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

Vectorized the evaluation of tensor expression (using SSE, AVX, NEON, ...)

Browse Source 
Added the ability to parallelize the evaluation of a tensor expression over multiple cpu cores.
Added the ability to offload the evaluation of a tensor expression to a GPU.
This commit is contained in:
Benoit Steiner
2014-05-16 15:08:05 -07:00
parent 0320f7e3a7
commit 7402fea0a8
17 changed files with 719 additions and 65 deletions
Download Patch File Download Diff File Expand all files Collapse all files

37
unsupported/test/cxx11_tensor_thread_pool.cpp Normal file
View File

@@ -0,0 +1,37 @@
// This file is part of Eigen, a lightweight C++ template library
// for linear algebra.
//
// Copyright (C) 2014 Benoit Steiner <benoit.steiner.goog@gmail.com>
//
// This Source Code Form is subject to the terms of the Mozilla
// Public License v. 2.0. If a copy of the MPL was not distributed
// with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
#define EIGEN_USE_THREADS
#include "main.h"
#include <Eigen/CXX11/Tensor>
using Eigen::Tensor;
void test_cxx11_tensor_thread_pool()
{
Eigen::Tensor<float, 3> in1(Eigen::array<ptrdiff_t, 3>(2,3,7));
Eigen::Tensor<float, 3> in2(Eigen::array<ptrdiff_t, 3>(2,3,7));
Eigen::Tensor<float, 3> out(Eigen::array<ptrdiff_t, 3>(2,3,7));
in1.setRandom();
in2.setRandom();
Eigen::ThreadPoolDevice thread_pool_device(3);
out.device(thread_pool_device) = in1 + in2 * 3.14;
for (int i = 0; i < 2; ++i) {
for (int j = 0; j < 3; ++j) {
for (int k = 0; k < 7; ++k) {
VERIFY_IS_APPROX(out(Eigen::array<ptrdiff_t, 3>(i,j,k)), in1(Eigen::array<ptrdiff_t, 3>(i,j,k)) + in2(Eigen::array<ptrdiff_t, 3>(i,j,k)) * 3.14f);
}
}
}
}