Added benchmarks for contraction on CPU.

bench/tensors/README

@@ -1,4 +1,6 @@
-Each benchmark comes in 2 flavors: one that runs on CPU, and one that runs on GPU.
+The tensor benchmark suite is made of several parts.
+
+The first part is a generic suite, in which each benchmark comes in 2 flavors: one that runs on CPU, and one that runs on GPU.
 
 To compile the floating point CPU benchmarks, simply call:
 g++ tensor_benchmarks_cpu.cc benchmark_main.cc -I ../../ -std=c++11 -O3 -DNDEBUG -pthread -mavx -o benchmarks_cpu
@@ -6,7 +8,8 @@ g++ tensor_benchmarks_cpu.cc benchmark_main.cc -I ../../ -std=c++11 -O3 -DNDEBUG
 To compile the floating point GPU benchmarks, simply call:
 nvcc tensor_benchmarks_gpu.cu benchmark_main.cc -I ../../ -std=c++11 -O2 -DNDEBUG -arch compute_35 -o benchmarks_gpu
 
-
+We also provide a version of the generic GPU tensor benchmarks that uses half floats (aka fp16) instead of regular floats. To compile these benchmarks, simply call the command line below. You'll need a recent GPU that supports compute capability 5.3 or higher to run them and nvcc 7.5 or higher to compile the code.
-To compile the half float GPU benchmarks, simply call the command line below. You'll need a recent GPU that supports compute capability 5.3 or higher to run them and nvcc 7.5 or higher to compile the code.
 nvcc tensor_benchmarks_fp16_gpu.cu benchmark_main.cc -I ../../ -std=c++11 -O2 -DNDEBUG -arch compute_53 -o benchmarks_fp16_gpu
 
+last but not least, we also provide a suite of benchmarks to measure the scalability of the contraction code on CPU. To compile these benchmarks, call 
+g++ contraction_benchmarks_cpu.cc benchmark_main.cc -I ../../ -std=c++11 -O3 -DNDEBUG -pthread -mavx -o benchmarks_cpu

bench/tensors/contraction_benchmarks_cpu.cc

@@ -0,0 +1,39 @@
+#define EIGEN_USE_THREADS
+
+#include <string>
+
+#include "tensor_benchmarks.h"
+
+#define CREATE_THREAD_POOL(threads)             \
+Eigen::ThreadPool pool(threads);                \
+Eigen::ThreadPoolDevice device(&pool, threads);
+
+
+// Contractions for number of threads ranging from 1 to 32
+// Dimensions are Rows, Cols, Depth
+#define BM_ContractionCPU(D1, D2, D3)                                         \
+  static void BM_##Contraction##_##D1##x##D2##x##D3(int iters, int Threads) { \
+    StopBenchmarkTiming();                                                    \
+    CREATE_THREAD_POOL(Threads);                                              \
+    BenchmarkSuite<Eigen::ThreadPoolDevice, float> suite(device, D1, D2, D3); \
+    suite.contraction(iters);                                                 \
+  }                                                                           \
+  BENCHMARK_RANGE(BM_##Contraction##_##D1##x##D2##x##D3, 1, 32);
+
+
+// Vector Matrix and Matrix Vector products
+BM_ContractionCPU(1, 2000, 500);
+BM_ContractionCPU(2000, 1, 500);
+
+// Various skinny matrices
+BM_ContractionCPU(250, 3, 512);
+BM_ContractionCPU(1500, 3, 512);
+
+BM_ContractionCPU(512, 800, 4);
+BM_ContractionCPU(512, 80, 800);
+BM_ContractionCPU(512, 80, 13522);
+BM_ContractionCPU(1, 80, 13522);
+
+BM_ContractionCPU(3200, 512, 4);
+BM_ContractionCPU(3200, 512, 80);
+BM_ContractionCPU(3200, 80, 512);