mirror of
https://github.com/pytorch/pytorch.git
synced 2025-10-21 05:34:18 +08:00
c6e9e9359f
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/39023 Reviewed By: orionr Differential Revision: D21702529 fbshipit-source-id: 6945bba95609102409850b105a8a091e33b8acc9
63 lines
1.7 KiB
C++
63 lines
1.7 KiB
C++
#include <gtest/gtest.h>
|
|
|
|
#include <torch/torch.h>
|
|
#include <ATen/native/Pow.h>
|
|
#include <torch/types.h>
|
|
#include <torch/utils.h>
|
|
#include <test/cpp/api/support.h>
|
|
#include <iostream>
|
|
#include <vector>
|
|
#include <type_traits>
|
|
#include <cstdlib>
|
|
|
|
|
|
struct DispatchTest : torch::test::SeedingFixture {};
|
|
|
|
TEST_F(DispatchTest, TestAVX2) {
|
|
const std::vector<int> ints {1, 2, 3, 4};
|
|
const std::vector<int> result {1, 4, 27, 256};
|
|
const auto vals_tensor = torch::tensor(ints);
|
|
const auto pows_tensor = torch::tensor(ints);
|
|
#ifdef _WIN32
|
|
_putenv("ATEN_CPU_CAPABILITY=avx2");
|
|
#else
|
|
setenv("ATEN_CPU_CAPABILITY", "avx2", 1);
|
|
#endif
|
|
const auto actual_pow_avx2 = vals_tensor.pow(pows_tensor);
|
|
for (int i = 0; i < 4; i++) {
|
|
ASSERT_EQ(result[i], actual_pow_avx2[i].item<int>());
|
|
}
|
|
}
|
|
|
|
TEST_F(DispatchTest, TestAVX) {
|
|
const std::vector<int> ints {1, 2, 3, 4};
|
|
const std::vector<int> result {1, 4, 27, 256};
|
|
const auto vals_tensor = torch::tensor(ints);
|
|
const auto pows_tensor = torch::tensor(ints);
|
|
#ifdef _WIN32
|
|
_putenv("ATEN_CPU_CAPABILITY=avx");
|
|
#else
|
|
setenv("ATEN_CPU_CAPABILITY", "avx", 1);
|
|
#endif
|
|
const auto actual_pow_avx = vals_tensor.pow(pows_tensor);
|
|
for (int i = 0; i < 4; i++) {
|
|
ASSERT_EQ(result[i], actual_pow_avx[i].item<int>());
|
|
}
|
|
}
|
|
|
|
TEST_F(DispatchTest, TestDefault) {
|
|
const std::vector<int> ints {1, 2, 3, 4};
|
|
const std::vector<int> result {1, 4, 27, 256};
|
|
const auto vals_tensor = torch::tensor(ints);
|
|
const auto pows_tensor = torch::tensor(ints);
|
|
#ifdef _WIN32
|
|
_putenv("ATEN_CPU_CAPABILITY=default");
|
|
#else
|
|
setenv("ATEN_CPU_CAPABILITY", "default", 1);
|
|
#endif
|
|
const auto actual_pow_default = vals_tensor.pow(pows_tensor);
|
|
for (int i = 0; i < 4; i++) {
|
|
ASSERT_EQ(result[i], actual_pow_default[i].item<int>());
|
|
}
|
|
}
|