devtools/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2026-04-10 11:34:33 +08:00
Code Issues Packages Projects Releases Wiki Activity

Clang-format tests, examples, libraries, benchmarks, etc.

Browse Source
This commit is contained in:
Antonio Sánchez
2023-12-05 21:22:55 +00:00
committed by Rasmus Munk Larsen
parent 3252ecc7a4
commit 46e9cdb7fe
876 changed files with 33453 additions and 37795 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
bench/tensors/tensor_benchmarks_cpu.cc
View File

@@ -4,9 +4,9 @@
#include "tensor_benchmarks.h"
#define CREATE_THREAD_POOL(threads) \
Eigen::ThreadPool pool(threads); \
Eigen::ThreadPoolDevice device(&pool, threads);
#define CREATE_THREAD_POOL(threads) \
Eigen::ThreadPool pool(threads); \
Eigen::ThreadPoolDevice device(&pool, threads);
// Simple functions
#define BM_FuncCPU(FUNC, THREADS) \
@@ -78,7 +78,6 @@ BM_FuncCPU(colReduction, 4);
BM_FuncCPU(colReduction, 8);
BM_FuncCPU(colReduction, 12);
// Contractions
#define BM_FuncWithInputDimsCPU(FUNC, D1, D2, D3, THREADS) \
static void BM_##FUNC##_##D1##x##D2##x##D3##_##THREADS##T(int iters, int N) { \
@@ -95,7 +94,6 @@ BM_FuncCPU(colReduction, 12);
} \
BENCHMARK_RANGE(BM_##FUNC##_##D1##x##D2##x##D3##_##THREADS##T, 10, 5000);
BM_FuncWithInputDimsCPU(contraction, N, N, N, 1);
BM_FuncWithInputDimsCPU(contraction, N, N, N, 4);
BM_FuncWithInputDimsCPU(contraction, N, N, N, 8);
@@ -132,15 +130,14 @@ BM_FuncWithInputDimsCPU(contraction, N, N, 1, 8);
BM_FuncWithInputDimsCPU(contraction, N, N, 1, 12);
BM_FuncWithInputDimsCPU(contraction, N, N, 1, 16);
// Convolutions
#define BM_FuncWithKernelDimsCPU(FUNC, DIM1, DIM2, THREADS) \
static void BM_##FUNC##_##DIM1##x##DIM2##_##THREADS##T(int iters, int N) { \
StopBenchmarkTiming(); \
CREATE_THREAD_POOL(THREADS); \
BenchmarkSuite<Eigen::ThreadPoolDevice, float> suite(device, N); \
suite.FUNC(iters, DIM1, DIM2); \
} \
#define BM_FuncWithKernelDimsCPU(FUNC, DIM1, DIM2, THREADS) \
static void BM_##FUNC##_##DIM1##x##DIM2##_##THREADS##T(int iters, int N) { \
StopBenchmarkTiming(); \
CREATE_THREAD_POOL(THREADS); \
BenchmarkSuite<Eigen::ThreadPoolDevice, float> suite(device, N); \
suite.FUNC(iters, DIM1, DIM2); \
} \
BENCHMARK_RANGE(BM_##FUNC##_##DIM1##x##DIM2##_##THREADS##T, 128, 5000);
BM_FuncWithKernelDimsCPU(convolution, 7, 1, 4);