kernel_factory.cc 14.3 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14
//   Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
//
// 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.

15
#include "paddle/phi/core/kernel_factory.h"
16

17
#include "glog/logging.h"
18
#include "paddle/phi/core/enforce.h"
19
#if defined(PADDLE_WITH_XPU) && !defined(PADDLE_WITH_XPU_KP)
20
#include "paddle/phi/backends/xpu/xpu_op_list.h"
21 22
#include "paddle/phi/core/compat/convert_utils.h"
#endif
23
#include "paddle/phi/core/compat/op_utils.h"
24
#include "paddle/utils/string/string_helper.h"
25

26
DECLARE_int32(low_precision_op_list);
27 28
DECLARE_bool(enable_api_kernel_fallback);

29
namespace phi {
30

31 32
const static Kernel empty_kernel;  // NOLINT

33 34
std::string KernelSelectionErrorMessage(const std::string& kernel_name,
                                        const KernelKey& target_key);
35

36 37 38 39 40 41 42 43 44
uint32_t KernelKey::Hash::operator()(const KernelKey& key) const {
  uint32_t hash_value = 0;
  // |----31-20------|---19-12---|---11-8----|---7-0---|
  // | For extension | DataType | DataLayout | Backend |
  hash_value |= static_cast<uint8_t>(key.backend());
  hash_value |=
      (static_cast<uint8_t>(key.layout()) << KernelKey::kBackendBitLength);
  hash_value |=
      (static_cast<uint16_t>(key.dtype())
45
       << (KernelKey::kBackendBitLength + KernelKey::kDataLayoutBitLength));
46 47 48 49 50 51 52 53
  return hash_value;
}

KernelFactory& KernelFactory::Instance() {
  static KernelFactory g_op_kernel_factory;
  return g_op_kernel_factory;
}

54 55 56 57 58 59 60 61 62 63 64
bool KernelFactory::HasCompatiblePhiKernel(const std::string& op_type) const {
  if (deprecated_op_names.find(op_type) == deprecated_op_names.end()) {
    if (phi::OpUtilsMap::Instance().Contains(op_type)) {
      return true;
    } else if (kernels_.find(op_type) != kernels_.end()) {
      return true;
    }
  }
  return false;
}

65 66
const Kernel& KernelFactory::SelectKernel(const std::string& kernel_name,
                                          const KernelKey& kernel_key) const {
67 68
  auto iter = kernels_.find(kernel_name);
  if (iter == kernels_.end()) {
69
    return empty_kernel;
70 71
  }
  auto kernel_iter = iter->second.find(kernel_key);
72 73 74 75 76 77 78
  if (kernel_iter == iter->second.end() &&
      kernel_key.layout() != phi::DataLayout::ALL_LAYOUT) {
    phi::KernelKey any_layout_kernel_key(
        kernel_key.backend(), phi::DataLayout::ALL_LAYOUT, kernel_key.dtype());
    kernel_iter = iter->second.find(any_layout_kernel_key);
  }

79
  if (kernel_iter == iter->second.end()) {
80
    return empty_kernel;
81
  }
82

83 84 85
  return kernel_iter->second;
}

86 87
KernelKeyMap KernelFactory::SelectKernelMap(
    const std::string& kernel_name) const {
88 89
  auto iter = kernels_.find(kernel_name);
  if (iter == kernels_.end()) {
90
    return KernelKeyMap();
91 92 93 94
  }
  return iter->second;
}

95 96
bool KernelFactory::HasKernel(const std::string& kernel_name,
                              const KernelKey& kernel_key) const {
97 98 99 100 101 102 103 104 105 106 107 108 109
  auto iter = kernels_.find(kernel_name);
  PADDLE_ENFORCE_NE(
      iter,
      kernels_.end(),
      phi::errors::NotFound("The kernel `%s` is not registered.", kernel_name));

  auto kernel_iter = iter->second.find(kernel_key);
  if (kernel_iter == iter->second.end()) {
    return false;
  }
  return true;
}

110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132
void KernelFactory::AddToLowPrecisionKernelList(
    const std::string& name,
    const paddle::experimental::DataType& kernel_key_type) {
  if (FLAGS_low_precision_op_list >= 1) {
    auto op_name = phi::TransToFluidOpName(name);
    if (op_name.find("_grad") != std::string::npos) {
      return;  // only record forward api
    }
    bool is_low_precision =
        (kernel_key_type == paddle::experimental::DataType::FLOAT16 ||
         kernel_key_type == paddle::experimental::DataType::BFLOAT16);
    bool need_record =
        FLAGS_low_precision_op_list == 1 ? is_low_precision : true;
    if (need_record) {
      low_precision_kernels_[op_name] += 1;
    }
  }
}

std::map<const std::string, int> KernelFactory::GetLowPrecisionKernelList() {
  return low_precision_kernels_;
}

133
KernelResult KernelFactory::SelectKernelOrThrowError(
134
    const std::string& kernel_name, const KernelKey& const_kernel_key) const {
135
  auto iter = kernels_.find(kernel_name);
136

137 138 139 140
  PADDLE_ENFORCE_NE(
      iter,
      kernels_.end(),
      phi::errors::NotFound("The kernel `%s` is not registered.", kernel_name));
141

142 143 144
  KernelKey kernel_key = KernelKey(const_kernel_key.backend(),
                                   phi::DataLayout::ALL_LAYOUT,
                                   const_kernel_key.dtype());
Z
zyfncg 已提交
145
#if defined(PADDLE_WITH_CUDA) || defined(PADDLE_WITH_HIP)
146
  if (kernel_key.backend() == Backend::GPUDNN) {
Z
zyfncg 已提交
147
    auto kernel_iter = iter->second.find(
148
        {Backend::GPUDNN, phi::DataLayout::ALL_LAYOUT, kernel_key.dtype()});
Z
zyfncg 已提交
149
    if (kernel_iter != iter->second.end()) {
150
      return {kernel_iter->second, false};
Z
zyfncg 已提交
151
    }
152 153
    kernel_key =
        KernelKey(Backend::GPU, kernel_key.layout(), kernel_key.dtype());
Z
zyfncg 已提交
154 155
  }
#endif
156
  auto kernel_iter = iter->second.find(kernel_key);
157

158 159 160 161
  PADDLE_ENFORCE_NE(
      kernel_iter == iter->second.end() && kernel_key.backend() == Backend::CPU,
      true,
      phi::errors::NotFound(
162
          "The kernel with key %s of kernel `%s` is not registered. %s",
163
          kernel_key,
164
          kernel_name,
165
          KernelSelectionErrorMessage(kernel_name, kernel_key)));
166

167
#if defined(PADDLE_WITH_XPU) && !defined(PADDLE_WITH_XPU_KP)
168 169
  VLOG(6) << "fluid_op_name: " << TransToFluidOpName(kernel_name);
  if ((FLAGS_enable_api_kernel_fallback && kernel_iter == iter->second.end()) ||
Q
QingshuChen 已提交
170 171
      !phi::backends::xpu::is_xpu_support_op(TransToFluidOpName(kernel_name),
                                             kernel_key.dtype())
172 173
#else
  if ((FLAGS_enable_api_kernel_fallback && kernel_iter == iter->second.end())
174 175
#endif
  ) {
176 177 178 179
    // Fallback CPU backend
    phi::KernelKey cpu_kernel_key(
        phi::Backend::CPU, kernel_key.layout(), kernel_key.dtype());
    kernel_iter = iter->second.find(cpu_kernel_key);
180 181 182 183 184

    PADDLE_ENFORCE_NE(
        kernel_iter,
        iter->second.end(),
        phi::errors::NotFound(
185 186
            "The kernel with key %s of kernel `%s` is not registered and "
            "fail to fallback to CPU one. %s",
187
            kernel_key,
188
            kernel_name,
189
            KernelSelectionErrorMessage(kernel_name, kernel_key)));
190 191 192 193 194 195

    VLOG(3) << "missing " << kernel_key.backend() << " kernel: " << kernel_name
            << ", expected_kernel_key:" << kernel_key
            << ", fallbacking to CPU one!";

    return {kernel_iter->second, true};
196 197
  }

198 199 200
  PADDLE_ENFORCE_NE(
      kernel_iter,
      iter->second.end(),
201
      phi::errors::NotFound(
202 203
          "The kernel with key %s of kernel `%s` is not registered. %s "
          "The current value of FLAGS_enable_api_kernel_fallback(bool,"
204 205
          " default true) is false. If you want to fallback this kernel"
          " to CPU one, please set the flag true before run again.",
206
          kernel_key,
207
          kernel_name,
208
          KernelSelectionErrorMessage(kernel_name, kernel_key)));
209

210
  return {kernel_iter->second, false};
211 212
}

213 214 215 216 217 218 219 220 221 222
const KernelArgsDef& KernelFactory::GetFirstKernelArgsDef(
    const std::string& kernel_name) const {
  auto iter = kernels_.find(kernel_name);
  PADDLE_ENFORCE_NE(
      iter,
      kernels_.end(),
      phi::errors::NotFound("The kernel `%s` is not registered.", kernel_name));
  return iter->second.cbegin()->second.args_def();
}

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
std::ostream& operator<<(std::ostream& os, AttributeType attr_type) {
  switch (attr_type) {
    case AttributeType::BOOL:
      os << "bool";
      break;
    case AttributeType::INT32:
      os << "int";
      break;
    case AttributeType::INT64:
      os << "int64_t";
      break;
    case AttributeType::FLOAT32:
      os << "float";
      break;
    case AttributeType::FLOAT64:
      os << "double";
      break;
    case AttributeType::STRING:
      os << "string";
      break;
    case AttributeType::BOOLS:
      os << "vector<bool>";
      break;
    case AttributeType::INT32S:
      os << "vector<int>";
      break;
    case AttributeType::INT64S:
      os << "vector<int64_t>";
      break;
    case AttributeType::FLOAT32S:
      os << "vector<float>";
      break;
    case AttributeType::FLOAT64S:
      os << "vector<double>";
      break;
    case AttributeType::STRINGS:
      os << "vector<string>";
      break;
    case AttributeType::SCALAR:
      os << "Scalar";
      break;
    case AttributeType::SCALARS:
      os << "vector<Scalar>";
      break;
    case AttributeType::INT_ARRAY:
      os << "IntArray";
      break;
    case AttributeType::DATA_TYPE:
      os << "DataType";
      break;
    case AttributeType::DATA_LAYOUT:
      os << "DataLayout";
      break;
    case AttributeType::PLACE:
      os << "Place";
      break;
    default:
      os << "Undefined";
  }
  return os;
}

285 286 287 288 289 290 291
// print kernel info with json format:
// {
//   "(CPU, Undefined(AnyLayout), complex64)": {
//   "input": ["CPU, NCHW, complex64", "CPU, NCHW, complex64"],
//   "output": ["CPU, NCHW, complex64"],
//   "attribute": ["i"]
// }
292
std::ostream& operator<<(std::ostream& os, const Kernel& kernel) {
293 294 295
  // input
  os << "{\"input\":[";
  bool need_comma = false;
296
  for (auto& in_def : kernel.args_def().input_defs()) {
297 298 299 300
    if (need_comma) os << ",";
    os << "\"" << in_def.backend << ", " << in_def.layout << ", "
       << in_def.dtype << "\"";
    need_comma = true;
301
  }
302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319
  os << "],";

  // output
  os << "\"output\":[";
  need_comma = false;
  for (auto& out_def : kernel.args_def().output_defs()) {
    if (need_comma) os << ",";
    os << "\"" << out_def.backend << ", " << out_def.layout << ", "
       << out_def.dtype << "\"";
    need_comma = true;
  }
  os << "],";

  // attr
  os << "\"attribute\":[";
  need_comma = false;
  for (auto& arg_def : kernel.args_def().attribute_defs()) {
    if (need_comma) os << ",";
320
    os << "\"" << arg_def.type_index << "\"";
321 322 323 324
    need_comma = true;
  }
  os << "]}";

325 326 327
  return os;
}

328 329 330 331 332 333 334 335 336 337 338 339 340 341 342
// print all kernels info with json format:
// {
//  "kernel_name1":
//      [
//        {
//          "(CPU, Undefined(AnyLayout), complex64)": {
//          "input": ["CPU, NCHW, complex64", "CPU, NCHW, complex64"],
//          "output": ["CPU, NCHW, complex64"],
//          "attribute": ["i"]
//        },
//        ...
//      ],
//    "kernel_name2": []
//    ...
// }
343
std::ostream& operator<<(std::ostream& os, KernelFactory& kernel_factory) {
344 345
  os << "{";
  bool need_comma_kernels = false;
346
  for (const auto& op_kernel_pair : kernel_factory.kernels()) {
347 348 349 350 351
    if (need_comma_kernels) {
      os << ",";
      os << std::endl;
    }
    os << "\"" << op_kernel_pair.first << " \":[" << std::endl;
352
    bool need_comma_per_kernel = false;
353
    for (const auto& kernel_pair : op_kernel_pair.second) {
354 355 356 357
      if (need_comma_per_kernel) {
        os << ",";
        os << std::endl;
      }
358 359
      os << "{\"" << kernel_pair.first << "\":" << kernel_pair.second << "}";
      need_comma_per_kernel = true;
360
    }
361 362
    os << "]";
    need_comma_kernels = true;
363
  }
364 365
  os << "}";

366 367 368
  return os;
}

369 370 371 372 373 374 375 376 377 378 379 380
// return all kernel selection error message of specific kernel_name:
// 1. If target_key not supports target backend, output "Selected wrong Backend
// ..."
// 2. If target_key not supports target datatype, output "Selected wrong
// DataType ..."
// 3. `target_key` is still not supported, output all kernel keys of
// corresponding kernel_name:
// {
//   (CPU, NCHW, [int8, int16, ...]);
//   (GPU, Undefined(AnyLayout), [float32, float64, ...]);
//   ...
// }
381 382
std::string KernelSelectionErrorMessage(const std::string& kernel_name,
                                        const KernelKey& target_key) {
383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414
  PADDLE_ENFORCE_NE(
      KernelFactory::Instance().kernels().find(kernel_name),
      KernelFactory::Instance().kernels().end(),
      phi::errors::NotFound("The kernel `%s` is not registered.", kernel_name));

  // Init data structure
  bool support_backend = false;
  bool support_dtype = false;
  std::unordered_map<std::string, std::vector<std::string>> all_kernel_key;
  std::unordered_set<std::string> backend_set;
  std::unordered_set<std::string> dtype_set;

  // Record all kernel information of kernel_name
  for (auto iter : KernelFactory::Instance().kernels()[kernel_name]) {
    KernelKey kernel_key = iter.first;
    if (kernel_key.backend() == target_key.backend()) {
      support_backend = true;
      if (kernel_key.dtype() == target_key.dtype()) {
        support_dtype = true;
      }
      dtype_set.insert(
          paddle::experimental::DataTypeToString(kernel_key.dtype()));
    }
    backend_set.insert(
        paddle::experimental::BackendToString(kernel_key.backend()));
    all_kernel_key[paddle::experimental::BackendToString(kernel_key.backend()) +
                   ", " + phi::DataLayoutToString(kernel_key.layout())]
        .push_back(paddle::experimental::DataTypeToString(kernel_key.dtype()));
  }
  // 1. If target_key not supports target backend, output "Selected wrong
  // Backend ..."
  if (!support_backend) {
415
    std::string error_message = paddle::string::join_strings(backend_set, ", ");
416 417 418 419 420 421 422
    return "Selected wrong Backend `" +
           paddle::experimental::BackendToString(target_key.backend()) +
           "`. Paddle support following Backends: " + error_message + ".";
  }
  // 2. If target_key not supports target datatype, output "Selected wrong
  // DataType ..."
  if (!support_dtype) {
423
    std::string error_message = paddle::string::join_strings(dtype_set, ", ");
424 425 426 427 428 429 430 431 432 433 434
    return "Selected wrong DataType `" +
           paddle::experimental::DataTypeToString(target_key.dtype()) +
           "`. Paddle support following DataTypes: " + error_message + ".";
  }
  // 3. `target_key` is still not supported, output all kernel keys of
  // corresponding kernel_name
  std::string message = "Currently, paddle support following kernel keys of `" +
                        kernel_name + "`: { ";
  for (auto iter = all_kernel_key.begin(); iter != all_kernel_key.end();
       ++iter) {
    std::vector<std::string>& dtype_vec = iter->second;
435 436
    message += "(" + iter->first + ", [";
    message += paddle::string::join_strings(dtype_vec, ", ");
437 438 439 440 441 442
    message += "]); ";
  }
  message += "}.";
  return message;
}

443
}  // namespace phi