.json index d1e252e29..65f7faaa1 100644 --- a/benchmarks/config/matmul/128x3072x768.json +++ b/benchmarks/config/matmul/128x3072x768.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "768", "3072" ], + "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "1", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "768", "3072" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "1", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "768", "3072" ], + "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "2", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "768", "3072" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "2", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/128x4096x1024.json b/benchmarks/config/matmul/128x4096x1024.json index e23dbfa32..4aebf866a 100644 --- a/benchmarks/config/matmul/128x4096x1024.json +++ b/benchmarks/config/matmul/128x4096x1024.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "1024", "4096" ], + "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "1", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "1024", "4096" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "0", "F", "64", "64", "64", "0", "1", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "1024", "4096" ], + "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "2", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "1024", "4096" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "0", "F", "64", "64", "64", "1", "2", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/128x768x2304.json b/benchmarks/config/matmul/128x768x2304.json index f3fc16452..8176fbcd1 100644 --- a/benchmarks/config/matmul/128x768x2304.json +++ b/benchmarks/config/matmul/128x768x2304.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "0", "F", "64", "64", "48", "0", "2304", "768" ], + "flags": [ "100", "128", "0", "F", "64", "64", "48", "0", "1", "2304", "768" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "0", "F", "64", "64", "48", "0", "2304", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "0", "F", "64", "64", "48", "0", "1", "2304", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "0", "F", "64", "64", "48", "1", "2304", "768" ], + "flags": [ "100", "128", "0", "F", "64", "64", "48", "1", "2", "2304", "768" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "0", "F", "64", "64", "48", "1", "2304", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "0", "F", "64", "64", "48", "1", "2", "2304", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/128x768x3072.json b/benchmarks/config/matmul/128x768x3072.json index d1ec88092..a20543301 100644 --- a/benchmarks/config/matmul/128x768x3072.json +++ b/benchmarks/config/matmul/128x768x3072.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "0", "F", "32", "32", "48", "0", "3072", "768" ], + "flags": [ "100", "128", "0", "F", "32", "32", "48", "0", "1", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "0", "F", "32", "32", "48", "0", "3072", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "0", "F", "32", "32", "48", "0", "1", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "0", "F", "32", "32", "48", "1", "3072", "768" ], + "flags": [ "100", "128", "0", "F", "32", "32", "48", "1", "2", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "0", "F", "32", "32", "48", "1", "3072", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "0", "F", "32", "32", "48", "1", "2", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/128x768x768.json b/benchmarks/config/matmul/128x768x768.json index 5f6e40852..27eaf67a5 100644 --- a/benchmarks/config/matmul/128x768x768.json +++ b/benchmarks/config/matmul/128x768x768.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "0", "F", "32", "64", "64", "0", "768", "768" ], + "flags": [ "100", "128", "0", "F", "32", "64", "64", "0", "1", "768", "768" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "0", "F", "32", "64", "64", "0", "768", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "0", "F", "32", "64", "64", "0", "1", "768", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "128", "0", "F", "32", "64", "64", "1", "768", "768" ], + "flags": [ "100", "128", "0", "F", "32", "64", "64", "1", "2", "768", "768" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "128", "0", "F", "32", "64", "64", "1", "768", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "128", "0", "F", "32", "64", "64", "1", "2", "768", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/256x1024x1024.json b/benchmarks/config/matmul/256x1024x1024.json index 14371a52b..7df79fd5f 100644 --- a/benchmarks/config/matmul/256x1024x1024.json +++ b/benchmarks/config/matmul/256x1024x1024.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1024", "1024" ], + "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "1024", "1024" ], + "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "2", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "2", "1024", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/256x1024x4096.json b/benchmarks/config/matmul/256x1024x4096.json index e7782042e..984f37d15 100644 --- a/benchmarks/config/matmul/256x1024x4096.json +++ b/benchmarks/config/matmul/256x1024x4096.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "4096", "1024" ], + "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "4096", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "4096", "1024" ], + "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "2", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "4096", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "2", "4096", "1024" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/256x3072x768.json b/benchmarks/config/matmul/256x3072x768.json index c1dec0325..01e7727f1 100644 --- a/benchmarks/config/matmul/256x3072x768.json +++ b/benchmarks/config/matmul/256x3072x768.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "768", "3072" ], + "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "768", "3072" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "768", "3072" ], + "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "2", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "768", "3072" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "2", "768", "3072" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/256x4096x1024.json b/benchmarks/config/matmul/256x4096x1024.json index 2ff90c36a..db2108d9a 100644 --- a/benchmarks/config/matmul/256x4096x1024.json +++ b/benchmarks/config/matmul/256x4096x1024.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1024", "4096" ], + "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1024", "4096" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "1024", "4096" ], + "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "2", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "1024", "4096" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "2", "1024", "4096" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/256x768x3072.json b/benchmarks/config/matmul/256x768x3072.json index 2b8cd3195..3558ec489 100644 --- a/benchmarks/config/matmul/256x768x3072.json +++ b/benchmarks/config/matmul/256x768x3072.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "3072", "768" ], + "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "3072", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "3072", "768" ], + "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "2", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "3072", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "2", "3072", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/matmul/256x768x768.json b/benchmarks/config/matmul/256x768x768.json index 93f76fe20..de4991004 100644 --- a/benchmarks/config/matmul/256x768x768.json +++ b/benchmarks/config/matmul/256x768x768.json @@ -5,14 +5,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "768", "768" ], + "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1", "768", "768" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "768", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "64", "64", "64", "0", "1", "768", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, @@ -22,14 +22,14 @@ "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", "environment": { "OMP_NUM_THREADS": "1" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "768", "768" ], + "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "2", "768", "768" ], "extensions": [ "(avx2|asimd)" ] }, "matmul_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "768", "768" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "64", "64", "64", "1", "2", "768", "768" ], "extensions": [ "(avx2|asimd)" ] } }}, diff --git a/benchmarks/config/omp/dnn-bf16.json b/benchmarks/config/omp/dnn-bf16.json index f03655586..d96532de1 100644 --- a/benchmarks/config/omp/dnn-bf16.json +++ b/benchmarks/config/omp/dnn-bf16.json @@ -4,29 +4,29 @@ "gemm_bf16_omp_2_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "2", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "32", "32", "32", "1", "1", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "2", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "32", "32", "32", "1", "2", "1024", "1024", "1024", "1024" ], "extensions": [] }, "gemm_bf16_omp_4_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "4", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "32", "32", "32", "1", "1", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "4", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "32", "32", "32", "1", "2", "1024", "1024", "1024", "1024" ], "extensions": [] }, "gemm_bf16_omp_8_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "8", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "32", "32", "32", "1", "1", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "8", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "32", "32", "32", "1", "2", "1024", "1024", "1024", "1024" ], "extensions": [] }, "gemm_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "32", "32", "32", "1", "1", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "32", "32", "32", "1", "2", "1024", "1024", "1024", "1024" ], "extensions": [] } }}, @@ -35,29 +35,29 @@ "mlp_bf16_omp_2_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "2", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "5", "F", "32", "32", "32", "1", "1", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "2", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "5", "F", "32", "32", "32", "1", "2", "1024", "1024", "1024", "1024" ], "extensions": [] }, "mlp_bf16_omp_4_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "4", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "5", "F", "32", "32", "32", "1", "1", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "4", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "5", "F", "32", "32", "32", "1", "2", "1024", "1024", "1024", "1024" ], "extensions": [] }, "mlp_bf16_omp_8_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "8", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "5", "F", "32", "32", "32", "1", "1", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "8", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "5", "F", "32", "32", "32", "1", "2", "1024", "1024", "1024", "1024" ], "extensions": [] }, "mlp_bf16_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "5", "F", "32", "32", "32", "1", "1", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "5", "F", "32", "32", "32", "1", "2", "1024", "1024", "1024", "1024" ], "extensions": [] } }} diff --git a/benchmarks/config/omp/dnn-fp32.json b/benchmarks/config/omp/dnn-fp32.json index c68de1b5f..d8eedbcab 100644 --- a/benchmarks/config/omp/dnn-fp32.json +++ b/benchmarks/config/omp/dnn-fp32.json @@ -4,29 +4,29 @@ "gemm_fp32_omp_2_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "2", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "32", "32", "32", "0", "0", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "2", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "32", "32", "32", "0", "1", "1024", "1024", "1024", "1024" ], "extensions": [] }, "gemm_fp32_omp_4_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "4", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "32", "32", "32", "0", "0", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "4", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "32", "32", "32", "0", "1", "1024", "1024", "1024", "1024" ], "extensions": [] }, "gemm_fp32_omp_8_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "8", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "32", "32", "32", "0", "0", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "8", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "32", "32", "32", "0", "1", "1024", "1024", "1024", "1024" ], "extensions": [] }, "gemm_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "0", "F", "32", "32", "32", "0", "0", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "0", "F", "32", "32", "32", "0", "1", "1024", "1024", "1024", "1024" ], "extensions": [] } }}, @@ -35,29 +35,29 @@ "mlp_fp32_omp_2_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "2", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "5", "F", "32", "32", "32", "0", "0", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "2", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "5", "F", "32", "32", "32", "0", "1", "1024", "1024", "1024", "1024" ], "extensions": [] }, "mlp_fp32_omp_4_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "4", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "5", "F", "32", "32", "32", "0", "0", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "4", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "5", "F", "32", "32", "32", "0", "1", "1024", "1024", "1024", "1024" ], "extensions": [] }, "mlp_fp32_omp_8_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "8", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "5", "F", "32", "32", "32", "0", "0", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "8", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "5", "F", "32", "32", "32", "0", "1", "1024", "1024", "1024", "1024" ], "extensions": [] }, "mlp_fp32_omp_16_dnn": { "type": "XSMM-DNN", "benchmark": "xsmm_dnn_mlp", - "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0" }, - "flags": [ "100", "256", "5", "F", "32", "32", "32", "0", "0", "1024", "1024", "1024", "1024" ], + "environment": { "OMP_NUM_THREADS": "16", "KMP_AFFINITY": "granularity=fine,verbose,compact,1,0", "LIBXSMM_DNN_DISABLE_SFC": "1" }, + "flags": [ "100", "256", "5", "F", "32", "32", "32", "0", "1", "1024", "1024", "1024", "1024" ], "extensions": [] } }} diff --git a/cmake/modules/xsmm-dnn.cmake b/cmake/modules/xsmm-dnn.cmake index c72c8d512..5a748788a 100644 --- a/cmake/modules/xsmm-dnn.cmake +++ b/cmake/modules/xsmm-dnn.cmake @@ -16,8 +16,8 @@ else() FetchContent_Declare( xsmm_dnn - URL https://github.com/libxsmm/libxsmm-dnn/archive/6e042517be4b975f431fb92d803ddc5d9dafcc64.tar.gz - URL_HASH SHA256=2d92577b865a823ec223a675bea1537a61b717a2b8fb9f06dd7e46a19183df60 + URL https://github.com/libxsmm/libxsmm-dnn/archive/6757af7d214a49b06e5289b9784b4b99b4a35aaf.tar.gz + URL_HASH SHA256=f7c879f2aed6a1562f3e15ea57637845181c6ae4267ec71975ab7b11cd16dfa4 ) FetchContent_GetProperties(xsmm_dnn) diff --git a/cmake/modules/xsmm.cmake b/cmake/modules/xsmm.cmake index 4b683a22a..d1d845afc 100644 --- a/cmake/modules/xsmm.cmake +++ b/cmake/modules/xsmm.cmake @@ -13,8 +13,8 @@ else() FetchContent_Declare( xsmm - URL https://github.com/libxsmm/libxsmm/archive/c5fe65139050d7460d6929b2b507e3e349687477.tar.gz - URL_HASH SHA256=c55467f2440e5d894907a6297e02c1f5c9f96fefa27a807fca3ad60bf7f4c5ab + URL https://github.com/libxsmm/libxsmm/archive/85851d4368f730069086e5acf65eaa3ae3e80852.tar.gz + URL_HASH SHA256=5b63f968ca0c618e0edf6760399a72c07cabed88b103e0ced11e30b321a824b7 ) FetchContent_GetProperties(xsmm)