device_context.cc 9.3 KB
Newer Older
1
/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved.
Q
qijun 已提交
2 3 4 5 6 7 8 9 10
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
Yi Wang 已提交
11
#include "paddle/fluid/platform/device_context.h"
12

13
#include <set>
14
#include <string>
Y
Yu Yang 已提交
15
#include <unordered_set>
16 17
#include <vector>

Y
Yi Wang 已提交
18
#include "paddle/fluid/memory/memory.h"
19 20 21
#ifdef PADDLE_WITH_CUDA
#include "paddle/fluid/framework/rw_lock.h"
#endif
22

Q
qijun 已提交
23 24 25
namespace paddle {
namespace platform {

D
dzhwinter 已提交
26 27
DeviceContextPool* DeviceContextPool::pool = nullptr;

Y
Yu Yang 已提交
28
platform::DeviceContext* DeviceContextPool::Get(const platform::Place& place) {
D
dzhwinter 已提交
29 30 31 32 33 34
  auto it = device_contexts_.find(place);
  if (it == device_contexts_.end()) {
    PADDLE_THROW(
        "'Place' is not supported, Please re-compile with WITH_GPU "
        "option");
  }
Y
Yu Yang 已提交
35
  return it->second.get();
D
dzhwinter 已提交
36 37 38 39 40
}

DeviceContextPool::DeviceContextPool(
    const std::vector<platform::Place>& places) {
  PADDLE_ENFORCE_GT(places.size(), 0);
Y
Yu Yang 已提交
41
  using PtrType = std::unique_ptr<DeviceContext>;
42
  std::set<Place> set;
Y
Yu Yang 已提交
43 44 45 46 47 48
  for (auto& p : places) {
    set.insert(p);
  }

  for (auto& p : set) {
    if (platform::is_cpu_place(p)) {
49
#ifdef PADDLE_WITH_MKLDNN
Y
Yu Yang 已提交
50 51
      device_contexts_.emplace(
          p, PtrType(new MKLDNNDeviceContext(boost::get<CPUPlace>(p))));
52
#else
Y
Yu Yang 已提交
53 54
      device_contexts_.emplace(
          p, PtrType(new CPUDeviceContext(boost::get<CPUPlace>(p))));
55
#endif
Y
Yu Yang 已提交
56
    } else if (platform::is_gpu_place(p)) {
D
dzhwinter 已提交
57
#ifdef PADDLE_WITH_CUDA
Y
Yu Yang 已提交
58 59
      device_contexts_.emplace(
          p, PtrType(new CUDADeviceContext(boost::get<CUDAPlace>(p))));
D
dzhwinter 已提交
60 61
#else
      PADDLE_THROW(
D
dzhwinter 已提交
62
          "'CUDAPlace' is not supported, Please re-compile with WITH_GPU "
D
dzhwinter 已提交
63
          "option");
C
chengduoZH 已提交
64 65 66 67 68 69 70 71 72 73
#endif
    } else if (platform::is_cuda_pinned_place(p)) {
#ifdef PADDLE_WITH_CUDA
      device_contexts_.emplace(
          p,
          PtrType(new CUDAPinnedDeviceContext(boost::get<CUDAPinnedPlace>(p))));
#else
      PADDLE_THROW(
          "'CUDAPlace' is not supported, Please re-compile with WITH_GPU "
          "option");
D
dzhwinter 已提交
74 75 76 77 78
#endif
    }
  }
}

79 80 81 82
CPUDeviceContext::CPUDeviceContext() {
  eigen_device_.reset(new Eigen::DefaultDevice());
}

D
dzhwinter 已提交
83
CPUDeviceContext::CPUDeviceContext(CPUPlace place) : place_(place) {
84 85 86 87 88 89 90
  eigen_device_.reset(new Eigen::DefaultDevice());
}

Eigen::DefaultDevice* CPUDeviceContext::eigen_device() const {
  return eigen_device_.get();
}

D
dzhwinter 已提交
91
Place CPUDeviceContext::GetPlace() const { return place_; }
92

93
#ifdef PADDLE_WITH_CUDA
94

Q
init  
qijun 已提交
95 96 97 98 99 100 101
class EigenCudaStreamDevice : public Eigen::StreamInterface {
 public:
  EigenCudaStreamDevice() : scratch_(nullptr), semaphore_(nullptr) {
    Eigen::initializeDeviceProp();
  }
  ~EigenCudaStreamDevice() override {}

D
dzhwinter 已提交
102
  void Reinitialize(const cudaStream_t* cuda_stream, CUDAPlace place) {
Q
init  
qijun 已提交
103 104 105 106 107 108 109 110 111 112 113 114
    stream_ = cuda_stream;
    place_ = place;
    device_prop_ = &Eigen::m_deviceProperties[place.device];
  }

  const cudaStream_t& stream() const override { return *stream_; }

  const cudaDeviceProp& deviceProperties() const override {
    return *device_prop_;
  }

  void* allocate(size_t num_bytes) const override {
115 116 117 118 119
    auto buf =
        paddle::memory::Alloc(place_, num_bytes, memory::Allocator::kTiny);
    void* retv = buf->ptr();
    allocations_[buf->ptr()] = std::move(buf);
    return retv;
Q
init  
qijun 已提交
120 121 122
  }

  void deallocate(void* buffer) const override {
123
    allocations_.erase(allocations_.find(buffer));
Q
init  
qijun 已提交
124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144
  }

  void* scratchpad() const override {
    if (scratch_ == NULL) {
      scratch_ = allocate(Eigen::kCudaScratchSize + sizeof(unsigned int));
    }
    return scratch_;
  }

  unsigned int* semaphore() const override {
    if (semaphore_ == NULL) {
      char* scratch =
          static_cast<char*>(scratchpad()) + Eigen::kCudaScratchSize;
      semaphore_ = reinterpret_cast<unsigned int*>(scratch);
      PADDLE_ENFORCE(
          cudaMemsetAsync(semaphore_, 0, sizeof(unsigned int), *stream_));
    }
    return semaphore_;
  }

 private:
D
dzhwinter 已提交
145
  CUDAPlace place_;
Q
init  
qijun 已提交
146 147
  const cudaStream_t* stream_;         // not owned;
  const cudaDeviceProp* device_prop_;  // not owned;
Q
qijun 已提交
148
  mutable void* scratch_;
Q
init  
qijun 已提交
149
  mutable unsigned int* semaphore_;
150 151
  mutable std::unordered_map<void*, std::unique_ptr<memory::Allocation>>
      allocations_;
Q
init  
qijun 已提交
152 153
};

154 155 156
class CudnnHolder {
 public:
  CudnnHolder(const cudaStream_t* stream, const CUDAPlace& place)
157
      : workspace_(nullptr), stream_(stream), place_(place) {
158 159 160 161 162 163 164 165 166
    PADDLE_ENFORCE(dynload::cudnnCreate(&cudnn_handle_));
    PADDLE_ENFORCE(dynload::cudnnSetStream(cudnn_handle_, *stream_));
  }

  cudnnHandle_t cudnn_handle() const { return cudnn_handle_; }

  void RunFunc(const std::function<void(void*)>& cudnn_func,
               size_t required_workspace_len) {
    std::lock_guard<std::mutex> lock(mtx_);
167
    if (required_workspace_len > WorkspaceSize()) {
168 169
      ReallocateWorkspace(required_workspace_len);
    }
170
    cudnn_func(workspace_->ptr());
171 172
  }

173 174 175 176 177 178 179 180
  ~CudnnHolder() { PADDLE_ENFORCE(dynload::cudnnDestroy(cudnn_handle_)); }

 private:
  size_t WorkspaceSize() const {
    if (workspace_ == nullptr) {
      return 0;
    } else {
      return workspace_->size();
181 182 183 184
    }
  }

  void ReallocateWorkspace(size_t required_workspace_len) {
185
    if (required_workspace_len <= WorkspaceSize()) {
186 187 188 189 190
      return;
    }
    if (workspace_ != nullptr) {
      // Maybe someone is using the current workspace
      PADDLE_ENFORCE(cudaStreamSynchronize(*stream_));
191
      workspace_.reset();
192
    }
193 194
    workspace_ = paddle::memory::Alloc(place_, required_workspace_len,
                                       memory::Allocator::kFluxHuge);
195 196 197
  }

  cudnnHandle_t cudnn_handle_;
198
  std::unique_ptr<memory::Allocation> workspace_;
199 200 201 202 203 204 205 206 207

  const cudaStream_t* stream_;  // not owned;
  const CUDAPlace place_;

  std::mutex mtx_;
};

CUDADeviceContext::CUDADeviceContext(CUDAPlace place)
    : place_(place), cudnn_holder_(nullptr) {
208
  SetDeviceId(place_.device);
K
Kexin Zhao 已提交
209
  compute_capability = GetCUDAComputeCapability(place_.device);
210 211
  multi_process = GetCUDAMultiProcessors(place_.device);
  max_threads_per_mp = GetCUDAMaxThreadsPerMultiProcessor(place_.device);
Q
init  
qijun 已提交
212 213 214
  PADDLE_ENFORCE(cudaStreamCreate(&stream_));
  eigen_stream_.reset(new EigenCudaStreamDevice());
  eigen_stream_->Reinitialize(&stream_, place);
215
  eigen_device_.reset(new Eigen::GpuDevice(eigen_stream_.get()));
216 217
  PADDLE_ENFORCE(dynload::cublasCreate(&cublas_handle_));
  PADDLE_ENFORCE(dynload::cublasSetStream(cublas_handle_, stream_));
D
dzhwinter 已提交
218
  if (dynload::HasCUDNN()) {
219
    cudnn_holder_.reset(new CudnnHolder(&stream_, place));
D
dzhwinter 已提交
220
  }
S
sneaxiy 已提交
221 222

  callback_manager_.reset(new StreamCallbackManager(stream_));
223 224 225 226
}

CUDADeviceContext::~CUDADeviceContext() {
  SetDeviceId(place_.device);
L
liaogang 已提交
227
  Wait();
S
sneaxiy 已提交
228
  WaitStreamCallback();
229
  PADDLE_ENFORCE(dynload::cublasDestroy(cublas_handle_));
230 231
  eigen_stream_.reset();
  eigen_device_.reset();
Q
init  
qijun 已提交
232
  PADDLE_ENFORCE(cudaStreamDestroy(stream_));
233 234
}

L
liaogang 已提交
235
Place CUDADeviceContext::GetPlace() const { return place_; }
236

L
liaogang 已提交
237
void CUDADeviceContext::Wait() const {
Q
init  
qijun 已提交
238
  PADDLE_ENFORCE(cudaStreamSynchronize(stream_));
239 240 241
  PADDLE_ENFORCE(cudaGetLastError());
}

K
Kexin Zhao 已提交
242 243 244 245
int CUDADeviceContext::GetComputeCapability() const {
  return compute_capability;
}

246 247 248 249
int CUDADeviceContext::GetMaxPhysicalThreadCount() const {
  return multi_process * max_threads_per_mp;
}

250 251 252 253
Eigen::GpuDevice* CUDADeviceContext::eigen_device() const {
  return eigen_device_.get();
}

254
cublasHandle_t CUDADeviceContext::cublas_handle() const {
255 256 257
  return cublas_handle_;
}

258 259 260 261 262 263 264 265
cudnnHandle_t CUDADeviceContext::cudnn_handle() const {
  return cudnn_holder_->cudnn_handle();
}

void CUDADeviceContext::RunCudnnFuncWithWorkspace(
    const std::function<void(void*)>& cudnn_func, size_t workspace_len) const {
  cudnn_holder_->RunFunc(cudnn_func, workspace_len);
}
266

267
cudaStream_t CUDADeviceContext::stream() const { return stream_; }
Q
qijun 已提交
268

C
chengduoZH 已提交
269 270 271 272 273 274 275 276 277 278 279 280 281 282
CUDAPinnedDeviceContext::CUDAPinnedDeviceContext() {
  eigen_device_.reset(new Eigen::DefaultDevice());
}

CUDAPinnedDeviceContext::CUDAPinnedDeviceContext(CUDAPinnedPlace place)
    : place_(place) {
  eigen_device_.reset(new Eigen::DefaultDevice());
}

Eigen::DefaultDevice* CUDAPinnedDeviceContext::eigen_device() const {
  return eigen_device_.get();
}

Place CUDAPinnedDeviceContext::GetPlace() const { return place_; }
L
Luo Tao 已提交
283
#endif
Q
qijun 已提交
284

T
tensor-tang 已提交
285 286
#ifdef PADDLE_WITH_MKLDNN
MKLDNNDeviceContext::MKLDNNDeviceContext(CPUPlace place)
287 288
    : CPUDeviceContext(place), engine_(mkldnn::engine::cpu, 0), p_blobs_() {
  p_blobs_.reset(new std::unordered_map<std::string, std::shared_ptr<void>>());
T
tensor-tang 已提交
289 290
}

291 292 293 294
void MKLDNNDeviceContext::SetBlob(const std::string& name,
                                  std::shared_ptr<void> data) const {
  std::unordered_map<std::string, std::shared_ptr<void>>* p;
  p = p_blobs_.get();
T
tensor-tang 已提交
295

296
  auto it = p->find(name);
T
tensor-tang 已提交
297

298 299 300 301 302
  if (it == p->end()) {
    (*p)[name] = data;  // create new blob
  } else {
    it->second = data;  // set data to existing blob
  }
T
tensor-tang 已提交
303

304
  return;
T
tensor-tang 已提交
305 306
}

307 308 309 310
std::shared_ptr<void> MKLDNNDeviceContext::GetBlob(
    const std::string& name) const {
  std::unordered_map<std::string, std::shared_ptr<void>>* p;
  p = p_blobs_.get();
T
tensor-tang 已提交
311

312
  auto it = p->find(name);
T
tensor-tang 已提交
313

314 315
  if (it != p->end()) {
    return it->second;
T
tensor-tang 已提交
316
  }
317 318

  return nullptr;
T
tensor-tang 已提交
319 320 321 322
}

#endif

Q
qijun 已提交
323
}  // namespace platform
Q
qijun 已提交
324
}  // namespace paddle