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

merging the CUDA and HIP implementation for the Tensor directory and the unit tests

Browse Source
This commit is contained in:
Deven Desai
2018-06-20 16:44:58 -04:00
parent cfdabbcc8f
commit 1bb6fa99a3
22 changed files with 1208 additions and 804 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
unsupported/test/cxx11_tensor_random_gpu.cu
View File

@@ -9,15 +9,16 @@
#define EIGEN_TEST_NO_LONGDOUBLE
#define EIGEN_TEST_NO_COMPLEX
#define EIGEN_TEST_FUNC cxx11_tensor_random_cuda
#define EIGEN_TEST_FUNC cxx11_tensor_random_gpu
#define EIGEN_DEFAULT_DENSE_INDEX_TYPE int
#define EIGEN_USE_GPU
#include "main.h"
#include <Eigen/CXX11/Tensor>
#include <Eigen/CXX11/src/Tensor/TensorGpuHipCudaDefines.h>
void test_cuda_random_uniform()
void test_gpu_random_uniform()
{
Tensor<float, 2> out(72,97);
out.setZero();
@@ -25,24 +26,24 @@ void test_cuda_random_uniform()
std::size_t out_bytes = out.size() * sizeof(float);
float* d_out;
cudaMalloc((void**)(&d_out), out_bytes);
gpuMalloc((void**)(&d_out), out_bytes);
Eigen::CudaStreamDevice stream;
Eigen::GpuStreamDevice stream;
Eigen::GpuDevice gpu_device(&stream);
Eigen::TensorMap<Eigen::Tensor<float, 2> > gpu_out(d_out, 72,97);
gpu_out.device(gpu_device) = gpu_out.random();
assert(cudaMemcpyAsync(out.data(), d_out, out_bytes, cudaMemcpyDeviceToHost, gpu_device.stream()) == cudaSuccess);
assert(cudaStreamSynchronize(gpu_device.stream()) == cudaSuccess);
assert(gpuMemcpyAsync(out.data(), d_out, out_bytes, gpuMemcpyDeviceToHost, gpu_device.stream()) == gpuSuccess);
assert(gpuStreamSynchronize(gpu_device.stream()) == gpuSuccess);
// For now we just check this code doesn't crash.
// TODO: come up with a valid test of randomness
}
void test_cuda_random_normal()
void test_gpu_random_normal()
{
Tensor<float, 2> out(72,97);
out.setZero();
@@ -50,9 +51,9 @@ void test_cuda_random_normal()
std::size_t out_bytes = out.size() * sizeof(float);
float* d_out;
cudaMalloc((void**)(&d_out), out_bytes);
gpuMalloc((void**)(&d_out), out_bytes);
Eigen::CudaStreamDevice stream;
Eigen::GpuStreamDevice stream;
Eigen::GpuDevice gpu_device(&stream);
Eigen::TensorMap<Eigen::Tensor<float, 2> > gpu_out(d_out, 72,97);
@@ -60,8 +61,8 @@ void test_cuda_random_normal()
Eigen::internal::NormalRandomGenerator<float> gen(true);
gpu_out.device(gpu_device) = gpu_out.random(gen);
assert(cudaMemcpyAsync(out.data(), d_out, out_bytes, cudaMemcpyDeviceToHost, gpu_device.stream()) == cudaSuccess);
assert(cudaStreamSynchronize(gpu_device.stream()) == cudaSuccess);
assert(gpuMemcpyAsync(out.data(), d_out, out_bytes, gpuMemcpyDeviceToHost, gpu_device.stream()) == gpuSuccess);
assert(gpuStreamSynchronize(gpu_device.stream()) == gpuSuccess);
}
static void test_complex()
@@ -77,9 +78,9 @@ static void test_complex()
}
void test_cxx11_tensor_random_cuda()
void test_cxx11_tensor_random_gpu()
{
CALL_SUBTEST(test_cuda_random_uniform());
CALL_SUBTEST(test_cuda_random_normal());
CALL_SUBTEST(test_gpu_random_uniform());
CALL_SUBTEST(test_gpu_random_normal());
CALL_SUBTEST(test_complex());
}