parallel_executor.cc 19.0 KB
Newer Older
Y
Yang Yang 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/* Copyright (c) 2016 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. */

#include "paddle/fluid/framework/parallel_executor.h"
Y
Yu Yang 已提交
16 17
#include "ThreadPool.h"
#include "executor.h"
Y
Yu Yang 已提交
18 19
#include "lod_tensor.h"
#include "op_registry.h"
Y
Yang Yang 已提交
20 21

namespace paddle {
Y
Yu Yang 已提交
22 23
namespace framework {

Y
Yu Yang 已提交
24 25 26 27 28 29
#ifdef PADDLE_WITH_CUDA

// FIXME: CHECK the return value of x;
#define NCCL_INVOKE(x) x
#endif

Y
Yu Yang 已提交
30 31
struct OpHandle;

Y
Yu Yang 已提交
32 33 34 35 36 37 38 39 40 41 42 43 44 45 46
struct VarHandleBase {
  virtual ~VarHandleBase() {}
  virtual std::string DebugString() const = 0;

  OpHandle *generated_op_;
  std::vector<OpHandle *> pending_ops_;
};

struct VarHandle : public VarHandleBase {
  std::string DebugString() const override {
    std::stringstream ss;
    ss << name_ << ":" << place_;
    return ss.str();
  }

Y
Yu Yang 已提交
47 48 49
  size_t version_;
  std::string name_;
  platform::Place place_;
Y
Yu Yang 已提交
50
};
Y
Yu Yang 已提交
51

Y
Yu Yang 已提交
52
struct DependencyVarHandle : public VarHandleBase {
Y
Yu Yang 已提交
53
  std::string DebugString() const override { return "Dependency Variable"; }
Y
Yu Yang 已提交
54 55 56
};

struct OpHandle {
Y
Yu Yang 已提交
57 58 59 60 61
  std::vector<VarHandleBase *> inputs_;
  std::vector<VarHandleBase *> outputs_;
  std::unordered_map<platform::Place, platform::DeviceContext *,
                     platform::PlaceHash>
      dev_ctx_;
Y
Yu Yang 已提交
62 63 64 65 66

  std::string DebugString() {
    std::stringstream ss;
    ss << "(";
    for (auto *var : inputs_) {
Y
Yu Yang 已提交
67
      ss << var->DebugString() << ", ";
Y
Yu Yang 已提交
68 69 70
    }
    ss << ") --> (";
    for (auto *var : outputs_) {
Y
Yu Yang 已提交
71
      ss << var->DebugString() << ", ";
Y
Yu Yang 已提交
72 73 74 75 76 77
    }
    ss << ")\n";
    return ss.str();
  }

  virtual ~OpHandle() {}
Y
Yu Yang 已提交
78

Y
Yu Yang 已提交
79
  virtual void Run() { PADDLE_THROW("Not implemented"); }
Y
Yu Yang 已提交
80
  virtual void Wait() {}
Y
Yu Yang 已提交
81 82 83 84
};

struct ComputationOpHandle : public OpHandle {
  std::unique_ptr<OperatorBase> op_;
Y
Yu Yang 已提交
85 86
  Scope *scope_;
  platform::Place place_;
Y
Yu Yang 已提交
87

Y
Yu Yang 已提交
88 89
  explicit ComputationOpHandle(const OpDesc &op_desc, Scope *scope,
                               platform::Place place)
Y
Yu Yang 已提交
90
      : op_(framework::OpRegistry::CreateOp(op_desc)),
Y
Yu Yang 已提交
91
        scope_(scope),
Y
Yu Yang 已提交
92 93 94 95
        place_(place) {}

  void Run() override {
    // Wait other op if necessary
Y
Yu Yang 已提交
96
    LOG(INFO) << "Run " << this << " " << DebugString();
Y
Yu Yang 已提交
97 98 99 100 101 102 103 104
    auto *cur_ctx = dev_ctx_[place_];
    for (auto *in : inputs_) {
      if (in->generated_op_ && in->generated_op_->dev_ctx_[place_] != cur_ctx) {
        in->generated_op_->Wait();
      }
    }

    op_->Run(*scope_, place_);
Y
Yu Yang 已提交
105
    LOG(INFO) << "Done " << this;
Y
Yu Yang 已提交
106
  }
Y
Yu Yang 已提交
107 108
};

Y
Yu Yang 已提交
109 110 111 112 113 114 115 116 117 118 119 120 121 122 123
struct ScaleLossGradOpHandle : public OpHandle {
  float coeff_;
  Scope *scope_;
  platform::Place place_;

  explicit ScaleLossGradOpHandle(size_t num_dev, Scope *scope,
                                 platform::Place place)
      : coeff_(static_cast<float>(1.0 / num_dev)),
        scope_(scope),
        place_(place) {}

  void Run() override {
    LOG(INFO) << "Run Scale Loss Grad";

    std::string var_name = static_cast<VarHandle *>(this->outputs_[0])->name_;
Y
Yu Yang 已提交
124

Y
Yu Yang 已提交
125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147
    float *tmp = scope_->FindVar(var_name)
                     ->GetMutable<framework::LoDTensor>()
                     ->mutable_data<float>(make_ddim({1}), place_);

    if (platform::is_cpu_place(place_)) {
      *tmp = coeff_;
    } else {
      memory::Copy(
          boost::get<platform::CUDAPlace>(place_), tmp, platform::CPUPlace(),
          &coeff_, sizeof(float),
          static_cast<platform::CUDADeviceContext *>(this->dev_ctx_[place_])
              ->stream());
    }
  }
};

struct NCCLAllReduceOpHandle : public OpHandle {
  void Run() override {
    if (this->inputs_.size() == 1) {
      return;  // No need to all reduce when GPU count = 1;
    }
  }
};
Y
Yu Yang 已提交
148 149 150

class ParallelExecutorPrivate {
 public:
Y
Yu Yang 已提交
151 152 153
  explicit ParallelExecutorPrivate(size_t num_threads = 12)
      : pool_(num_threads) {}

Y
Yu Yang 已提交
154 155
  std::unordered_map<platform::Place, Scope *, platform::PlaceHash>
      local_scopes_;
Y
Yu Yang 已提交
156

Y
Stash  
Yu Yang 已提交
157 158
  std::vector<platform::Place> places_;

Y
Yu Yang 已提交
159 160 161 162 163 164 165 166 167 168 169 170 171 172 173
#ifdef PADDLE_WITH_CUDA
  struct NCCLContext {
    std::unique_ptr<platform::CUDADeviceContext> ctx_;
    ncclComm_t comm;

    explicit NCCLContext(int dev_id) {
      ctx_.reset(new platform::CUDADeviceContext(platform::CUDAPlace(dev_id)));
    }

    cudaStream_t stream() const { return ctx_->stream(); }

    int device_id() const {
      return boost::get<platform::CUDAPlace>(ctx_->GetPlace()).device;
    }

Y
Update  
Yu Yang 已提交
174 175
    static void InitNCCLContext(std::unordered_map<int, NCCLContext> &contexts,
                                const std::vector<platform::Place> &places) {
Y
Yu Yang 已提交
176 177 178 179 180
      std::vector<ncclComm_t> comms;
      std::vector<int> devs;
      comms.resize(contexts.size());
      devs.reserve(contexts.size());

Y
Update  
Yu Yang 已提交
181 182
      for (auto &p : places) {
        devs.push_back(boost::get<platform::CUDAPlace>(p).device);
Y
Yu Yang 已提交
183 184 185 186 187 188
      }

      NCCL_INVOKE(platform::dynload::ncclCommInitAll(
          &comms[0], static_cast<int>(contexts.size()), &devs[0]));

      int i = 0;
Y
Update  
Yu Yang 已提交
189 190
      for (auto &dev_id : devs) {
        contexts.at(dev_id).comm = comms[i++];
Y
Yu Yang 已提交
191 192 193 194
      }
    }
  };

Y
Update  
Yu Yang 已提交
195
  std::unordered_map<int, NCCLContext> communication_streams_;
Y
Yu Yang 已提交
196 197 198 199 200 201 202 203

  NCCLContext &GetNCCLCtx(platform::Place p) {
    int dev_id = boost::get<platform::CUDAPlace>(p).device;
    return communication_streams_.at(dev_id);
  }

#endif

Y
Yu Yang 已提交
204 205 206 207 208 209 210 211 212 213 214 215 216
  platform::DeviceContext *CommunicationDevCtx(const platform::Place &place) {
    if (platform::is_cpu_place(place) || local_scopes_.size() == 1) {
      return const_cast<platform::DeviceContext *>(
          platform::DeviceContextPool::Instance().Get(place));
    } else {
#ifdef PADDLE_WITH_CUDA
      return GetNCCLCtx(place).ctx_.get();
#else
      PADDLE_THROW("Not compiled with CUDA")
#endif
    }
  }

Y
Yu Yang 已提交
217 218 219 220 221 222
  platform::Place main_place_;

  std::unordered_map<platform::Place,
                     std::unordered_map<std::string, std::map<int, VarHandle>>,
                     platform::PlaceHash>
      vars_;
Y
Yu Yang 已提交
223 224
  std::unordered_set<std::unique_ptr<VarHandleBase>> dep_vars_;

Y
Yu Yang 已提交
225
  std::vector<std::unique_ptr<OpHandle>> ops_;
Y
Yu Yang 已提交
226

Y
Yu Yang 已提交
227
  // Use a simpler thread pool, might be faster.
Y
Yu Yang 已提交
228
  ThreadPool pool_;
Y
Yu Yang 已提交
229 230

  std::unique_ptr<platform::EnforceNotMet> exception_;
Y
Yu Yang 已提交
231 232 233 234
};

// TODO(yy): Move this function somewhere
ncclDataType_t ToNCCLDataType(std::type_index type) {
Y
Stash  
Yu Yang 已提交
235 236 237 238 239 240 241 242 243
  if (type == typeid(float)) {  // NOLINT
    return ncclFloat;
  } else if (type == typeid(double)) {  // NOLINT
    return ncclDouble;
  } else if (type == typeid(int)) {  // NOLINT
    return ncclInt;
  } else {
    PADDLE_THROW("Not supported");
  }
Y
Yu Yang 已提交
244 245 246 247 248 249 250 251
}

ParallelExecutor::ParallelExecutor(
    const std::vector<platform::Place> &places,
    const std::unordered_set<std::string> &params,
    const ProgramDesc &startup_program, const ProgramDesc &main_program,
    const std::string &loss_var_name, Scope *scope)
    : member_(new ParallelExecutorPrivate()) {
Y
Stash  
Yu Yang 已提交
252 253
  member_->places_ = places;

Y
Yu Yang 已提交
254 255 256 257 258 259 260 261 262 263
  // Step 1. RunStartupProgram and Bcast the params to devs.
  Executor exe(places[0]);
  exe.Run(startup_program, scope, 0);
  // Create local scopes
  for (auto &place : places) {
    member_->local_scopes_[place] = &scope->NewScope();
  }
  member_->main_place_ = places[0];

  // Bcast Parameters to all GPUs
Y
Yu Yang 已提交
264 265 266 267
  if (platform::is_gpu_place(member_->main_place_) &&
      member_->local_scopes_.size() != 1) {  // Is CUDA
    BuildNCCLCommunicator();
    BCastParamsToGPUs(startup_program);
Y
Yu Yang 已提交
268 269 270 271 272 273
  }
  // Startup Program has been run. All local scopes has correct parameters.

  // Step 2. Convert main_program to SSA form and dependency graph. Also, insert
  // ncclOp
  ConstructDependencyGraph(params, main_program, loss_var_name);
Y
Yu Yang 已提交
274 275 276 277 278 279 280 281 282 283 284 285 286

  // Step 3. Create vars in each scope;
  for (auto &pair : member_->local_scopes_) {
    auto *scope = pair.second;

    for (auto *var : main_program.Block(0).AllVars()) {
      if (scope->FindVar(var->Name()) != nullptr) {
        continue;
      }

      InitializeVariable(scope->Var(var->Name()), var->GetType());
    }
  }
Y
Yu Yang 已提交
287 288 289 290 291
}

void ParallelExecutor::ConstructDependencyGraph(
    const std::unordered_set<std::string> &params,
    const ProgramDesc &main_program, const std::string &loss_var_name) const {
Y
Yu Yang 已提交
292
  std::unordered_set<std::string> grads;
Y
Yu Yang 已提交
293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309
  for (auto &each_param : params) {
    grads.insert(each_param + "@GRAD");
  }

  bool is_forwarding = true;
  for (auto *op : main_program.Block(0).AllOps()) {
    bool change_forward = false;

    if (!is_forwarding) {
      // FIXME(yy): Do not hard code like this
      if (op->OutputArgumentNames().size() == 1 &&
          op->OutputArgumentNames()[0] == loss_var_name + "@GRAD") {
        continue;  // Drop fill 1. for backward coeff;
      }
    }

    for (auto &pair : member_->local_scopes_) {
Y
Yu Yang 已提交
310 311
      member_->ops_.emplace_back(
          new ComputationOpHandle(*op, pair.second, pair.first));
Y
Yu Yang 已提交
312
      auto *op_handle = member_->ops_.back().get();
Y
Yu Yang 已提交
313 314
      op_handle->dev_ctx_[pair.first] = const_cast<platform::DeviceContext *>(
          platform::DeviceContextPool::Instance().Get(pair.first));
Y
Yu Yang 已提交
315 316 317 318 319 320 321

      auto var_names = op->InputArgumentNames();

      for (auto &each_var_name : var_names) {
        auto &place = pair.first;
        VarHandle *var = GetVarHandle(each_var_name, place);
        op_handle->inputs_.emplace_back(var);
Y
Yu Yang 已提交
322
        var->pending_ops_.emplace_back(op_handle);
Y
Yu Yang 已提交
323 324 325 326 327 328 329 330 331 332 333
      }
      var_names = op->OutputArgumentNames();

      for (auto &each_var_name : var_names) {
        auto &place = pair.first;
        GenerateVar(op_handle, each_var_name, place);
      }

      if (is_forwarding) {
        if (var_names.size() == 1 && var_names[0] == loss_var_name) {
          // Insert ScaleCost OpHandle
Y
Yu Yang 已提交
334 335
          member_->ops_.emplace_back(new ScaleLossGradOpHandle(
              this->member_->local_scopes_.size(), pair.second, pair.first));
Y
Yu Yang 已提交
336
          op_handle = member_->ops_.back().get();
Y
Yu Yang 已提交
337 338 339 340

          op_handle->dev_ctx_[pair.first] =
              member_->CommunicationDevCtx(pair.first);

Y
Yu Yang 已提交
341
          auto &place = pair.first;
Y
Yu Yang 已提交
342 343 344 345 346 347
          // FIXME: Currently ScaleLossGradOp only use device_count as scale
          // factor. So it does not depend on any other operators.
          // VarHandle *loss = GetVarHandle(loss_var_name, place);
          // loss->pending_ops_.emplace_back(op_handle);
          // op_handle->inputs_.emplace_back(loss);

Y
Yu Yang 已提交
348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375
          GenerateVar(op_handle, loss_var_name + "@GRAD", place);
          change_forward = true;
          LOG(INFO) << "Scale Loss " << op_handle->DebugString();
        }
      }
    }

    if (change_forward) {
      is_forwarding = false;
    }

    if (!is_forwarding) {
      auto var_names = op->OutputArgumentNames();
      for (auto &og : var_names) {
        if (grads.count(og) != 0) {  // is param grad
          // Insert NCCL AllReduce Op
          member_->ops_.emplace_back(new NCCLAllReduceOpHandle());
          auto *op_handle = member_->ops_.back().get();

          for (auto &pair : member_->local_scopes_) {
            auto &place = pair.first;
            auto &vars = member_->vars_[place][og];

            if (vars.empty()) {  // This device has no data. continue.
              continue;
            }
            auto *prev_grad = &vars[vars.size() - 1];
            op_handle->inputs_.emplace_back(prev_grad);
Y
Yu Yang 已提交
376
            prev_grad->pending_ops_.emplace_back(op_handle);
Y
Yu Yang 已提交
377 378 379 380 381 382
            auto &var = vars[vars.size()];
            var.place_ = place;
            var.generated_op_ = op_handle;
            var.name_ = og;
            var.version_ = vars.size() - 1;
            op_handle->outputs_.emplace_back(&var);
Y
Yu Yang 已提交
383 384 385 386 387

            for (auto &pair : member_->local_scopes_) {
              op_handle->dev_ctx_[pair.first] =
                  member_->CommunicationDevCtx(pair.first);
            }
Y
Yu Yang 已提交
388 389 390 391 392
          }
        }
      }
    }
  }
Y
Yu Yang 已提交
393

Y
Yu Yang 已提交
394 395 396
  /*
    Dependency graph has been constructed. However, there are still data
    harzaeds need to be handled.
Y
Yu Yang 已提交
397
   */
Y
Yu Yang 已提交
398 399
  PolishGraphToSupportDataHarzaeds();
}
Y
Yu Yang 已提交
400

Y
Yu Yang 已提交
401 402 403 404 405 406 407 408
/**
 * We only handle write after read(WAR), since it should not have a write
 * after write in program. If there are write after write operators, we need
 * prune them.
 *
 * https://en.wikipedia.org/wiki/Hazard_(computer_architecture)#Write_after_read_(WAR)
 */
void ParallelExecutor::PolishGraphToSupportDataHarzaeds() const {
Y
Yu Yang 已提交
409 410 411 412 413 414 415 416 417 418 419
  for (auto &place_pair : member_->vars_) {
    for (auto &name_pair : place_pair.second) {
      if (name_pair.second.size() <= 1) {
        return;
      }
      auto it_new = name_pair.second.rbegin();
      auto it_old = name_pair.second.rbegin();
      ++it_old;
      for (; it_old != name_pair.second.rend(); it_new = it_old, ++it_old) {
        auto *write_op = it_new->second.generated_op_;
        auto &read_ops = it_old->second.pending_ops_;
Y
Yu Yang 已提交
420 421 422 423 424 425 426 427 428
        auto *ex_write_op = it_old->second.generated_op_;

        if (ex_write_op == nullptr) {  // Nobody write this var.
          continue;
        }

        LOG(INFO) << "Link " << it_new->second.DebugString() << " From "
                  << it_old->second.version_ << " To "
                  << it_new->second.version_;
Y
Yu Yang 已提交
429 430 431

        for (auto *read_op : read_ops) {
          // Manually add a dependency var from read_op to write_op;
Y
Yu Yang 已提交
432 433 434 435
          if (read_op == write_op) {
            // Read Write is the same op.
            continue;
          }
Y
Yu Yang 已提交
436 437

          auto *dep_var = new DependencyVarHandle();
Y
Yu Yang 已提交
438

Y
Yu Yang 已提交
439 440 441 442 443 444 445 446 447 448
          dep_var->generated_op_ = read_op;
          read_op->outputs_.emplace_back(dep_var);

          dep_var->pending_ops_.emplace_back(write_op);
          write_op->inputs_.emplace_back(dep_var);
          member_->dep_vars_.emplace(dep_var);
        }
      }
    }
  }
Y
Yu Yang 已提交
449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483
}

void ParallelExecutor::GenerateVar(OpHandle *op_handle,
                                   const std::string &each_var_name,
                                   const platform::Place &place) const {
  auto &vars = member_->vars_[place][each_var_name];
  size_t version = vars.size();
  auto &var = vars[version];
  var.version_ = version;
  var.generated_op_ = op_handle;
  var.name_ = each_var_name;
  var.place_ = place;
  op_handle->outputs_.emplace_back(&var);
}

VarHandle *ParallelExecutor::GetVarHandle(const std::string &each_var_name,
                                          const platform::Place &place) const {
  auto &var_holders = member_->vars_[place];
  auto &var_holder = var_holders[each_var_name];
  VarHandle *var = nullptr;
  if (var_holder.empty()) {
    auto &init_var = var_holder[0];
    init_var.place_ = place;
    init_var.name_ = each_var_name;
    init_var.generated_op_ = nullptr;
    init_var.version_ = 0;
    var = &init_var;
  } else {
    var = &var_holder.rbegin()->second;
  }
  return var;
}

void ParallelExecutor::BCastParamsToGPUs(
    const ProgramDesc &startup_program) const {
Y
Yu Yang 已提交
484
#ifdef PADDLE_WITH_CUDA
Y
Yu Yang 已提交
485
  auto *main_scope = member_->local_scopes_[member_->main_place_];
Y
Yu Yang 已提交
486

Y
Yu Yang 已提交
487 488 489 490 491 492 493 494
  for (auto *var_desc : startup_program.Block(0).AllVars()) {
    if (var_desc->GetType() == proto::VarType::LOD_TENSOR) {
      auto &main_tensor =
          main_scope->FindVar(var_desc->Name())->Get<LoDTensor>();
      ncclDataType_t data_type = ToNCCLDataType(main_tensor.type());
      auto &dims = main_tensor.dims();
      size_t numel = main_tensor.numel();

Y
Stash  
Yu Yang 已提交
495
      platform::dynload::ncclGroupStart();
Y
Yu Yang 已提交
496

Y
Update  
Yu Yang 已提交
497 498 499 500 501 502 503 504 505 506 507 508
      for (size_t i = 0; i < member_->places_.size(); ++i) {
        auto place = member_->places_[i];
        void *buffer;
        if (i == 0) {
          buffer = const_cast<void *>(main_tensor.data<void>());
        } else {
          auto local_scope = member_->local_scopes_[place];
          auto *t = local_scope->Var(var_desc->Name())->GetMutable<LoDTensor>();
          t->Resize(dims);
          buffer = t->mutable_data(place, main_tensor.type());
        }

Y
Stash  
Yu Yang 已提交
509
        auto &nccl_ctx = member_->GetNCCLCtx(place);
Y
Update  
Yu Yang 已提交
510
        platform::dynload::ncclBcast(buffer, numel, data_type, 0, nccl_ctx.comm,
Y
Stash  
Yu Yang 已提交
511
                                     nccl_ctx.stream());
Y
Yu Yang 已提交
512
      }
Y
Stash  
Yu Yang 已提交
513 514 515
      platform::dynload::ncclGroupEnd();
    }
  }
Y
Yu Yang 已提交
516

Y
Yu Yang 已提交
517
  // Debug code, bias should be 1.0f.
Y
Stash  
Yu Yang 已提交
518 519
  for (auto &pair : member_->local_scopes_) {
    member_->GetNCCLCtx(pair.first).ctx_->Wait();
Y
Yu Yang 已提交
520

Y
Stash  
Yu Yang 已提交
521
    auto &b = pair.second->FindVar("fc_0.b_0")->Get<framework::LoDTensor>();
Y
Stash  
Yu Yang 已提交
522 523 524 525
    framework::LoDTensor cpu;
    framework::TensorCopy(b, platform::CPUPlace(), &cpu);
    platform::DeviceContextPool::Instance().Get(b.place())->Wait();
    LOG(INFO) << *cpu.data<float>();
Y
Yu Yang 已提交
526
  }
Y
Stash  
Yu Yang 已提交
527

Y
Yu Yang 已提交
528 529 530 531
#else
  PADDLE_THROW("Not compiled with CUDA");
#endif
}
Y
Yu Yang 已提交
532

Y
Yu Yang 已提交
533 534 535 536 537
void ParallelExecutor::BuildNCCLCommunicator() const {
#ifdef PADDLE_WITH_CUDA
  for (auto &place_pair : member_->local_scopes_) {
    auto place = place_pair.first;
    int dev_id = boost::get<platform::CUDAPlace>(place).device;
Y
Yu Yang 已提交
538

Y
Yu Yang 已提交
539 540
    member_->communication_streams_.emplace(
        dev_id, ParallelExecutorPrivate::NCCLContext(dev_id));
Y
Yu Yang 已提交
541
  }
Y
Yu Yang 已提交
542 543

  ParallelExecutorPrivate::NCCLContext::InitNCCLContext(
Y
Update  
Yu Yang 已提交
544
      member_->communication_streams_, member_->places_);
Y
Yu Yang 已提交
545
#endif
Y
Yu Yang 已提交
546 547 548 549 550
}

std::vector<LoDTensor> ParallelExecutor::Run(
    const std::vector<std::string> &fetch_tensors) {
  // Version --> VarHandle
Y
Yu Yang 已提交
551
  member_->exception_.reset();
Y
Yu Yang 已提交
552
  std::unordered_map<VarHandleBase *, bool> pending_vars;
Y
Yu Yang 已提交
553 554 555 556 557
  std::unordered_map<OpHandle *, size_t> pending_ops;

  for (auto &place_pair : member_->vars_) {
    for (auto &name_pair : place_pair.second) {
      for (auto &version_pair : name_pair.second) {
Y
Yu Yang 已提交
558 559
        pending_vars[&version_pair.second] =
            version_pair.second.generated_op_ == nullptr;
Y
Yu Yang 已提交
560 561 562 563
      }
    }
  }

Y
Yu Yang 已提交
564 565 566 567
  for (auto &var : member_->dep_vars_) {
    pending_vars[var.get()] = var->generated_op_ == nullptr;
  }

Y
Yu Yang 已提交
568 569
  std::vector<OpHandle *> to_run;

Y
Yu Yang 已提交
570
  for (auto &op : member_->ops_) {
Y
Yu Yang 已提交
571 572 573 574 575 576 577 578 579
    if (op->inputs_.empty()) {  // Special case, Op has no input.
      to_run.emplace_back(op.get());
    } else {
      pending_ops.insert({op.get(), op->inputs_.size()});
    }
  }

  for (auto *op : to_run) {
    RunOp(pending_vars, op);
Y
Yu Yang 已提交
580 581
  }

Y
Yu Yang 已提交
582
  while (!pending_ops.empty()) {
Y
Yu Yang 已提交
583
    VarHandleBase *ready_var = nullptr;
Y
Yu Yang 已提交
584 585 586
    for (auto &pair : pending_vars) {
      if (pair.second) {
        ready_var = pair.first;
Y
Yu Yang 已提交
587 588
      }
    }
Y
Yu Yang 已提交
589 590

    if (ready_var == nullptr) {
Y
Yu Yang 已提交
591 592 593 594 595 596 597
      // FIXME use conditional var instead of busy wait.

      if (member_->exception_) {
        throw * member_->exception_;
      }

      std::this_thread::yield();
Y
Yu Yang 已提交
598
      continue;
Y
Yu Yang 已提交
599 600
    }

Y
Yu Yang 已提交
601 602
    pending_vars.erase(ready_var);

Y
Yu Yang 已提交
603
    to_run.clear();
Y
Yu Yang 已提交
604 605 606 607 608 609

    for (auto *op : ready_var->pending_ops_) {
      auto &deps = pending_ops[op];
      --deps;
      if (deps == 0) {
        to_run.emplace_back(op);
Y
Yu Yang 已提交
610 611 612 613 614
      }
    }

    for (auto *op : to_run) {
      pending_ops.erase(op);
Y
Yu Yang 已提交
615
      RunOp(pending_vars, op);
Y
Yu Yang 已提交
616 617 618 619
    }
  }
  return std::vector<LoDTensor>();
}
Y
Yu Yang 已提交
620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644

void ParallelExecutor::RunOp(
    std::unordered_map<VarHandleBase *, bool> &pending_vars,
    OpHandle *op) const {
  std::vector<bool *> ready_buffer;
  for (auto *var : op->outputs_) {
    ready_buffer.emplace_back(&pending_vars[var]);
  }

  auto op_run = [ready_buffer, op, this] {
    try {
      // TODO(yy) Check Previous Op has same dev ctx.
      op->Run();
      for (auto *ready : ready_buffer) {
        *ready = true;
      }
    } catch (platform::EnforceNotMet ex) {
      member_->exception_.reset(new platform::EnforceNotMet(ex));
    } catch (...) {
      LOG(FATAL) << "Unknown exception catched";
    }
  };

  member_->pool_.enqueue(op_run);
}
Y
Yu Yang 已提交
645
}  // namespace framework
Y
Yang Yang 已提交
646
}  // namespace paddle