mixed_vector_test.cu 2.5 KB
Newer Older
D
dzhwinter 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13
/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.

   Licensed under the Apache License, Version 2.0 (the "License");
   you may not use this file except in compliance with the License.
   You may obtain a copy of the License at

   http://www.apache.org/licenses/LICENSE-2.0

   Unless required by applicable law or agreed to in writing, software
   distributed under the License is distributed on an "AS IS" BASIS,
   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
   See the License for the specific language governing permissions and
   limitations under the License. */
Y
Yu Yang 已提交
14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83
#include <cuda_runtime.h>

#include "glog/logging.h"
#include "gtest/gtest.h"
#include "paddle/framework/mixed_vector.h"
#include "paddle/platform/gpu_info.h"

template <typename T>
using vec = paddle::framework::Vector<T>;

TEST(mixed_vector, CPU_VECTOR) {
  vec<int> tmp;
  for (int i = 0; i < 10; ++i) {
    tmp.push_back(i);
  }
  ASSERT_EQ(tmp.size(), 10);
  vec<int> tmp2;
  tmp2 = tmp;
  ASSERT_EQ(tmp2.size(), 10);
  for (int i = 0; i < 10; ++i) {
    ASSERT_EQ(tmp2[i], i);
    ASSERT_EQ(tmp2[i], tmp[i]);
  }
  int cnt = 0;
  for (auto& t : tmp2) {
    ASSERT_EQ(t, cnt);
    ++cnt;
  }
}

static __global__ void multiply_10(int* ptr) {
  for (int i = 0; i < 10; ++i) {
    ptr[i] *= 10;
  }
}

cudaStream_t GetCUDAStream(paddle::platform::CUDAPlace place) {
  return reinterpret_cast<const paddle::platform::CUDADeviceContext*>(
             paddle::platform::DeviceContextPool::Instance().Get(place))
      ->stream();
}

TEST(mixed_vector, GPU_VECTOR) {
  vec<int> tmp;
  for (int i = 0; i < 10; ++i) {
    tmp.push_back(i);
  }
  ASSERT_EQ(tmp.size(), 10);
  paddle::platform::CUDAPlace gpu(0);

  multiply_10<<<1, 1, 0, GetCUDAStream(gpu)>>>(tmp.MutableData(gpu));

  for (int i = 0; i < 10; ++i) {
    ASSERT_EQ(tmp[i], i * 10);
  }
}

TEST(mixed_vector, MultiGPU) {
  if (paddle::platform::GetCUDADeviceCount() < 2) {
    LOG(WARNING) << "Skip mixed_vector.MultiGPU since there are not multiple "
                    "GPUs in your machine.";
    return;
  }

  vec<int> tmp;
  for (int i = 0; i < 10; ++i) {
    tmp.push_back(i);
  }
  ASSERT_EQ(tmp.size(), 10);
  paddle::platform::CUDAPlace gpu0(0);
Y
Fix CI  
Yu Yang 已提交
84
  paddle::platform::SetDeviceId(0);
Y
Yu Yang 已提交
85 86
  multiply_10<<<1, 1, 0, GetCUDAStream(gpu0)>>>(tmp.MutableData(gpu0));
  paddle::platform::CUDAPlace gpu1(1);
Y
Fix CI  
Yu Yang 已提交
87 88 89
  auto* gpu1_ptr = tmp.MutableData(gpu1);
  paddle::platform::SetDeviceId(1);
  multiply_10<<<1, 1, 0, GetCUDAStream(gpu1)>>>(gpu1_ptr);
Y
Yu Yang 已提交
90 91 92 93
  for (int i = 0; i < 10; ++i) {
    ASSERT_EQ(tmp[i], i * 100);
  }
}