executor.cc 15.8 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 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
Yi Wang 已提交
15
#include "paddle/fluid/framework/executor.h"
Y
Yang Yang 已提交
16

17
#include "paddle/fluid/framework/channel.h"
Y
Yi Wang 已提交
18 19 20 21 22
#include "paddle/fluid/framework/feed_fetch_method.h"
#include "paddle/fluid/framework/lod_rank_table.h"
#include "paddle/fluid/framework/lod_tensor_array.h"
#include "paddle/fluid/framework/op_registry.h"
#include "paddle/fluid/framework/reader.h"
G
gongweibao 已提交
23
#include "paddle/fluid/operators/detail/macros.h"
Y
Yi Wang 已提交
24
#include "paddle/fluid/platform/place.h"
X
Xin Pan 已提交
25
#include "paddle/fluid/platform/profiler.h"
Y
Yang Yu 已提交
26

D
dzhwinter 已提交
27
DECLARE_bool(benchmark);
28
DEFINE_bool(use_mkldnn, false, "Use MKLDNN to run");
Q
qijun 已提交
29 30 31

namespace paddle {
namespace framework {
X
Xin Pan 已提交
32 33 34 35 36
namespace {
// block id starts from 0. This id is used to represent the codeblock
// wrapping the first block 0.
int kProgramId = -1;
}  // namespace
Q
qijun 已提交
37

Q
Qiao Longfei 已提交
38 39 40
ExecutorPrepareContext::ExecutorPrepareContext(
    const framework::ProgramDesc& prog, size_t block_id)
    : prog_(prog), block_id_(block_id) {}
Y
Yu Yang 已提交
41

Q
Qiao Longfei 已提交
42 43 44
ExecutorPrepareContext::~ExecutorPrepareContext() {
  VLOG(5) << "destroy ExecutorPrepareContext";
}
Y
Yu Yang 已提交
45

D
dzhwinter 已提交
46
Executor::Executor(const platform::Place& place) : place_(place) {}
Q
qijun 已提交
47

W
Wu Yi 已提交
48
#ifdef PADDLE_WITH_DISTRIBUTE
Y
Yancey1989 已提交
49
void Executor::BeginPass() {
Y
Yancey1989 已提交
50 51 52
  ::paddle::operators::distributed::RPCClient::GetInstance<
      ::paddle::operators::distributed::GRPCClient>()
      ->SendBeginPass();
Y
Yancey1989 已提交
53 54 55
}

void Executor::EndPass() {
Y
Yancey1989 已提交
56 57 58
  ::paddle::operators::distributed::RPCClient::GetInstance<
      ::paddle::operators::distributed::GRPCClient>()
      ->SendEndPass();
W
Wu Yi 已提交
59 60 61
}
#endif

Y
Stash  
Yu Yang 已提交
62
void InitializeVariable(Variable* var, proto::VarType::Type var_type) {
63
  if (var_type == proto::VarType::LOD_TENSOR) {
Q
QI JUN 已提交
64
    var->GetMutable<LoDTensor>();
65
  } else if (var_type == proto::VarType::SELECTED_ROWS) {
Q
QI JUN 已提交
66
    var->GetMutable<SelectedRows>();
67
  } else if (var_type == proto::VarType::FEED_MINIBATCH) {
Q
QI JUN 已提交
68
    var->GetMutable<FeedFetchList>();
69
  } else if (var_type == proto::VarType::FETCH_LIST) {
Q
QI JUN 已提交
70
    var->GetMutable<FeedFetchList>();
71
  } else if (var_type == proto::VarType::STEP_SCOPES) {
Y
Yu Yang 已提交
72
    var->GetMutable<std::vector<framework::Scope>>();
73
  } else if (var_type == proto::VarType::LOD_RANK_TABLE) {
Y
Yu Yang 已提交
74
    var->GetMutable<LoDRankTable>();
75
  } else if (var_type == proto::VarType::LOD_TENSOR_ARRAY) {
Y
Yu Yang 已提交
76
    var->GetMutable<LoDTensorArray>();
77
  } else if (var_type == proto::VarType::PLACE_LIST) {
Y
Yang Yu 已提交
78
    var->GetMutable<platform::PlaceList>();
79
  } else if (var_type == proto::VarType::READER) {
F
fengjiayi 已提交
80
    var->GetMutable<ReaderHolder>();
81 82
  } else if (var_type == proto::VarType::CHANNEL) {
    var->GetMutable<ChannelHolder>();
T
typhoonzero 已提交
83 84
  } else if (var_type == proto::VarType::RAW) {
    // GetMutable will be called in operator
Q
QI JUN 已提交
85 86
  } else {
    PADDLE_THROW(
Y
Yu Yang 已提交
87
        "Variable type %d is not in "
F
fengjiayi 已提交
88
        "[LOD_TENSOR, SELECTED_ROWS, FEED_MINIBATCH, FETCH_LIST, "
T
typhoonzero 已提交
89
        "LOD_RANK_TABLE, PLACE_LIST, READER, CHANNEL, RAW]",
Y
Yu Yang 已提交
90
        var_type);
Q
QI JUN 已提交
91 92 93
  }
}

L
Liu Yiqun 已提交
94 95 96
void Executor::CreateVariables(const ProgramDesc& pdesc, Scope* scope,
                               int block_id) {
  auto& global_block = pdesc.Block(block_id);
97 98 99 100 101 102 103 104 105 106 107 108 109 110

  const Scope* ancestor_scope = scope;
  while (ancestor_scope->parent()) {
    ancestor_scope = ancestor_scope->parent();
  }

  if (ancestor_scope != scope) {
    for (auto& var : global_block.AllVars()) {
      if (var->Name() == framework::kEmptyVarName) {
        continue;
      }

      if (var->Persistable()) {
        auto* ptr = const_cast<Scope*>(ancestor_scope)->Var(var->Name());
111
        InitializeVariable(ptr, var->GetType());
112 113 114 115
        VLOG(3) << "Create Variable " << var->Name()
                << " global, which pointer is " << ptr;
      } else {
        auto* ptr = scope->Var(var->Name());
116
        InitializeVariable(ptr, var->GetType());
117 118 119 120 121 122 123
        VLOG(3) << "Create Variable " << var->Name()
                << " locally, which pointer is " << ptr;
      }
    }
  } else {
    for (auto& var : global_block.AllVars()) {
      auto* ptr = scope->Var(var->Name());
124
      InitializeVariable(ptr, var->GetType());
125 126 127 128 129 130
      VLOG(3) << "Create variable " << var->Name() << ", which pointer is "
              << ptr;
    }
  }
}

Y
Yu Yang 已提交
131
void Executor::Run(const ProgramDesc& pdesc, Scope* scope, int block_id,
T
typhoonzero 已提交
132
                   bool create_local_scope, bool create_vars) {
X
Xin Pan 已提交
133
  platform::RecordBlock b(block_id);
134
  if (FLAGS_use_mkldnn) EnableMKLDNN(pdesc);
Q
Qiao Longfei 已提交
135 136
  auto ctx = Prepare(pdesc, block_id);
  RunPreparedContext(ctx.get(), scope, create_local_scope, create_vars);
Q
qijun 已提交
137 138
}

139 140 141 142 143 144 145
// Check whether the block already has feed operators and feed_holder.
// Return false if the block does not have any feed operators.
// If some feed operators have been prepended to the block, check that
// the info contained in these feed operators matches the feed_targets
// and feed_holder_name. Raise exception when any mismatch is found.
// Return true if the block has feed operators and holder of matching info.
static bool has_feed_operators(
146
    const BlockDesc& block,
L
Liu Yiqun 已提交
147
    const std::map<std::string, const LoDTensor*>& feed_targets,
148 149
    const std::string& feed_holder_name) {
  size_t feed_count = 0;
150
  for (auto* op : block.AllOps()) {
151 152
    if (op->Type() == kFeedOpType) {
      feed_count++;
L
Liu Yiqun 已提交
153
      // The input variable's name of feed_op should be feed_holder_name.
154 155 156 157 158 159 160 161 162 163 164 165 166 167 168
      PADDLE_ENFORCE_EQ(op->Input("X")[0], feed_holder_name,
                        "Input to feed op should be '%s'", feed_holder_name);
      std::string feed_target_name = op->Output("Out")[0];
      PADDLE_ENFORCE(
          feed_targets.find(feed_target_name) != feed_targets.end(),
          "Feed operator output name '%s' cannot be found in 'feed_targets'",
          feed_target_name);
    }
  }

  if (feed_count > 0) {
    PADDLE_ENFORCE_EQ(
        feed_count, feed_targets.size(),
        "The number of feed operators should match 'feed_targets'");

169
    if (!feed_holder_name.empty()) {
L
Liu Yiqun 已提交
170
      // When feed operator are present, so should be feed_holder.
171 172 173 174 175 176 177
      auto var = block.FindVar(feed_holder_name);
      PADDLE_ENFORCE_NOT_NULL(var, "Block should already have a '%s' variable",
                              feed_holder_name);
      PADDLE_ENFORCE_EQ(var->GetType(), proto::VarType::FEED_MINIBATCH,
                        "'%s' variable should be 'FEED_MINIBATCH' type",
                        feed_holder_name);
    }
178 179 180 181 182 183 184 185 186 187 188 189
  }

  return feed_count > 0;
}

// Check whether the block already has fetch operators and fetch_holder.
// Return false if the block does not have any fetch operators.
// If some fetch operators have been appended to the block, check that
// the info contained in these fetch operators matches the fetch_targets
// and fetch_holder_name. Raise exception when any mismatch is found.
// Return true if the block has fetch operators and holder of matching info.
static bool has_fetch_operators(
L
Liu Yiqun 已提交
190 191
    const BlockDesc& block,
    const std::map<std::string, LoDTensor*>& fetch_targets,
192 193
    const std::string& fetch_holder_name) {
  size_t fetch_count = 0;
194
  for (auto* op : block.AllOps()) {
195 196
    if (op->Type() == kFetchOpType) {
      fetch_count++;
L
Liu Yiqun 已提交
197
      // The output variable's name of fetch_op should be fetch_holder_name.
198 199 200 201 202 203 204 205 206 207 208 209 210 211 212
      PADDLE_ENFORCE_EQ(op->Output("Out")[0], fetch_holder_name,
                        "Output of fetch op should be '%s'", fetch_holder_name);
      std::string fetch_target_name = op->Input("X")[0];
      PADDLE_ENFORCE(
          fetch_targets.find(fetch_target_name) != fetch_targets.end(),
          "Fetch operator input name '%s' cannot be found in 'fetch_targets'",
          fetch_target_name);
    }
  }

  if (fetch_count > 0) {
    PADDLE_ENFORCE_EQ(
        fetch_count, fetch_targets.size(),
        "The number of fetch operators should match 'fetch_targets'");

213
    if (!fetch_holder_name.empty()) {
L
Liu Yiqun 已提交
214
      // When fetch operator are present, so should be fetch_holder.
215 216 217 218 219 220 221
      auto var = block.FindVar(fetch_holder_name);
      PADDLE_ENFORCE_NOT_NULL(var, "Block should already have a '%s' variable",
                              fetch_holder_name);
      PADDLE_ENFORCE_EQ(var->GetType(), proto::VarType::FETCH_LIST,
                        "'%s' variable should be 'FETCH_LIST' type",
                        fetch_holder_name);
    }
222 223 224 225 226 227
  }

  return fetch_count > 0;
}

void Executor::Run(const ProgramDesc& program, Scope* scope,
228 229
                   std::map<std::string, const LoDTensor*>* feed_targets,
                   std::map<std::string, LoDTensor*>* fetch_targets,
W
Wu Yi 已提交
230 231
                   bool create_local_scope, bool create_vars,
                   const std::string& feed_holder_name,
232
                   const std::string& fetch_holder_name) {
X
Xin Pan 已提交
233
  platform::RecordBlock b(kProgramId);
234
  if (FLAGS_use_mkldnn) EnableMKLDNN(program);
235
  bool has_feed_ops =
236
      has_feed_operators(program.Block(0), *feed_targets, feed_holder_name);
237
  bool has_fetch_ops =
238
      has_fetch_operators(program.Block(0), *fetch_targets, fetch_holder_name);
239 240

  ProgramDesc* copy_program = const_cast<ProgramDesc*>(&program);
S
sneaxiy 已提交
241
  std::unique_ptr<ProgramDesc> unique_ptr_of_copy_program;
242
  if (!has_feed_ops || !has_fetch_ops) {
S
sneaxiy 已提交
243 244
    unique_ptr_of_copy_program.reset(new ProgramDesc(program));
    copy_program = unique_ptr_of_copy_program.get();
245
  }
246 247
  auto* global_block = copy_program->MutableBlock(0);

248
  if (!has_feed_ops) {
249 250
    // create feed_holder variable
    auto* feed_holder = global_block->Var(feed_holder_name);
251
    feed_holder->SetType(proto::VarType::FEED_MINIBATCH);
252 253 254
    feed_holder->SetPersistable(true);

    int i = 0;
255
    for (auto& feed_target : (*feed_targets)) {
256 257 258 259 260 261 262 263 264 265 266 267 268 269 270
      std::string var_name = feed_target.first;
      VLOG(3) << "feed target's name: " << var_name;

      // prepend feed op
      auto* op = global_block->PrependOp();
      op->SetType(kFeedOpType);
      op->SetInput("X", {feed_holder_name});
      op->SetOutput("Out", {var_name});
      op->SetAttr("col", {static_cast<int>(i)});
      op->CheckAttrs();

      i++;
    }
  }

271
  if (!has_fetch_ops) {
272 273
    // create fetch_holder variable
    auto* fetch_holder = global_block->Var(fetch_holder_name);
274
    fetch_holder->SetType(proto::VarType::FETCH_LIST);
275 276 277
    fetch_holder->SetPersistable(true);

    int i = 0;
278
    for (auto& fetch_target : (*fetch_targets)) {
279 280 281 282 283 284 285 286 287 288 289 290 291 292 293
      std::string var_name = fetch_target.first;
      VLOG(3) << "fetch target's name: " << var_name;

      // append fetch op
      auto* op = global_block->AppendOp();
      op->SetType(kFetchOpType);
      op->SetInput("X", {var_name});
      op->SetOutput("Out", {fetch_holder_name});
      op->SetAttr("col", {static_cast<int>(i)});
      op->CheckAttrs();

      i++;
    }
  }

294
  auto ctx = Prepare(*copy_program, 0);
W
Wu Yi 已提交
295 296 297
  RunPreparedContext(ctx.get(), scope, feed_targets, fetch_targets,
                     create_local_scope, create_vars, feed_holder_name,
                     fetch_holder_name);
298 299
}

Q
Qiao Longfei 已提交
300 301
std::unique_ptr<ExecutorPrepareContext> Executor::Prepare(
    const ProgramDesc& program, int block_id) {
Q
Qiyang Min 已提交
302 303
  std::unique_ptr<ExecutorPrepareContext> ctx(
      new ExecutorPrepareContext(program, block_id));
Y
Yu Yang 已提交
304 305 306 307 308
  PADDLE_ENFORCE_LT(static_cast<size_t>(block_id), program.Size());
  auto& block = program.Block(block_id);
  for (auto& op_desc : block.AllOps()) {
    ctx->ops_.push_back(OpRegistry::CreateOp(*op_desc));
  }
Q
Qiyang Min 已提交
309
  return ctx;
Y
Yu Yang 已提交
310 311
}

T
refine  
typhoonzero 已提交
312
std::vector<std::shared_ptr<ExecutorPrepareContext>> Executor::Prepare(
T
typhoonzero 已提交
313 314 315 316 317 318 319 320 321 322 323 324 325 326
    const ProgramDesc& program, const std::vector<int>& block_ids) {
  std::vector<std::shared_ptr<ExecutorPrepareContext>> result;
  for (auto& bid : block_ids) {
    auto* ctx = new ExecutorPrepareContext(program, bid);
    PADDLE_ENFORCE_LT(static_cast<size_t>(bid), program.Size());
    auto& block = program.Block(bid);
    for (auto& op_desc : block.AllOps()) {
      ctx->ops_.push_back(OpRegistry::CreateOp(*op_desc));
    }
    result.push_back(std::shared_ptr<ExecutorPrepareContext>(ctx));
  }
  return result;
}

Y
Yu Yang 已提交
327
void Executor::RunPreparedContext(ExecutorPrepareContext* ctx, Scope* scope,
Q
qiaolongfei 已提交
328 329
                                  bool create_local_scope, bool create_vars,
                                  bool keep_kids) {
Y
Yu Yang 已提交
330 331 332 333
  Scope* local_scope = scope;
  if (create_vars) {
    if (create_local_scope) {
      local_scope = &scope->NewScope();
334 335
    }
    CreateVariables(ctx->prog_, local_scope, ctx->block_id_);
L
Liu Yiqun 已提交
336
  }
Y
Yu Yang 已提交
337 338

  for (auto& op : ctx->ops_) {
Q
qiaolongfei 已提交
339
    VLOG(4) << place_ << " " << op->DebugStringEx(local_scope);
340
    op->Run(*local_scope, place_);
Q
qiaolongfei 已提交
341 342 343 344
    // NOTE! Please do not delete this line, it's usefull because the debug
    // string before and after op.run are different, after run the output
    // will have right shape which is usefull for debug.
    VLOG(3) << place_ << " " << op->DebugStringEx(local_scope);
Y
Yang Yang 已提交
345

Y
Yu Yang 已提交
346 347 348 349 350
    if (FLAGS_benchmark) {
      VLOG(2) << "Memory used after operator " + op->Type() + " running: "
              << memory::memory_usage(place_);
    }
  }
351
  platform::DeviceContextPool::Instance().Get(place_)->Wait();
Q
qiaolongfei 已提交
352
  if (local_scope != scope) {
Y
Yu Yang 已提交
353
    scope->DeleteScope(local_scope);
354
  } else {
Q
qiaolongfei 已提交
355 356 357 358 359
    if (!keep_kids) {
      // By default, we should delete all kid scopes after run executor because
      // some operators may create local scope when running, such as while_op.
      // But when while_op also create a local executor to run it's sub block,
      // the sub scopes it created should not be dropped immediately, because
Q
qiaolongfei 已提交
360 361
      // while_grad_op will use some variables created during while_op run, so
      // we need to keep the kids and wait for the outer executor to drop them.
Q
qiaolongfei 已提交
362 363
      scope->DropKids();
    }
Y
Yu Yang 已提交
364
  }
Q
qiaolongfei 已提交
365

Y
Yu Yang 已提交
366 367 368 369 370 371 372 373
  if (FLAGS_benchmark) {
    VLOG(2) << "-------------------------------------------------------";
    VLOG(2) << "Memory used after deleting local scope: "
            << memory::memory_usage(place_);
    VLOG(2) << "-------------------------------------------------------";
  }
}

374 375
void Executor::RunPreparedContext(
    ExecutorPrepareContext* ctx, Scope* scope,
376
    std::map<std::string, const LoDTensor*>* feed_targets,
W
Wu Yi 已提交
377 378 379
    std::map<std::string, LoDTensor*>* fetch_targets, bool create_local_scope,
    bool create_vars, const std::string& feed_holder_name,
    const std::string& fetch_holder_name) {
380 381
  auto& global_block = ctx->prog_.Block(ctx->block_id_);

382
  PADDLE_ENFORCE(
383
      has_feed_operators(global_block, *feed_targets, feed_holder_name),
384 385
      "Program in ExecutorPrepareContext should has feed_ops.");
  PADDLE_ENFORCE(
386
      has_fetch_operators(global_block, *fetch_targets, fetch_holder_name),
387 388
      "Program in the prepared context should has fetch_ops.");

389 390 391 392 393
  // map the data of feed_targets to feed_holder
  for (auto* op : global_block.AllOps()) {
    if (op->Type() == kFeedOpType) {
      std::string feed_target_name = op->Output("Out")[0];
      int idx = boost::get<int>(op->GetAttr("col"));
394 395
      SetFeedVariable(scope, *(*feed_targets)[feed_target_name],
                      feed_holder_name, idx);
396 397 398
    }
  }

W
Wu Yi 已提交
399
  RunPreparedContext(ctx, scope, create_local_scope, create_vars);
400 401 402 403 404 405

  // obtain the data of fetch_targets from fetch_holder
  for (auto* op : global_block.AllOps()) {
    if (op->Type() == kFetchOpType) {
      std::string fetch_target_name = op->Input("X")[0];
      int idx = boost::get<int>(op->GetAttr("col"));
406
      *(*fetch_targets)[fetch_target_name] =
407 408 409 410 411
          GetFetchVariable(*scope, fetch_holder_name, idx);
    }
  }
}

412 413 414 415 416 417 418 419 420 421 422
void Executor::EnableMKLDNN(const ProgramDesc& program) {
#ifdef PADDLE_WITH_MKLDNN
  VLOG(3) << "use_mkldnn=True";
  for (size_t bid = 0; bid < program.Size(); ++bid) {
    auto* block = const_cast<ProgramDesc&>(program).MutableBlock(bid);
    for (auto* op : block->AllOps()) {
      if (op->HasAttr("use_mkldnn")) {
        op->SetAttr("use_mkldnn", true);
      }
    }
  }
423 424 425
#else
  LOG(WARNING)
      << "'MKLDNN' is not supported, Please re-compile with WITH_MKLDNN option";
426 427 428
#endif
}

Q
qijun 已提交
429 430
}  // namespace framework
}  // namespace paddle