hl_cuda_cudnn.cc 41.9 KB
Newer Older
1
/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.
Z
zhangjinchao01 已提交
2 3 4 5 6 7 8 9 10 11 12 13 14

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
15
#include "hl_cuda_cudnn.h"
Z
zhangjinchao01 已提交
16 17 18 19
#include <cudnn.h>
#include <mutex>
#include "hl_cuda_cudnn.ph"
#include "hl_dso_loader.h"
Y
Yu Yang 已提交
20
#include "hl_thread.ph"
21
#include "paddle/utils/CommandLineParser.h"
Y
Yu Yang 已提交
22
#include "paddle/utils/Logging.h"
23

24 25 26 27
DEFINE_int32(cudnn_conv_workspace_limit_in_mb,
             4096,
             "Specify cuDNN max workspace limit, in units MB, "
             "4096MB=4GB by default.");
Z
zhangjinchao01 已提交
28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43

namespace dynload {

std::once_flag cudnn_dso_flag;
void* cudnn_dso_handle = nullptr;

/**
 * The following macro definition can generate structs
 * (for each function) to dynamic load cudbnn routine
 * via operator overloading: operator ()
 *
 * note: default dynamic linked libs
 **/

#ifdef PADDLE_USE_DSO

44 45 46 47 48 49 50 51 52
#define DYNAMIC_LOAD_CUDNN_WRAP(__name)                                     \
  struct DynLoad__##__name {                                                \
    template <typename... Args>                                             \
    auto operator()(Args... args) -> decltype(__name(args...)) {            \
      using cudnn_func = decltype(__name(args...)) (*)(Args...);            \
      std::call_once(cudnn_dso_flag, GetCudnnDsoHandle, &cudnn_dso_handle); \
      void* p_##__name = dlsym(cudnn_dso_handle, #__name);                  \
      return reinterpret_cast<cudnn_func>(p_##__name)(args...);             \
    }                                                                       \
Z
zhangjinchao01 已提交
53 54 55 56
  } __name; /* struct DynLoad__##__name */

#else

57 58 59 60 61 62
#define DYNAMIC_LOAD_CUDNN_WRAP(__name)                          \
  struct DynLoad__##__name {                                     \
    template <typename... Args>                                  \
    auto operator()(Args... args) -> decltype(__name(args...)) { \
      return __name(args...);                                    \
    }                                                            \
Z
zhangjinchao01 已提交
63 64 65 66 67 68 69 70
  } __name; /* struct DynLoad__##__name */

#endif

/**
 * include all needed cudnn functions in HPPL
 * different cudnn version has different interfaces
 **/
71
// clang-format off
Z
zhangjinchao01 已提交
72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98
#define CUDNN_DNN_ROUTINE_EACH(__macro)                   \
  __macro(cudnnSetTensor4dDescriptor)                     \
  __macro(cudnnSetTensor4dDescriptorEx)                   \
  __macro(cudnnGetConvolutionNdForwardOutputDim)          \
  __macro(cudnnGetConvolutionForwardAlgorithm)            \
  __macro(cudnnCreateTensorDescriptor)                    \
  __macro(cudnnDestroyTensorDescriptor)                   \
  __macro(cudnnCreateFilterDescriptor)                    \
  __macro(cudnnSetFilter4dDescriptor)                     \
  __macro(cudnnSetPooling2dDescriptor)                    \
  __macro(cudnnDestroyFilterDescriptor)                   \
  __macro(cudnnCreateConvolutionDescriptor)               \
  __macro(cudnnCreatePoolingDescriptor)                   \
  __macro(cudnnDestroyPoolingDescriptor)                  \
  __macro(cudnnSetConvolution2dDescriptor)                \
  __macro(cudnnDestroyConvolutionDescriptor)              \
  __macro(cudnnCreate)                                    \
  __macro(cudnnDestroy)                                   \
  __macro(cudnnSetStream)                                 \
  __macro(cudnnActivationForward)                         \
  __macro(cudnnConvolutionForward)                        \
  __macro(cudnnConvolutionBackwardBias)                   \
  __macro(cudnnGetConvolutionForwardWorkspaceSize)        \
  __macro(cudnnTransformTensor)                           \
  __macro(cudnnPoolingForward)                            \
  __macro(cudnnPoolingBackward)                           \
  __macro(cudnnSoftmaxBackward)                           \
99 100 101
  __macro(cudnnSoftmaxForward)                            \
  __macro(cudnnGetVersion)                                \
  __macro(cudnnGetErrorString)
Z
zhangjinchao01 已提交
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122
CUDNN_DNN_ROUTINE_EACH(DYNAMIC_LOAD_CUDNN_WRAP)

#define CUDNN_DNN_ROUTINE_EACH_R2(__macro)                \
  __macro(cudnnAddTensor)                                 \
  __macro(cudnnConvolutionBackwardData)                   \
  __macro(cudnnConvolutionBackwardFilter)
CUDNN_DNN_ROUTINE_EACH_R2(DYNAMIC_LOAD_CUDNN_WRAP)

// APIs available after R3:
#if CUDNN_VERSION >= 3000
#define CUDNN_DNN_ROUTINE_EACH_AFTER_R3(__macro)              \
  __macro(cudnnGetConvolutionBackwardFilterWorkspaceSize)     \
  __macro(cudnnGetConvolutionBackwardDataAlgorithm)           \
  __macro(cudnnGetConvolutionBackwardFilterAlgorithm)         \
  __macro(cudnnGetConvolutionBackwardDataWorkspaceSize)
CUDNN_DNN_ROUTINE_EACH_AFTER_R3(DYNAMIC_LOAD_CUDNN_WRAP)
#undef CUDNN_DNN_ROUTINE_EACH_AFTER_R3
#endif


// APIs available after R4:
123
#if CUDNN_VERSION >= 4007
Z
zhangjinchao01 已提交
124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143
#define CUDNN_DNN_ROUTINE_EACH_AFTER_R4(__macro)             \
  __macro(cudnnBatchNormalizationForwardTraining)            \
  __macro(cudnnBatchNormalizationForwardInference)           \
  __macro(cudnnBatchNormalizationBackward)
CUDNN_DNN_ROUTINE_EACH_AFTER_R4(DYNAMIC_LOAD_CUDNN_WRAP)
#undef CUDNN_DNN_ROUTINE_EACH_AFTER_R4
#endif

// APIs in R5
#if CUDNN_VERSION >= 5000
#define CUDNN_DNN_ROUTINE_EACH_R5(__macro)                    \
  __macro(cudnnCreateActivationDescriptor)                    \
  __macro(cudnnSetActivationDescriptor)                       \
  __macro(cudnnGetActivationDescriptor)                       \
  __macro(cudnnDestroyActivationDescriptor)
CUDNN_DNN_ROUTINE_EACH_R5(DYNAMIC_LOAD_CUDNN_WRAP)
#undef CUDNN_DNN_ROUTINE_EACH_R5
#endif

#undef CUDNN_DNN_ROUTINE_EACH
144
// clang-format on
Z
zhangjinchao01 已提交
145 146 147
} /* namespace dynload */

/**
148
 * Check build-in cudnn function using glog and it **does not**
Z
zhangjinchao01 已提交
149 150
 * support << operator for more details error info.
 */
151 152 153 154 155
#define CHECK_CUDNN(cudnnFunc)                                         \
  do {                                                                 \
    cudnnStatus_t cudnnStat = cudnnFunc;                               \
    CHECK_EQ(CUDNN_STATUS_SUCCESS, cudnnStat)                          \
        << "Cudnn Error: " << dynload::cudnnGetErrorString(cudnnStat); \
156
  } while (0)
Z
zhangjinchao01 已提交
157 158 159 160

bool g_is_libcudnn_init = false;
int g_cudnn_lib_version = 0;

161 162
void hl_cudnn_desc_init(cudnnTensorDescriptor_t* cudnn_desc) {
  CHECK_CUDNN(dynload::cudnnCreateTensorDescriptor(cudnn_desc));
Z
zhangjinchao01 已提交
163 164
}

165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187
void hl_cudnn_init(cudnnHandle_t* cudnn_handle, cudaStream_t stream) {
  size_t cudnn_dso_ver = dynload::cudnnGetVersion();
  size_t cudnn_dso_major = cudnn_dso_ver / 1000;
  size_t cudnn_cuh_major = CUDNN_VERSION / 1000;

  // Compare cudnn header version with that of cudnn.so.
  CHECK((cudnn_cuh_major < 4 && cudnn_dso_major < 4) ||
        (cudnn_cuh_major == cudnn_dso_major))
      << "[cudnn init] libcudnn v" << cudnn_dso_major << " with header v"
      << cudnn_cuh_major << " unmatched!\n"
      << "PaddlePaddle Requirement: "
      << "(header v[2-3] with libcudnn v[2-3]) Or "
      << "(header v4 with libcudnn v4) Or "
      << "(header v5 with libcudnn v5).";

  CHECK(!(CUDNN_VERSION >= 5000 && CUDA_VERSION < 7050))
      << "cudnn v5 requires cuda version >= 7.5";

  CHECK_CUDNN(dynload::cudnnCreate(cudnn_handle));
  CHECK_CUDNN(dynload::cudnnSetStream(*cudnn_handle, stream));

  g_is_libcudnn_init = true;
  g_cudnn_lib_version = cudnn_dso_ver;
Z
zhangjinchao01 已提交
188 189
}

190
int hl_get_cudnn_lib_version() { return g_cudnn_lib_version; }
Z
zhangjinchao01 已提交
191 192 193 194 195 196 197 198 199 200 201 202 203

void hl_conv_workspace(hl_tensor_descriptor input,
                       hl_tensor_descriptor output,
                       hl_filter_descriptor filter,
                       hl_convolution_descriptor conv,
                       int* convFwdAlgo,
                       size_t* fwdLimitBytes,
                       int* convBwdDataAlgo,
                       size_t* bwdDataLimitBytes,
                       int* convBwdFilterAlgo,
                       size_t* bwdFilterLimitBytes) {
#if CUDNN_VERSION >= 4000

204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288
  CHECK_NOTNULL(input);
  CHECK_NOTNULL(output);
  CHECK_NOTNULL(filter);
  CHECK_NOTNULL(conv);

  // Specify workspace limit directly
  size_t memoryLimitBytes =
      (1LL << 20) * FLAGS_cudnn_conv_workspace_limit_in_mb;

  // cudnn convolution forward configuration
  cudnnTensorDescriptor_t fwd_src_desc = GET_TENSOR_DESCRIPTOR(input);
  cudnnTensorDescriptor_t fwd_dest_desc = GET_TENSOR_DESCRIPTOR(output);
  cudnnFilterDescriptor_t fwd_filter_desc = GET_FILTER_DESCRIPTOR(filter);
  cudnnConvolutionDescriptor_t fwd_conv_desc = GET_CONVOLUTION_DESCRIPTOR(conv);

  CHECK_CUDNN(dynload::cudnnGetConvolutionForwardAlgorithm(
      t_resource.cudnn_handle,
      fwd_src_desc,
      fwd_filter_desc,
      fwd_conv_desc,
      fwd_dest_desc,
      CUDNN_CONVOLUTION_FWD_SPECIFY_WORKSPACE_LIMIT,
      memoryLimitBytes,
      reinterpret_cast<cudnnConvolutionFwdAlgo_t*>(convFwdAlgo)));

  CHECK_CUDNN(dynload::cudnnGetConvolutionForwardWorkspaceSize(
      t_resource.cudnn_handle,
      fwd_src_desc,
      fwd_filter_desc,
      fwd_conv_desc,
      fwd_dest_desc,
      static_cast<cudnnConvolutionFwdAlgo_t>(*convFwdAlgo),
      fwdLimitBytes));

  // cudnn convolution backward data configuration
  cudnnFilterDescriptor_t bwd_data_filter_desc = GET_FILTER_DESCRIPTOR(filter);
  cudnnTensorDescriptor_t bwd_data_diff_desc = GET_TENSOR_DESCRIPTOR(output);
  cudnnTensorDescriptor_t bwd_data_grad_desc = GET_TENSOR_DESCRIPTOR(input);
  cudnnConvolutionDescriptor_t bwd_data_conv_desc =
      GET_CONVOLUTION_DESCRIPTOR(conv);

  CHECK_CUDNN(dynload::cudnnGetConvolutionBackwardDataAlgorithm(
      t_resource.cudnn_handle,
      bwd_data_filter_desc,
      bwd_data_diff_desc,
      bwd_data_conv_desc,
      bwd_data_grad_desc,
      CUDNN_CONVOLUTION_BWD_DATA_SPECIFY_WORKSPACE_LIMIT,
      memoryLimitBytes,
      reinterpret_cast<cudnnConvolutionBwdDataAlgo_t*>(convBwdDataAlgo)));

  CHECK_CUDNN(dynload::cudnnGetConvolutionBackwardDataWorkspaceSize(
      t_resource.cudnn_handle,
      bwd_data_filter_desc,
      bwd_data_diff_desc,
      bwd_data_conv_desc,
      bwd_data_grad_desc,
      static_cast<cudnnConvolutionBwdDataAlgo_t>(*convBwdDataAlgo),
      bwdDataLimitBytes));

  // cudnn convolution backward filter configuration
  cudnnTensorDescriptor_t bwd_filter_src_desc = GET_TENSOR_DESCRIPTOR(input);
  cudnnTensorDescriptor_t bwd_filter_diff_desc = GET_TENSOR_DESCRIPTOR(output);
  cudnnConvolutionDescriptor_t bwd_filter_conv_desc =
      GET_CONVOLUTION_DESCRIPTOR(conv);
  cudnnFilterDescriptor_t bwd_filter_grad_desc = GET_FILTER_DESCRIPTOR(filter);

  CHECK_CUDNN(dynload::cudnnGetConvolutionBackwardFilterAlgorithm(
      t_resource.cudnn_handle,
      bwd_filter_src_desc,
      bwd_filter_diff_desc,
      bwd_filter_conv_desc,
      bwd_filter_grad_desc,
      CUDNN_CONVOLUTION_BWD_FILTER_SPECIFY_WORKSPACE_LIMIT,
      memoryLimitBytes,
      reinterpret_cast<cudnnConvolutionBwdFilterAlgo_t*>(convBwdFilterAlgo)));

  CHECK_CUDNN(dynload::cudnnGetConvolutionBackwardFilterWorkspaceSize(
      t_resource.cudnn_handle,
      bwd_filter_src_desc,
      bwd_filter_diff_desc,
      bwd_filter_conv_desc,
      bwd_filter_grad_desc,
      static_cast<cudnnConvolutionBwdFilterAlgo_t>(*convBwdFilterAlgo),
      bwdFilterLimitBytes));
Z
zhangjinchao01 已提交
289 290 291 292 293 294 295 296

#endif
}

void hl_create_tensor_descriptor(hl_tensor_descriptor* image_desc,
                                 int batch_size,
                                 int feature_maps,
                                 int height,
297
                                 int width) {
298
  CHECK_NOTNULL(image_desc);
Z
zhangjinchao01 已提交
299

300 301 302
  cudnn_tensor_descriptor hl_desc =
      (cudnn_tensor_descriptor)malloc(sizeof(_cudnn_tensor_descriptor));
  CHECK_NOTNULL(hl_desc);
Z
zhangjinchao01 已提交
303

304
#ifndef PADDLE_TYPE_DOUBLE
305
  cudnnDataType_t data_type = CUDNN_DATA_FLOAT;
Z
zhangjinchao01 已提交
306
#else
307
  cudnnDataType_t data_type = CUDNN_DATA_DOUBLE;
Z
zhangjinchao01 已提交
308
#endif
309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326
  CHECK_CUDNN(dynload::cudnnCreateTensorDescriptor(&hl_desc->desc));

  CHECK_CUDNN(dynload::cudnnSetTensor4dDescriptor(hl_desc->desc,
                                                  CUDNN_TENSOR_NCHW,
                                                  data_type,
                                                  batch_size,
                                                  feature_maps,
                                                  height,
                                                  width));

  hl_desc->format = CUDNN_TENSOR_NCHW;
  hl_desc->data_type = data_type;
  hl_desc->batch_size = batch_size;
  hl_desc->feature_maps = feature_maps;
  hl_desc->height = height;
  hl_desc->width = width;

  *image_desc = (hl_tensor_descriptor)hl_desc;
Z
zhangjinchao01 已提交
327 328 329
}

void hl_create_tensor_descriptor(hl_tensor_descriptor* image_desc) {
330
  CHECK_NOTNULL(image_desc);
Z
zhangjinchao01 已提交
331

332 333 334
  cudnn_tensor_descriptor hl_desc =
      (cudnn_tensor_descriptor)malloc(sizeof(_cudnn_tensor_descriptor));
  CHECK_NOTNULL(hl_desc);
Z
zhangjinchao01 已提交
335

336
#ifndef PADDLE_TYPE_DOUBLE
337
  cudnnDataType_t data_type = CUDNN_DATA_FLOAT;
Z
zhangjinchao01 已提交
338
#else
339
  cudnnDataType_t data_type = CUDNN_DATA_DOUBLE;
Z
zhangjinchao01 已提交
340
#endif
341
  CHECK_CUDNN(dynload::cudnnCreateTensorDescriptor(&hl_desc->desc));
Z
zhangjinchao01 已提交
342

343
  hl_desc->data_type = data_type;
Z
zhangjinchao01 已提交
344

345
  *image_desc = (hl_tensor_descriptor)hl_desc;
Z
zhangjinchao01 已提交
346 347 348 349 350 351
}

void hl_tensor_reshape(hl_tensor_descriptor image_desc,
                       int batch_size,
                       int feature_maps,
                       int height,
352
                       int width) {
353 354 355 356 357 358 359 360 361 362 363 364 365
  const int stride_w = 1;
  const int stride_h = width * stride_w;
  const int stride_c = height * stride_h;
  const int stride_n = feature_maps * stride_c;
  return hl_tensor_reshape(image_desc,
                           batch_size,
                           feature_maps,
                           height,
                           width,
                           stride_n,
                           stride_c,
                           stride_h,
                           stride_w);
Z
zhangjinchao01 已提交
366 367 368 369 370 371 372 373 374 375
}

void hl_tensor_reshape(hl_tensor_descriptor image_desc,
                       int batch_size,
                       int feature_maps,
                       int height,
                       int width,
                       int nStride,
                       int cStride,
                       int hStride,
376
                       int wStride) {
377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396
  CHECK_NOTNULL(image_desc);

  cudnn_tensor_descriptor hl_desc = (cudnn_tensor_descriptor)image_desc;
  CHECK_NOTNULL(hl_desc->desc);

  CHECK_CUDNN(dynload::cudnnSetTensor4dDescriptorEx(hl_desc->desc,
                                                    hl_desc->data_type,
                                                    batch_size,
                                                    feature_maps,
                                                    height,
                                                    width,
                                                    nStride,
                                                    cStride,
                                                    hStride,
                                                    wStride));

  hl_desc->batch_size = batch_size;
  hl_desc->feature_maps = feature_maps;
  hl_desc->height = height;
  hl_desc->width = width;
Z
zhangjinchao01 已提交
397 398
}

399
void hl_destroy_tensor_descriptor(hl_tensor_descriptor image_desc) {
400
  CHECK_NOTNULL(image_desc);
Z
zhangjinchao01 已提交
401

402 403
  cudnn_tensor_descriptor hl_desc = (cudnn_tensor_descriptor)image_desc;
  CHECK_NOTNULL(hl_desc->desc);
Z
zhangjinchao01 已提交
404

405
  CHECK_CUDNN(dynload::cudnnDestroyTensorDescriptor(hl_desc->desc));
Z
zhangjinchao01 已提交
406

407
  hl_desc->desc = NULL;
Z
zhangjinchao01 已提交
408

409
  free(image_desc);
Z
zhangjinchao01 已提交
410 411 412 413 414 415 416 417 418
}

void hl_create_pooling_descriptor(hl_pooling_descriptor* pooling_desc,
                                  hl_pooling_mode_t mode,
                                  int height,
                                  int width,
                                  int height_padding,
                                  int width_padding,
                                  int stride_height,
419
                                  int stride_width) {
420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444
  cudnnPoolingMode_t cudnn_mode;
  switch (mode) {
    case HL_POOLING_MAX:
      cudnn_mode = CUDNN_POOLING_MAX;
      break;
    case HL_POOLING_AVERAGE:
      cudnn_mode = CUDNN_POOLING_AVERAGE_COUNT_INCLUDE_PADDING;
      break;
    case HL_POOLING_AVERAGE_EXCLUDE_PADDING:
      cudnn_mode = CUDNN_POOLING_AVERAGE_COUNT_EXCLUDE_PADDING;
      break;
    default:
      LOG(FATAL) << "parameter mode error";
  }

  CHECK_NOTNULL(pooling_desc);

  cudnn_pooling_descriptor hl_pooling_desc =
      (cudnn_pooling_descriptor)malloc(sizeof(_cudnn_pooling_descriptor));
  CHECK_NOTNULL(hl_pooling_desc);

  CHECK_CUDNN(dynload::cudnnCreatePoolingDescriptor(&hl_pooling_desc->desc));

  CHECK_CUDNN(dynload::cudnnSetPooling2dDescriptor(hl_pooling_desc->desc,
                                                   cudnn_mode,
Z
zhangjinchao01 已提交
445
#if CUDNN_VERSION >= 5000
446
                                                   CUDNN_PROPAGATE_NAN,
Z
zhangjinchao01 已提交
447
#endif
448 449 450 451 452 453 454 455 456 457 458 459 460 461
                                                   height,
                                                   width,
                                                   height_padding,
                                                   width_padding,
                                                   stride_height,
                                                   stride_width));

  hl_pooling_desc->mode = cudnn_mode;
  hl_pooling_desc->window_height = height;
  hl_pooling_desc->window_width = width;
  hl_pooling_desc->stride_height = stride_height;
  hl_pooling_desc->stride_width = stride_width;

  *pooling_desc = (hl_pooling_descriptor)hl_pooling_desc;
Z
zhangjinchao01 已提交
462 463
}

464
void hl_destroy_pooling_descriptor(hl_pooling_descriptor pooling_desc) {
465
  CHECK_NOTNULL(pooling_desc);
Z
zhangjinchao01 已提交
466

467
  cudnn_pooling_descriptor hl_pooling = (cudnn_pooling_descriptor)pooling_desc;
Z
zhangjinchao01 已提交
468

469 470
  CHECK_NOTNULL(hl_pooling->desc);
  CHECK_CUDNN(dynload::cudnnDestroyPoolingDescriptor(hl_pooling->desc));
Z
zhangjinchao01 已提交
471

472
  hl_pooling->desc = NULL;
Z
zhangjinchao01 已提交
473

474
  free(pooling_desc);
Z
zhangjinchao01 已提交
475 476 477 478 479 480
}

void hl_pooling_forward(hl_tensor_descriptor input,
                        real* input_image,
                        hl_tensor_descriptor output,
                        real* output_image,
481
                        hl_pooling_descriptor pooling) {
482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505
  cudnnPoolingDescriptor_t pooling_desc;
  cudnnTensorDescriptor_t input_desc;
  cudnnTensorDescriptor_t output_desc;

  CHECK_NOTNULL(input);
  CHECK_NOTNULL(output);
  CHECK_NOTNULL(pooling);
  CHECK_NOTNULL(input_image);
  CHECK_NOTNULL(output_image);

  real alpha = 1.0f;
  real beta = 1.0f;
  input_desc = ((cudnn_tensor_descriptor)input)->desc;
  output_desc = ((cudnn_tensor_descriptor)output)->desc;
  pooling_desc = ((cudnn_pooling_descriptor)pooling)->desc;
  CHECK_CUDNN(dynload::cudnnPoolingForward(t_resource.cudnn_handle,
                                           pooling_desc,
                                           &alpha,
                                           input_desc,
                                           input_image,
                                           &beta,
                                           output_desc,
                                           output_image));
  CHECK_SYNC("hl_pooling_forward failed");
Z
zhangjinchao01 已提交
506 507 508 509 510 511 512 513
}

void hl_pooling_backward(hl_tensor_descriptor input,
                         real* input_image,
                         real* input_image_grad,
                         hl_tensor_descriptor output,
                         real* output_image,
                         real* output_image_grad,
514
                         hl_pooling_descriptor pooling) {
515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543
  cudnnPoolingDescriptor_t pooling_desc;
  cudnnTensorDescriptor_t input_desc;
  cudnnTensorDescriptor_t output_desc;

  CHECK_NOTNULL(input);
  CHECK_NOTNULL(output);
  CHECK_NOTNULL(pooling);
  CHECK_NOTNULL(input_image);
  CHECK_NOTNULL(input_image_grad);
  CHECK_NOTNULL(output_image);
  CHECK_NOTNULL(output_image_grad);

  real alpha = 1.0f;
  real beta = 1.0f;
  input_desc = ((cudnn_tensor_descriptor)input)->desc;
  output_desc = ((cudnn_tensor_descriptor)output)->desc;
  pooling_desc = ((cudnn_pooling_descriptor)pooling)->desc;
  CHECK_CUDNN(dynload::cudnnPoolingBackward(t_resource.cudnn_handle,
                                            pooling_desc,
                                            &alpha,
                                            output_desc,
                                            output_image,
                                            output_desc,
                                            output_image_grad,
                                            input_desc,
                                            input_image,
                                            &beta,
                                            input_desc,
                                            input_image_grad));
Z
zhangjinchao01 已提交
544 545 546 547 548 549 550
  CHECK_SYNC("hl_pooling_backward failed");
}

void hl_create_filter_descriptor(hl_filter_descriptor* filter,
                                 int input_feature_maps,
                                 int output_feature_maps,
                                 int height,
551
                                 int width) {
552
  CHECK_NOTNULL(filter);
Z
zhangjinchao01 已提交
553

554 555 556
  cudnn_filter_descriptor hl_filter =
      (cudnn_filter_descriptor)malloc(sizeof(_cudnn_filter_descriptor));
  CHECK_NOTNULL(hl_filter);
Z
zhangjinchao01 已提交
557

558
  CHECK_CUDNN(dynload::cudnnCreateFilterDescriptor(&hl_filter->desc));
Z
zhangjinchao01 已提交
559

560
#ifndef PADDLE_TYPE_DOUBLE
561
  cudnnDataType_t data_type = CUDNN_DATA_FLOAT;
Z
zhangjinchao01 已提交
562
#else
563
  cudnnDataType_t data_type = CUDNN_DATA_DOUBLE;
Z
zhangjinchao01 已提交
564
#endif
565 566
  CHECK_CUDNN(dynload::cudnnSetFilter4dDescriptor(hl_filter->desc,
                                                  data_type,
Z
zhangjinchao01 已提交
567
#if CUDNN_VERSION >= 5000
568
                                                  CUDNN_TENSOR_NCHW,
Z
zhangjinchao01 已提交
569
#endif
570 571 572 573 574 575 576 577 578 579 580 581
                                                  output_feature_maps,
                                                  input_feature_maps,
                                                  height,
                                                  width));

  hl_filter->data_type = data_type;
  hl_filter->output_feature_maps = output_feature_maps;
  hl_filter->input_feature_maps = input_feature_maps;
  hl_filter->filter_height = height;
  hl_filter->filter_width = width;

  *filter = (hl_filter_descriptor)hl_filter;
Z
zhangjinchao01 已提交
582 583
}

584
void hl_destroy_filter_descriptor(hl_filter_descriptor filter) {
585
  CHECK_NOTNULL(filter);
Z
zhangjinchao01 已提交
586

587 588
  cudnn_filter_descriptor hl_filter = (cudnn_filter_descriptor)filter;
  CHECK_NOTNULL(hl_filter->desc);
Z
zhangjinchao01 已提交
589

590
  CHECK_CUDNN(dynload::cudnnDestroyFilterDescriptor(hl_filter->desc));
Z
zhangjinchao01 已提交
591

592
  hl_filter->desc = NULL;
Z
zhangjinchao01 已提交
593

594
  free(filter);
Z
zhangjinchao01 已提交
595 596 597 598 599 600 601 602
}

void hl_create_convolution_descriptor(hl_convolution_descriptor* conv,
                                      hl_tensor_descriptor image,
                                      hl_filter_descriptor filter,
                                      int padding_height,
                                      int padding_width,
                                      int stride_height,
603
                                      int stride_width) {
604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632
  CHECK_NOTNULL(conv);

  cudnn_convolution_descriptor hl_conv = (cudnn_convolution_descriptor)malloc(
      sizeof(_cudnn_convolution_descriptor));

  CHECK_NOTNULL(hl_conv);
  CHECK_CUDNN(dynload::cudnnCreateConvolutionDescriptor(&hl_conv->desc));

  cudnnConvolutionMode_t mode = CUDNN_CROSS_CORRELATION;
  CHECK_CUDNN(dynload::cudnnSetConvolution2dDescriptor(hl_conv->desc,
                                                       padding_height,
                                                       padding_width,
                                                       stride_height,
                                                       stride_width,
                                                       1,
                                                       1,
                                                       mode));

  hl_conv->input_image = image;
  hl_conv->filter = filter;
  hl_conv->padding_height = padding_height;
  hl_conv->padding_width = padding_width;
  hl_conv->stride_height = stride_height;
  hl_conv->stride_width = stride_width;
  hl_conv->upscalex = 1;
  hl_conv->upscaley = 1;
  hl_conv->mode = mode;

  *conv = (hl_convolution_descriptor)hl_conv;
Z
zhangjinchao01 已提交
633 634 635 636 637 638 639 640
}

void hl_reset_convolution_descriptor(hl_convolution_descriptor conv,
                                     hl_tensor_descriptor image,
                                     hl_filter_descriptor filter,
                                     int padding_height,
                                     int padding_width,
                                     int stride_height,
641
                                     int stride_width) {
642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666
  CHECK_NOTNULL(conv);
  CHECK_NOTNULL(image);
  CHECK_NOTNULL(filter);

  cudnnConvolutionDescriptor_t conv_desc = GET_CONVOLUTION_DESCRIPTOR(conv);
  cudnnConvolutionMode_t mode = CUDNN_CROSS_CORRELATION;
  CHECK_CUDNN(dynload::cudnnSetConvolution2dDescriptor(conv_desc,
                                                       padding_height,
                                                       padding_width,
                                                       stride_height,
                                                       stride_width,
                                                       1,
                                                       1,
                                                       mode));

  cudnn_convolution_descriptor hl_conv = (cudnn_convolution_descriptor)conv;
  hl_conv->input_image = image;
  hl_conv->filter = filter;
  hl_conv->padding_height = padding_height;
  hl_conv->padding_width = padding_width;
  hl_conv->stride_height = stride_height;
  hl_conv->stride_width = stride_width;
  hl_conv->upscalex = 1;
  hl_conv->upscaley = 1;
  hl_conv->mode = mode;
Z
zhangjinchao01 已提交
667 668
}

669
void hl_destroy_convolution_descriptor(hl_convolution_descriptor conv) {
670
  CHECK_NOTNULL(conv);
Z
zhangjinchao01 已提交
671

672 673
  cudnn_convolution_descriptor hl_conv = (cudnn_convolution_descriptor)conv;
  CHECK_NOTNULL(hl_conv->desc);
Z
zhangjinchao01 已提交
674

675 676
  CHECK_CUDNN(dynload::cudnnDestroyConvolutionDescriptor(hl_conv->desc));
  hl_conv->desc = NULL;
Z
zhangjinchao01 已提交
677

678
  free(conv);
Z
zhangjinchao01 已提交
679 680 681 682 683 684 685 686 687 688 689 690
}

void hl_convolution_forward(hl_tensor_descriptor input,
                            real* input_data,
                            hl_tensor_descriptor output,
                            real* output_data,
                            hl_filter_descriptor filter,
                            real* filter_data,
                            hl_convolution_descriptor conv,
                            void* gpuWorkSpace,
                            size_t sizeInBytes,
                            int convFwdAlgo) {
691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717
  CHECK_NOTNULL(input);
  CHECK_NOTNULL(output);
  CHECK_NOTNULL(filter);
  CHECK_NOTNULL(conv);
  CHECK_NOTNULL(input_data);
  CHECK_NOTNULL(output_data);
  CHECK_NOTNULL(filter_data);
  cudnnTensorDescriptor_t src_desc = GET_TENSOR_DESCRIPTOR(input);
  cudnnTensorDescriptor_t dest_desc = GET_TENSOR_DESCRIPTOR(output);
  cudnnFilterDescriptor_t filter_desc = GET_FILTER_DESCRIPTOR(filter);
  cudnnConvolutionDescriptor_t conv_desc = GET_CONVOLUTION_DESCRIPTOR(conv);
  real alpha = 1.0f;
  real beta = 1.0f;
  CHECK_CUDNN(dynload::cudnnConvolutionForward(
      t_resource.cudnn_handle,
      &alpha,
      src_desc,
      input_data,
      filter_desc,
      filter_data,
      conv_desc,
      static_cast<cudnnConvolutionFwdAlgo_t>(convFwdAlgo),
      gpuWorkSpace,
      sizeInBytes,
      &beta,
      dest_desc,
      output_data));
Z
zhangjinchao01 已提交
718 719 720 721 722 723
  CHECK_SYNC("hl_convolution_forward failed");
}

void hl_convolution_forward_add_bias(hl_tensor_descriptor bias,
                                     real* bias_data,
                                     hl_tensor_descriptor output,
724
                                     real* output_data) {
725 726 727 728 729 730 731 732 733 734 735
  CHECK_NOTNULL(bias);
  CHECK_NOTNULL(output);
  CHECK_NOTNULL(bias_data);
  CHECK_NOTNULL(output_data);

  cudnnTensorDescriptor_t output_desc = GET_TENSOR_DESCRIPTOR(output);
  cudnnTensorDescriptor_t bias_desc = GET_TENSOR_DESCRIPTOR(bias);
  real alpha = 1.0f;
  real beta = 1.0f;

  CHECK_CUDNN(dynload::cudnnAddTensor(t_resource.cudnn_handle,
Z
zhangjinchao01 已提交
736
#if CUDNN_VERSION < 4000
737
                                      CUDNN_ADD_SAME_C,
Z
zhangjinchao01 已提交
738
#endif
739 740 741 742 743 744
                                      &alpha,
                                      bias_desc,
                                      bias_data,
                                      &beta,
                                      output_desc,
                                      output_data));
Z
zhangjinchao01 已提交
745 746 747 748 749 750
  CHECK_SYNC("hl_convolution_forward_add_bias failed");
}

void hl_convolution_backward_bias(hl_tensor_descriptor bias,
                                  real* bias_grad_data,
                                  hl_tensor_descriptor output,
751
                                  real* output_grad_data) {
752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767
  CHECK_NOTNULL(bias);
  CHECK_NOTNULL(output);
  CHECK_NOTNULL(bias_grad_data);
  CHECK_NOTNULL(output_grad_data);

  real alpha = 1.0f;
  real beta = 1.0f;
  cudnnTensorDescriptor_t diff_desc = GET_TENSOR_DESCRIPTOR(output);
  cudnnTensorDescriptor_t bias_desc = GET_TENSOR_DESCRIPTOR(bias);
  CHECK_CUDNN(dynload::cudnnConvolutionBackwardBias(t_resource.cudnn_handle,
                                                    &alpha,
                                                    diff_desc,
                                                    output_grad_data,
                                                    &beta,
                                                    bias_desc,
                                                    bias_grad_data));
Z
zhangjinchao01 已提交
768 769 770 771 772 773 774 775 776 777 778 779 780
  CHECK_SYNC("hl_convolution_backward_bias failed");
}

void hl_convolution_backward_filter(hl_tensor_descriptor input,
                                    real* input_data,
                                    hl_tensor_descriptor output,
                                    real* output_grad_data,
                                    hl_filter_descriptor filter,
                                    real* filter_grad_data,
                                    hl_convolution_descriptor conv,
                                    void* gpuWorkSpace,
                                    size_t sizeInBytes,
                                    int convBwdFilterAlgo) {
781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803
  CHECK_NOTNULL(input);
  CHECK_NOTNULL(output);
  CHECK_NOTNULL(filter);
  CHECK_NOTNULL(conv);
  CHECK_NOTNULL(input_data);
  CHECK_NOTNULL(output_grad_data);
  CHECK_NOTNULL(filter_grad_data);

  real alpha = 1.0f;
  real beta = 1.0f;
  cudnnTensorDescriptor_t src_desc = GET_TENSOR_DESCRIPTOR(input);
  cudnnTensorDescriptor_t diff_desc = GET_TENSOR_DESCRIPTOR(output);
  cudnnConvolutionDescriptor_t conv_desc = GET_CONVOLUTION_DESCRIPTOR(conv);
  cudnnFilterDescriptor_t grad_desc = GET_FILTER_DESCRIPTOR(filter);

  CHECK_CUDNN(dynload::cudnnConvolutionBackwardFilter(
      t_resource.cudnn_handle,
      &alpha,
      src_desc,
      input_data,
      diff_desc,
      output_grad_data,
      conv_desc,
Z
zhangjinchao01 已提交
804
#if CUDNN_VERSION >= 4000
805 806 807
      static_cast<cudnnConvolutionBwdFilterAlgo_t>(convBwdFilterAlgo),
      gpuWorkSpace,
      sizeInBytes,
Z
zhangjinchao01 已提交
808
#endif
809 810 811
      &beta,
      grad_desc,
      filter_grad_data));
Z
zhangjinchao01 已提交
812 813 814 815 816 817 818 819 820 821 822 823 824
  CHECK_SYNC("hl_convolution_backward_filter failed");
}

void hl_convolution_backward_data(hl_tensor_descriptor input,
                                  real* input_data_grad,
                                  hl_tensor_descriptor output,
                                  real* output_grad_data,
                                  hl_filter_descriptor filter,
                                  real* filter_data,
                                  hl_convolution_descriptor conv,
                                  void* gpuWorkSpace,
                                  size_t sizeInBytes,
                                  int convBwdDataAlgo) {
825 826 827 828 829 830 831 832 833 834 835 836 837 838 839
  real alpha = 1.0f;
  real beta = 1.0f;
  cudnnFilterDescriptor_t filter_desc = GET_FILTER_DESCRIPTOR(filter);
  cudnnTensorDescriptor_t diff_desc = GET_TENSOR_DESCRIPTOR(output);
  cudnnTensorDescriptor_t grad_desc = GET_TENSOR_DESCRIPTOR(input);
  cudnnConvolutionDescriptor_t conv_desc = GET_CONVOLUTION_DESCRIPTOR(conv);

  CHECK_CUDNN(dynload::cudnnConvolutionBackwardData(
      t_resource.cudnn_handle,
      &alpha,
      filter_desc,
      filter_data,
      diff_desc,
      output_grad_data,
      conv_desc,
Z
zhangjinchao01 已提交
840
#if CUDNN_VERSION >= 4000
841 842 843
      static_cast<cudnnConvolutionBwdDataAlgo_t>(convBwdDataAlgo),
      gpuWorkSpace,
      sizeInBytes,
Z
zhangjinchao01 已提交
844
#endif
845 846 847
      &beta,
      grad_desc,
      input_data_grad));
Z
zhangjinchao01 已提交
848 849 850
  CHECK_SYNC("hl_convolution_backward_data failed");
}

851
void hl_softmax_forward(real* input, real* output, int height, int width) {
852
#ifndef PADDLE_TYPE_DOUBLE
853
  cudnnDataType_t data_type = CUDNN_DATA_FLOAT;
Z
zhangjinchao01 已提交
854
#else
855
  cudnnDataType_t data_type = CUDNN_DATA_DOUBLE;
Z
zhangjinchao01 已提交
856
#endif
857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875
  CHECK_CUDNN(dynload::cudnnSetTensor4dDescriptor(t_resource.cudnn_desc,
                                                  CUDNN_TENSOR_NCHW,
                                                  data_type,
                                                  height,
                                                  width,
                                                  1,
                                                  1));

  real alpha = 1.0f;
  real beta = 0.0f;
  CHECK_CUDNN(dynload::cudnnSoftmaxForward(t_resource.cudnn_handle,
                                           CUDNN_SOFTMAX_ACCURATE,
                                           CUDNN_SOFTMAX_MODE_CHANNEL,
                                           &alpha,
                                           t_resource.cudnn_desc,
                                           input,
                                           &beta,
                                           t_resource.cudnn_desc,
                                           output));
Z
zhangjinchao01 已提交
876 877 878
  CHECK_SYNC("hl_softmax_forward failed");
}

879 880
void hl_softmax_backward(real* output_value,
                         real* output_grad,
Z
zhangjinchao01 已提交
881
                         int height,
882
                         int width) {
883
#ifndef PADDLE_TYPE_DOUBLE
884
  cudnnDataType_t data_type = CUDNN_DATA_FLOAT;
Z
zhangjinchao01 已提交
885
#else
886
  cudnnDataType_t data_type = CUDNN_DATA_DOUBLE;
Z
zhangjinchao01 已提交
887
#endif
888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908
  CHECK_CUDNN(dynload::cudnnSetTensor4dDescriptor(t_resource.cudnn_desc,
                                                  CUDNN_TENSOR_NCHW,
                                                  data_type,
                                                  height,
                                                  width,
                                                  1,
                                                  1));

  real alpha = 1.0f;
  real beta = 0.0f;
  CHECK_CUDNN(dynload::cudnnSoftmaxBackward(t_resource.cudnn_handle,
                                            CUDNN_SOFTMAX_ACCURATE,
                                            CUDNN_SOFTMAX_MODE_CHANNEL,
                                            &alpha,
                                            t_resource.cudnn_desc,
                                            output_value,
                                            t_resource.cudnn_desc,
                                            output_grad,
                                            &beta,
                                            t_resource.cudnn_desc,
                                            output_grad));
Z
zhangjinchao01 已提交
909 910 911 912
  CHECK_SYNC("hl_softmax_backward failed");
}

void hl_batch_norm_forward_training(hl_tensor_descriptor inputDesc,
913
                                    real* input,
Z
zhangjinchao01 已提交
914
                                    hl_tensor_descriptor outputDesc,
915
                                    real* output,
Z
zhangjinchao01 已提交
916
                                    hl_tensor_descriptor bnParamDesc,
917 918
                                    real* scale,
                                    real* bias,
Z
zhangjinchao01 已提交
919
                                    double factor,
920 921
                                    real* runningMean,
                                    real* runningInvVar,
Z
zhangjinchao01 已提交
922
                                    double epsilon,
923 924
                                    real* savedMean,
                                    real* savedVar) {
925
#if CUDNN_VERSION >= 4007
Z
zhangjinchao01 已提交
926 927 928
  if ((NULL != runningMean && NULL == runningInvVar) ||
      (NULL == runningMean && NULL != runningInvVar)) {
    LOG(FATAL) << "runningMean and runningInvVar can be NULL "
929
               << "but only at the same time.";
Z
zhangjinchao01 已提交
930 931 932 933 934 935 936 937 938 939 940 941 942
  }
  if ((NULL != savedMean && NULL == savedVar) ||
      (NULL == savedMean && NULL != savedVar)) {
    LOG(FATAL) << "savedMean and savedVar can be NULL "
               << "but only at the same time.";
  }

  cudnnTensorDescriptor_t xDesc = GET_TENSOR_DESCRIPTOR(inputDesc);
  cudnnTensorDescriptor_t yDesc = GET_TENSOR_DESCRIPTOR(outputDesc);
  cudnnTensorDescriptor_t bnDesc = GET_TENSOR_DESCRIPTOR(bnParamDesc);
  real alpha = 1.0f;
  real beta = 1.0f;
  cudnnBatchNormMode_t mode = CUDNN_BATCHNORM_SPATIAL;
943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960
  CHECK_CUDNN(
      dynload::cudnnBatchNormalizationForwardTraining(t_resource.cudnn_handle,
                                                      mode,
                                                      &alpha,
                                                      &beta,
                                                      xDesc,
                                                      input,
                                                      yDesc,
                                                      output,
                                                      bnDesc,
                                                      scale,
                                                      bias,
                                                      factor,
                                                      runningMean,
                                                      runningInvVar,
                                                      epsilon,
                                                      savedMean,
                                                      savedVar));
Z
zhangjinchao01 已提交
961 962 963

  CHECK_SYNC("hl_batch_norm_forward_training failed");
#else
964
  LOG(FATAL) << "CudnnBatchNorm requires cudnn version >= 4007. "
Z
zhangjinchao01 已提交
965 966 967 968 969
             << "But cudnn lib version is " << g_cudnn_lib_version;
#endif
}

void hl_batch_norm_forward_inference(hl_tensor_descriptor inputDesc,
970 971 972 973 974 975 976 977 978
                                     real* input,
                                     hl_tensor_descriptor outputDesc,
                                     real* output,
                                     hl_tensor_descriptor bnParamDesc,
                                     real* scale,
                                     real* bias,
                                     real* estimatedMean,
                                     real* estimatedInvVar,
                                     double epsilon) {
979
#if CUDNN_VERSION >= 4007
Z
zhangjinchao01 已提交
980 981 982 983 984 985
  cudnnTensorDescriptor_t xDesc = GET_TENSOR_DESCRIPTOR(inputDesc);
  cudnnTensorDescriptor_t yDesc = GET_TENSOR_DESCRIPTOR(outputDesc);
  cudnnTensorDescriptor_t bnDesc = GET_TENSOR_DESCRIPTOR(bnParamDesc);
  real alpha = 1.0f;
  real beta = 1.0f;
  cudnnBatchNormMode_t mode = CUDNN_BATCHNORM_SPATIAL;
986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000
  CHECK_CUDNN(
      dynload::cudnnBatchNormalizationForwardInference(t_resource.cudnn_handle,
                                                       mode,
                                                       &alpha,
                                                       &beta,
                                                       xDesc,
                                                       input,
                                                       yDesc,
                                                       output,
                                                       bnDesc,
                                                       scale,
                                                       bias,
                                                       estimatedMean,
                                                       estimatedInvVar,
                                                       epsilon));
Z
zhangjinchao01 已提交
1001 1002 1003

  CHECK_SYNC("hl_batch_norm_forward_inference failed");
#else
1004
  LOG(FATAL) << "CudnnBatchNorm requires cudnn version >= 4007. "
Z
zhangjinchao01 已提交
1005 1006 1007 1008 1009
             << "But cudnn lib version is " << g_cudnn_lib_version;
#endif
}

void hl_batch_norm_backward(hl_tensor_descriptor inputDesc,
1010
                            real* input,
Z
zhangjinchao01 已提交
1011
                            hl_tensor_descriptor outGradDesc,
1012
                            real* outGrad,
Z
zhangjinchao01 已提交
1013
                            hl_tensor_descriptor inGradDesc,
1014
                            real* inGrad,
Z
zhangjinchao01 已提交
1015
                            hl_tensor_descriptor dBnParamDesc,
1016 1017 1018
                            real* scale,
                            real* scaleGrad,
                            real* biasGrad,
Z
zhangjinchao01 已提交
1019
                            double epsilon,
1020 1021
                            real* savedMean,
                            real* savedInvVar) {
1022
#if CUDNN_VERSION >= 4007
Z
zhangjinchao01 已提交
1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035
  if ((NULL != savedMean && NULL == savedInvVar) ||
      (NULL == savedMean && NULL != savedInvVar)) {
    LOG(FATAL) << "savedMean and savedVar can be NULL "
               << "but only at the same time.";
  }

  cudnnTensorDescriptor_t xDesc = GET_TENSOR_DESCRIPTOR(inputDesc);
  cudnnTensorDescriptor_t dyDesc = GET_TENSOR_DESCRIPTOR(outGradDesc);
  cudnnTensorDescriptor_t dxDesc = GET_TENSOR_DESCRIPTOR(inGradDesc);
  cudnnTensorDescriptor_t bnDesc = GET_TENSOR_DESCRIPTOR(dBnParamDesc);
  real alpha = 1.0f;
  real beta = 1.0f;
  cudnnBatchNormMode_t mode = CUDNN_BATCHNORM_SPATIAL;
1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
  CHECK_CUDNN(dynload::cudnnBatchNormalizationBackward(t_resource.cudnn_handle,
                                                       mode,
                                                       &alpha,
                                                       &beta,
                                                       &alpha,
                                                       &beta,
                                                       xDesc,
                                                       input,
                                                       dyDesc,
                                                       outGrad,
                                                       dxDesc,
                                                       inGrad,
                                                       bnDesc,
                                                       scale,
                                                       scaleGrad,
                                                       biasGrad,
                                                       epsilon,
                                                       savedMean,
                                                       savedInvVar));
Z
zhangjinchao01 已提交
1055 1056 1057

  CHECK_SYNC("hl_batch_norm_backward failed");
#else
1058
  LOG(FATAL) << "CudnnBatchNorm requires cudnn version >= 4007. "
Z
zhangjinchao01 已提交
1059 1060 1061
             << "But cudnn lib version is " << g_cudnn_lib_version;
#endif
}