1 // This file is part of Eigen, a lightweight C++ template library
2 // for linear algebra.
3 //
4 // Copyright (C) 2016 Benoit Steiner <benoit.steiner.goog@gmail.com>
5 //
6 // This Source Code Form is subject to the terms of the Mozilla
7 // Public License v. 2.0. If a copy of the MPL was not distributed
8 // with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
9
10 #define EIGEN_TEST_NO_LONGDOUBLE
11 #define EIGEN_TEST_NO_COMPLEX
12
13 #define EIGEN_DEFAULT_DENSE_INDEX_TYPE int
14 #define EIGEN_USE_GPU
15
16 #include "main.h"
17 #include <unsupported/Eigen/CXX11/Tensor>
18
19 #include <Eigen/CXX11/src/Tensor/TensorGpuHipCudaDefines.h>
20
21 using Eigen::Tensor;
22 typedef Tensor<float, 1>::DimensionPair DimPair;
23
24 template<int DataLayout>
test_gpu_cumsum(int m_size,int k_size,int n_size)25 void test_gpu_cumsum(int m_size, int k_size, int n_size)
26 {
27 std::cout << "Testing for (" << m_size << "," << k_size << "," << n_size << ")" << std::endl;
28 Tensor<float, 3, DataLayout> t_input(m_size, k_size, n_size);
29 Tensor<float, 3, DataLayout> t_result(m_size, k_size, n_size);
30 Tensor<float, 3, DataLayout> t_result_gpu(m_size, k_size, n_size);
31
32 t_input.setRandom();
33
34 std::size_t t_input_bytes = t_input.size() * sizeof(float);
35 std::size_t t_result_bytes = t_result.size() * sizeof(float);
36
37 float* d_t_input;
38 float* d_t_result;
39
40 gpuMalloc((void**)(&d_t_input), t_input_bytes);
41 gpuMalloc((void**)(&d_t_result), t_result_bytes);
42
43 gpuMemcpy(d_t_input, t_input.data(), t_input_bytes, gpuMemcpyHostToDevice);
44
45 Eigen::GpuStreamDevice stream;
46 Eigen::GpuDevice gpu_device(&stream);
47
48 Eigen::TensorMap<Eigen::Tensor<float, 3, DataLayout> >
49 gpu_t_input(d_t_input, Eigen::array<int, 3>(m_size, k_size, n_size));
50 Eigen::TensorMap<Eigen::Tensor<float, 3, DataLayout> >
51 gpu_t_result(d_t_result, Eigen::array<int, 3>(m_size, k_size, n_size));
52
53 gpu_t_result.device(gpu_device) = gpu_t_input.cumsum(1);
54 t_result = t_input.cumsum(1);
55
56 gpuMemcpy(t_result_gpu.data(), d_t_result, t_result_bytes, gpuMemcpyDeviceToHost);
57 for (DenseIndex i = 0; i < t_result.size(); i++) {
58 if (fabs(t_result(i) - t_result_gpu(i)) < 1e-4f) {
59 continue;
60 }
61 if (Eigen::internal::isApprox(t_result(i), t_result_gpu(i), 1e-4f)) {
62 continue;
63 }
64 std::cout << "mismatch detected at index " << i << ": " << t_result(i)
65 << " vs " << t_result_gpu(i) << std::endl;
66 assert(false);
67 }
68
69 gpuFree((void*)d_t_input);
70 gpuFree((void*)d_t_result);
71 }
72
73
EIGEN_DECLARE_TEST(cxx11_tensor_scan_gpu)74 EIGEN_DECLARE_TEST(cxx11_tensor_scan_gpu)
75 {
76 CALL_SUBTEST_1(test_gpu_cumsum<ColMajor>(128, 128, 128));
77 CALL_SUBTEST_2(test_gpu_cumsum<RowMajor>(128, 128, 128));
78 }
79