fleet.cc 31.2 KB
Newer Older
T
tangwei12 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14
/* Copyright (c) 2020 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 16
#include "paddle/fluid/distributed/ps/wrapper/fleet.h"

17 18
#include <google/protobuf/text_format.h>

19 20
#include "paddle/fluid/distributed/ps/service/communicator/communicator.h"
#include "paddle/fluid/distributed/ps/table/table.h"
D
danleifeng 已提交
21 22 23 24
#include "paddle/fluid/distributed/ps/wrapper/fleet.h"
#if defined PADDLE_WITH_HETERPS && defined PADDLE_WITH_PSCORE
#include "paddle/fluid/framework/fleet/ps_gpu_wrapper.h"
#endif
T
tangwei12 已提交
25 26 27 28 29 30 31 32 33 34 35 36 37 38

namespace paddle {
namespace distributed {

using framework::LoDTensor;
using framework::ProgramDesc;
using framework::VarDesc;
using framework::Variable;

const uint32_t MAX_FEASIGN_NUM = 1024 * 100 * 100;
std::shared_ptr<FleetWrapper> FleetWrapper::s_instance_ = NULL;
bool FleetWrapper::is_initialized_ = false;

std::shared_ptr<paddle::distributed::PSCore> FleetWrapper::pserver_ptr_ = NULL;
39 40 41 42 43 44 45 46 47 48 49 50 51 52
std::shared_ptr<paddle::distributed::PSClient> FleetWrapper::worker_ptr_ = NULL;

int FleetWrapper::RegisterHeterCallback(HeterCallBackFunc handler) {
  VLOG(0) << "RegisterHeterCallback support later";
  return 0;
}

int32_t FleetWrapper::CopyTable(const uint64_t src_table_id,
                                const uint64_t dest_table_id) {
  VLOG(0) << "CopyTable support later";
  return 0;
}

int32_t FleetWrapper::CopyTableByFeasign(
53 54
    const uint64_t src_table_id,
    const uint64_t dest_table_id,
55 56 57 58
    const std::vector<uint64_t>& feasign_list) {
  VLOG(0) << "CopyTableByFeasign support later";
  return 0;
}
T
tangwei12 已提交
59 60 61 62 63 64 65 66 67 68 69 70 71 72

void FleetWrapper::SetClient2ClientConfig(int request_timeout_ms,
                                          int connect_timeout_ms,
                                          int max_retry) {
  client2client_request_timeout_ms_ = request_timeout_ms;
  client2client_connect_timeout_ms_ = connect_timeout_ms;
  client2client_max_retry_ = max_retry;
}

void FleetWrapper::LoadSparseOnServer(const std::string& path,
                                      const std::string& meta,
                                      uint32_t table_id) {
  VLOG(3) << "load sparse table " << table_id << " with " << path << " meta "
          << meta;
Z
zhaocaibei123 已提交
73
  pserver_ptr_->_server_ptr->GetTable(table_id)->Load(path, meta);
T
tangwei12 已提交
74 75
}

76 77
void FleetWrapper::InitServer(
    const std::string& dist_desc,
78 79 80
    const std::vector<std::string>& host_sign_list,
    int index,
    int trainers,
81
    const std::vector<framework::ProgramDesc>& server_sub_program) {
T
tangwei12 已提交
82 83 84 85
  if (!is_initialized_) {
    VLOG(3) << "Going to init server";
    pserver_ptr_ = std::shared_ptr<paddle::distributed::PSCore>(
        new paddle::distributed::PSCore());
86 87 88 89 90 91
    pserver_ptr_->InitServer(dist_desc,
                             &host_sign_list,
                             host_sign_list.size(),
                             index,
                             trainers,
                             server_sub_program);
T
tangwei12 已提交
92 93 94 95 96 97
    is_initialized_ = true;
  } else {
    VLOG(3) << "Server can be initialized only once";
  }
}

98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116
void FleetWrapper::InitGFlag(const std::string& gflags) {
  VLOG(3) << "Init With Gflags:" << gflags;
  std::vector<std::string> flags = paddle::string::split_string(gflags);
  if (flags.size() < 1) {
    flags.push_back("-max_body_size=314217728");
    flags.push_back("-bthread_concurrency=40");
    flags.push_back("-socket_max_unwritten_bytes=2048000000");
    flags.push_back("-max_connection_pool_size=1950");
  }
  auto it = flags.begin();
  flags.insert(it, "exe default");
  char* flags_ptr[flags.size()];
  for (size_t i = 0; i < flags.size(); ++i) {
    flags_ptr[i] = (char*)(flags[i].c_str());  // NOLINT
  }
  int params_cnt = flags.size();
  char** params_ptr = &(flags_ptr[0]);
  ::GFLAGS_NAMESPACE::ParseCommandLineFlags(&params_cnt, &params_ptr, true);
}
T
tangwei12 已提交
117

118 119 120 121 122 123 124 125 126 127 128 129 130 131
void FleetWrapper::InitWorker(const std::string& dist_desc,
                              const std::vector<std::string>& host_sign_list,
                              int index) {
  if (!is_initialized_) {
    // not used, just for psclient's init
    // TODO(zhaocaibei123): remove this later
    std::map<uint64_t, std::vector<paddle::distributed::Region>>
        dense_pull_regions;

    if (worker_ptr_.get() == nullptr) {
      paddle::distributed::PSParameter ps_param;
      google::protobuf::TextFormat::ParseFromString(dist_desc, &ps_param);
      InitGFlag(ps_param.init_gflags());
      int servers = host_sign_list.size();
Z
zhaocaibei123 已提交
132
      ps_env_.SetPsServers(&host_sign_list, servers);
133
      worker_ptr_ = std::shared_ptr<paddle::distributed::PSClient>(
Z
zhaocaibei123 已提交
134 135
          paddle::distributed::PSClientFactory::Create(ps_param));
      worker_ptr_->Configure(ps_param, dense_pull_regions, ps_env_, index);
D
danleifeng 已提交
136 137 138 139 140 141 142
#if defined PADDLE_WITH_HETERPS && defined PADDLE_WITH_PSCORE
      VLOG(3) << "FleetWrapper::InitWorker InitializeGPUServer";
      auto* accessor = worker_ptr_->GetTableAccessor(0);
      auto ps_gpu_wrapper = paddle::framework::PSGPUWrapper::GetInstance();
      ps_gpu_wrapper->InitializeGPUServer(ps_param);
      ps_gpu_wrapper->SetTableAccessor(accessor);
#endif
143
    }
T
tangwei12 已提交
144
  } else {
145
    VLOG(3) << "Client can be initialized only once";
T
tangwei12 已提交
146 147 148 149 150
  }
}

void FleetWrapper::StopServer() {
  VLOG(3) << "Going to stop server";
Z
zhaocaibei123 已提交
151
  auto status = worker_ptr_->StopServer();
T
tangwei12 已提交
152 153 154 155 156
  status.wait();
}

void FleetWrapper::FinalizeWorker() {
  VLOG(3) << "Going to finalize worker";
Z
zhaocaibei123 已提交
157
  worker_ptr_->FinalizeWorker();
T
tangwei12 已提交
158 159 160 161 162 163 164 165 166 167
}

void FleetWrapper::BarrierWithTable(uint32_t barrier_type) {
  VLOG(3) << "Going to Barrier worker";
  auto* communicator = Communicator::GetInstance();
  communicator->BarrierWithTable(barrier_type);
}

uint64_t FleetWrapper::RunServer(const std::string& ip, uint32_t port) {
  VLOG(3) << "Going to run server with ip " << ip << " port " << port;
Z
zhaocaibei123 已提交
168
  auto ret = pserver_ptr_->RunServer(ip, port);
T
tangwei12 已提交
169 170 171 172 173
  return ret;
}

std::vector<uint64_t> FleetWrapper::GetClientsInfo() {
  VLOG(3) << "Going to get client info";
Z
zhaocaibei123 已提交
174
  std::vector<uint64_t> res = ps_env_.GetClientInfo();
175 176 177
  for (auto rr : res) {
    VLOG(2) << "FleetWrapper::GetClientInfo " << rr;
  }
Z
zhaocaibei123 已提交
178
  return res;
T
tangwei12 已提交
179 180
}

181 182
int FleetWrapper::SetClients(std::vector<uint64_t>& host_sign_list) {
  int node = host_sign_list.size();
Z
zhaocaibei123 已提交
183
  return ps_env_.SetPsClients(host_sign_list.data(), node);
184 185
}

T
tangwei12 已提交
186
void FleetWrapper::CreateClient2ClientConnection() {
Z
zhaocaibei123 已提交
187
  VLOG(1) << "Going to create client2client connection";
Z
zhaocaibei123 已提交
188 189 190
  worker_ptr_->CreateClient2ClientConnection(client2client_request_timeout_ms_,
                                             client2client_connect_timeout_ms_,
                                             client2client_max_retry_);
T
tangwei12 已提交
191 192
}

193
std::future<int32_t> FleetWrapper::PullSparseVarsAsync(
194 195 196 197 198 199
    const Scope& scope,
    const uint64_t table_id,
    const std::vector<std::string>& var_names,
    std::vector<uint64_t>* fea_keys,
    std::vector<std::vector<float>>* fea_values,
    int fea_value_dim) {
200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228
  fea_keys->clear();
  fea_keys->resize(0);
  fea_keys->reserve(MAX_FEASIGN_NUM);
  for (auto name : var_names) {
    Variable* var = scope.FindVar(name);
    if (var == nullptr) {
      continue;
    }
    LoDTensor* tensor = var->GetMutable<LoDTensor>();
    CHECK(tensor != nullptr) << "tensor of var " << name << " is null";
    int64_t* ids = tensor->data<int64_t>();
    size_t len = tensor->numel();
    for (auto i = 0u; i < len; ++i) {
      if (ids[i] == 0u) {
        continue;
      }
      fea_keys->push_back(static_cast<uint64_t>(ids[i]));
    }
  }
  fea_values->resize(fea_keys->size() + 1);
  for (auto& t : *fea_values) {
    t.resize(fea_value_dim);
  }
  std::vector<float*> pull_result_ptr;
  for (auto& t : *fea_values) {
    pull_result_ptr.push_back(t.data());
  }

  bool training = true;
229 230
  return pserver_ptr_->_worker_ptr->PullSparse(pull_result_ptr.data(),
                                               table_id,
Z
zhaocaibei123 已提交
231
                                               fea_keys->data(),
232 233
                                               fea_keys->size(),
                                               training);
234 235
}

T
tangwei12 已提交
236
void FleetWrapper::PullSparseVarsSync(
237 238 239 240 241 242
    const Scope& scope,
    const uint64_t table_id,
    const std::vector<std::string>& var_names,
    std::vector<uint64_t>* fea_keys,
    std::vector<std::vector<float>>* fea_values,
    int fea_value_dim,
T
tangwei12 已提交
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
    const std::vector<std::string>& var_emb_names) {
  std::vector<std::future<int32_t>> pull_sparse_status;
  pull_sparse_status.resize(0);
  fea_keys->clear();
  fea_keys->resize(0);
  fea_keys->reserve(MAX_FEASIGN_NUM);
  for (size_t var_index = 0; var_index < var_names.size(); ++var_index) {
    const std::string& name = var_names[var_index];
    Variable* var = scope.FindVar(name);
    if (var == nullptr) {
      continue;
    }
    LoDTensor* tensor = var->GetMutable<LoDTensor>();
    CHECK(tensor != nullptr) << "tensor of var " << name << " is null";
    int64_t* ids = tensor->data<int64_t>();
    size_t len = tensor->numel();

    // skip slots which do not have embedding
    const std::string& emb_name = var_emb_names[var_index];
    Variable* emb_var = scope.FindVar(emb_name);
    if (emb_var == nullptr) {
      continue;
    }

    for (auto i = 0u; i < len; ++i) {
      if (ids[i] == 0u) {
        continue;
      }
      fea_keys->push_back(static_cast<uint64_t>(ids[i]));
    }
  }
  fea_values->resize(fea_keys->size() + 1);
  for (auto& t : *fea_values) {
    t.resize(fea_value_dim);
  }
  std::vector<float*> pull_result_ptr;
  for (auto& t : *fea_values) {
    pull_result_ptr.push_back(t.data());
  }
282
  bool training = true;
283 284 285 286 287
  auto status = pserver_ptr_->_worker_ptr->PullSparse(pull_result_ptr.data(),
                                                      table_id,
                                                      fea_keys->data(),
                                                      fea_keys->size(),
                                                      training);
T
tangwei12 已提交
288 289 290 291 292 293 294 295 296 297 298 299
  pull_sparse_status.push_back(std::move(status));
  for (auto& t : pull_sparse_status) {
    t.wait();
    auto status = t.get();
    if (status != 0) {
      LOG(ERROR) << "fleet pull sparse failed, status[" << status << "]";
      sleep(sleep_seconds_before_fail_exit_);
      exit(-1);
    }
  }
}

300 301 302
// is_training is true means training, false means inference, the behavior is
// different on pserver

303 304
void FleetWrapper::PullSparseToTensorSync(const uint64_t table_id,
                                          int fea_dim,
T
tangwei12 已提交
305 306
                                          uint64_t padding_id,
                                          platform::Place place,
307
                                          bool is_training,
T
tangwei12 已提交
308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335
                                          std::vector<const LoDTensor*>* inputs,
                                          std::vector<LoDTensor*>* outputs) {
  std::vector<uint64_t> fea_keys;
  std::vector<float*> pull_result_ptr;
  fea_keys.reserve(MAX_FEASIGN_NUM / 100);
  pull_result_ptr.reserve(MAX_FEASIGN_NUM / 100);
  std::vector<float> init_value(fea_dim, 0);
  framework::LoDTensor* output = nullptr;
  float* output_data = nullptr;
  size_t output_index = -1;
  size_t output_len = 0;
  for (size_t index = 0; index < inputs->size(); ++index) {
    const framework::LoDTensor* tensor = inputs->at(index);
    const int64_t* ids = tensor->data<int64_t>();
    size_t len = tensor->numel();
    for (size_t i = 0; i < len; ++i, output_len += fea_dim) {
      if (!output || output_len == size_t(output->numel())) {
        ++output_index;
        CHECK(output_index < outputs->size());  // NOLINT
        output = outputs->at(output_index);
        output->set_lod(tensor->lod());
        output_data = output->mutable_data<float>(place);
        output_len = 0;
        CHECK(output->numel() % fea_dim == 0);  // NOLINT
        CHECK(output_data != nullptr);          // NOLINT
      }
      uint64_t real_id = static_cast<uint64_t>(ids[i]);
      if (real_id == padding_id) {
336 337
        memcpy(output_data + output_len,
               init_value.data(),
T
tangwei12 已提交
338 339 340 341 342 343 344
               sizeof(float) * fea_dim);
        continue;
      }
      fea_keys.push_back(real_id);
      pull_result_ptr.push_back(output_data + output_len);
    }
  }
Z
zhaocaibei123 已提交
345

346 347 348 349 350
  auto status = worker_ptr_->PullSparse(pull_result_ptr.data(),
                                        table_id,
                                        fea_keys.data(),
                                        fea_keys.size(),
                                        is_training);
T
tangwei12 已提交
351 352 353 354 355 356 357 358 359
  status.wait();
  auto ret = status.get();
  if (ret != 0) {
    LOG(ERROR) << "fleet pull sparse failed, status[" << ret << "]";
    sleep(sleep_seconds_before_fail_exit_);
  }
}

void FleetWrapper::PullDenseVarsAsync(
360 361
    const Scope& scope,
    const uint64_t tid,
T
tangwei12 已提交
362
    const std::vector<std::string>& var_names,
363 364
    std::vector<std::future<int32_t>>* pull_dense_status,
    bool in_cpu) {
Z
zhaocaibei123 已提交
365
  auto& regions = regions_[tid];
T
tangwei12 已提交
366 367 368 369 370 371 372 373 374 375 376 377 378
  regions.clear();
  regions.resize(var_names.size());
  for (auto i = 0u; i < var_names.size(); ++i) {
    std::string varname = var_names[i];
    if (!in_cpu) {
      varname = var_names[i] + "pin";
    }
    Variable* var = scope.FindVar(varname);
    LoDTensor* tensor = var->GetMutable<LoDTensor>();
    float* w = tensor->data<float>();
    paddle::distributed::Region reg(w, tensor->numel());
    regions[i] = std::move(reg);
  }
Z
zhaocaibei123 已提交
379 380

  auto status = worker_ptr_->PullDense(regions.data(), regions.size(), tid);
T
tangwei12 已提交
381 382 383 384
  pull_dense_status->push_back(std::move(status));
}

void FleetWrapper::PullDenseVarsSync(
385 386
    const Scope& scope,
    const uint64_t tid,
T
tangwei12 已提交
387
    const std::vector<std::string>& var_names) {
Z
zhaocaibei123 已提交
388
  auto& regions = regions_[tid];
T
tangwei12 已提交
389 390 391 392 393
  regions.clear();
  regions.reserve(var_names.size());
  for (auto& t : var_names) {
    Variable* var = scope.FindVar(t);
    LoDTensor* tensor = var->GetMutable<LoDTensor>();
394 395 396 397 398
    if (!platform::is_gpu_place(tensor->place())) {
      float* w = tensor->data<float>();
      paddle::distributed::Region reg(w, tensor->numel());
      regions.emplace_back(std::move(reg));
    }
T
tangwei12 已提交
399
  }
Z
zhaocaibei123 已提交
400
  auto status = worker_ptr_->PullDense(regions.data(), regions.size(), tid);
T
tangwei12 已提交
401 402 403 404
  status.wait();
}

void FleetWrapper::PushDenseParamSync(
405 406
    const Scope& scope,
    const uint64_t table_id,
T
tangwei12 已提交
407 408 409 410 411 412 413
    const std::vector<std::string>& var_names) {
  auto place = platform::CPUPlace();
  std::vector<paddle::distributed::Region> regions;
  for (auto& t : var_names) {
    Variable* var = scope.FindVar(t);
    CHECK(var != nullptr) << "var[" << t << "] not found";
    LoDTensor* tensor = var->GetMutable<LoDTensor>();
414 415 416 417 418
    if (!platform::is_gpu_place(tensor->place())) {
      float* g = tensor->mutable_data<float>(place);
      paddle::distributed::Region reg(g, tensor->numel());
      regions.emplace_back(std::move(reg));
    }
T
tangwei12 已提交
419
  }
420
  auto push_status =
Z
zhaocaibei123 已提交
421
      worker_ptr_->PushDenseParam(regions.data(), regions.size(), table_id);
T
tangwei12 已提交
422 423 424 425 426 427
  push_status.wait();
  auto status = push_status.get();
  CHECK(status == 0) << "push dense param failed, status[" << status << "]";
}

void FleetWrapper::PushDenseVarsSync(
428 429
    Scope* scope,
    const uint64_t table_id,
T
tangwei12 已提交
430 431 432
    const std::vector<std::string>& var_names) {}

void FleetWrapper::PushDenseVarsAsync(
433 434
    const Scope& scope,
    const uint64_t table_id,
T
tangwei12 已提交
435
    const std::vector<std::string>& var_names,
436 437
    std::vector<std::future<int32_t>>* push_sparse_status,
    float scale_datanorm,
T
tangwei12 已提交
438
    int batch_size) {
Z
zhaocaibei123 已提交
439 440 441 442 443 444
  auto place = platform::CPUPlace();
  std::vector<paddle::distributed::Region> regions;
  for (auto& t : var_names) {
    Variable* var = scope.FindVar(t);
    CHECK(var != nullptr) << "var[" << t << "] not found";
    LoDTensor* tensor = var->GetMutable<LoDTensor>();
445
    int count = tensor->numel();
Z
zhaocaibei123 已提交
446
    float* g = tensor->mutable_data<float>(place);
447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462
    // TODO(zhaocaibei123): how to get batch_size in op?
    if (scale_datanorm >= 0) {
      if (t.find(".batch_size@GRAD") != std::string::npos ||
          t.find(".batch_sum@GRAD") != std::string::npos) {
        Eigen::Map<Eigen::MatrixXf> mat(g, 1, count);
        float scale = 1.0 / batch_size;
        mat *= scale;
      } else if (t.find(".batch_square_sum@GRAD") != std::string::npos) {
        VLOG(3) << "epsilon: " << scale_datanorm;
        for (int i = 0; i < count; ++i) {
          g[i] = (g[i] - batch_size * scale_datanorm) / batch_size +
                 batch_size * scale_datanorm;
        }
      }
    }

Z
zhaocaibei123 已提交
463 464 465 466 467 468 469
    paddle::distributed::Region reg(g, tensor->numel());
    regions.emplace_back(std::move(reg));
    VLOG(3) << "FleetWrapper::PushDenseVarsAsync Var " << t << " talbe_id "
            << table_id << " Temp_data[0] " << g[0] << " Temp_data[-1] "
            << g[tensor->numel() - 1];
  }

Z
zhaocaibei123 已提交
470 471
  auto push_status =
      worker_ptr_->PushDense(regions.data(), regions.size(), table_id);
T
tangwei12 已提交
472 473 474
}

void FleetWrapper::PushSparseVarsAsync(
475 476
    const Scope& scope,
    const uint64_t table_id,
T
tangwei12 已提交
477 478 479 480 481 482 483
    const std::string& grad_varname,
    std::vector<std::future<int32_t>>* push_sparse_status) {
  std::vector<std::string> varnames;
  varnames.push_back(grad_varname);

  auto* communicator = Communicator::GetInstance();
  PADDLE_ENFORCE_EQ(
484 485
      communicator->Check(table_id),
      true,
T
tangwei12 已提交
486 487 488 489 490 491
      platform::errors::InvalidArgument(
          "can not find table: %s, please check your config", table_id));
  communicator->Send(varnames, scope);
}

void FleetWrapper::PushSparseVarsWithLabelAsync(
492 493 494 495
    const Scope& scope,
    const uint64_t table_id,
    const std::vector<uint64_t>& fea_keys,
    const std::vector<float>& fea_labels,
T
tangwei12 已提交
496
    const std::vector<std::string>& sparse_key_names,
497 498
    const std::vector<std::string>& sparse_grad_names,
    const int emb_dim,
T
tangwei12 已提交
499
    std::vector<std::vector<float>>* push_values,
500 501 502 503 504 505
    std::vector<std::future<int32_t>>* push_sparse_status,
    const int batch_size,
    const bool use_cvm,
    const bool dump_slot,
    std::vector<uint64_t>* sparse_push_keys,
    const bool no_cvm) {
T
tangwei12 已提交
506 507 508 509 510
  // not support
  return;
}

void FleetWrapper::PushSparseFromTensorWithLabelAsync(
511 512 513 514 515 516 517 518
    const Scope& scope,
    const uint64_t table_id,
    int fea_dim,
    uint64_t padding_id,
    bool scale_sparse,
    const std::string& accesor,
    const std::string& click_name,
    platform::Place place,
T
tangwei12 已提交
519 520 521 522 523 524 525
    const std::vector<std::string>& input_names,
    std::vector<const LoDTensor*>* inputs,
    std::vector<const LoDTensor*>* outputs) {
  // not support
  return;
}

Z
zhaocaibei123 已提交
526
void FleetWrapper::PushSparseFromTensorAsync(
527 528 529 530 531
    const uint64_t table_id,
    int fea_dim,
    uint64_t padding_id,
    platform::Place place,
    std::vector<const LoDTensor*>* inputs,
532
    std::vector<int>& slots,
533 534 535 536
    const LoDTensor* shows,
    const LoDTensor* clks,
    std::vector<LoDTensor*>* outputs,
    bool use_cvm_op) {
537
  CHECK(slots.size() == inputs->size());
Z
zhaocaibei123 已提交
538
  int batch_size = -1;
Z
zhaocaibei123 已提交
539
  bool batch_size_consist = true;
Z
zhaocaibei123 已提交
540
  for (auto* input : *inputs) {
D
danleifeng 已提交
541
    size_t cur_batch_size =
Z
zhaocaibei123 已提交
542 543
        input->lod().size() ? input->lod()[0].size() - 1 : input->dims()[0];
    if (batch_size == -1) {
D
danleifeng 已提交
544 545
      batch_size = int(cur_batch_size);
    } else if (batch_size != int(cur_batch_size)) {
Z
zhaocaibei123 已提交
546 547 548
      // CHECK(batch_size == cur_batch_size);  // NOLINT
      batch_size_consist = false;
      break;
Z
zhaocaibei123 已提交
549 550 551 552
    }
  }
  CHECK(batch_size > 0);  // NOLINT

D
danleifeng 已提交
553
  size_t show_size =
Z
zhaocaibei123 已提交
554
      shows->lod().size() ? shows->lod()[0].size() - 1 : shows->dims()[0];
D
danleifeng 已提交
555 556
  CHECK(show_size == size_t(batch_size) || show_size == 1);
  size_t clk_size =
Z
zhaocaibei123 已提交
557
      clks->lod().size() ? clks->lod()[0].size() - 1 : clks->dims()[0];
D
danleifeng 已提交
558
  CHECK(clk_size == size_t(batch_size) || clk_size == 1);
Z
zhaocaibei123 已提交
559

560
  CHECK(outputs->size() == inputs->size());
Z
zhaocaibei123 已提交
561 562 563 564 565 566 567 568 569 570 571 572
  std::vector<uint64_t> push_keys;
  push_keys.reserve(MAX_FEASIGN_NUM / 100);
  std::vector<std::vector<float>> push_values;
  push_values.reserve(MAX_FEASIGN_NUM / 100);
  size_t output_len = 0;
  size_t input_idx = 0;

  VLOG(2) << "fleet.cc::emb_dim: " << fea_dim;

  // TODO(zhaocaibei123): check type of show/clk is int? float? uint64?
  // const long int* show_tensor = shows->data<int64_t>();
  // const long int* clk_tensor = clks->data<int64_t>();
573 574
  const float* show_tensor = shows->data<float>();
  const float* clk_tensor = clks->data<float>();
Z
zhaocaibei123 已提交
575 576

  for (size_t index = 0; index < inputs->size(); ++index) {
577 578 579 580 581 582 583 584
    framework::LoDTensor* g_tensor = outputs->at(index);
    float* g = g_tensor->data<float>();
    // no cvm
    if (batch_size_consist) {  // TODO(zhaocaibei123): add config
                               // scale_sparse_gradient_with_batch_size_
      Eigen::Map<
          Eigen::Matrix<float, Eigen::Dynamic, Eigen::Dynamic, Eigen::RowMajor>>
          g_mat(g, g_tensor->numel() / fea_dim, fea_dim);
585 586 587 588 589
      if (use_cvm_op) {
        g_mat.rightCols(fea_dim - 2) *= batch_size;
      } else {
        g_mat.rightCols(fea_dim) *= batch_size;
      }
590 591
    }

Z
zhaocaibei123 已提交
592 593 594
    const framework::LoDTensor* tensor = inputs->at(index);
    const int64_t* ids = tensor->data<int64_t>();
    size_t len = tensor->numel();
595
    output_len = 0;
Z
zhaocaibei123 已提交
596 597

    if (tensor->lod().size() > 0) {
Z
zhangchunle 已提交
598
      for (size_t i = 0; i < tensor->lod()[0].size() - 1; ++i) {
599
        for (size_t j = tensor->lod()[0][i]; j < tensor->lod()[0][i + 1];
Z
zhaocaibei123 已提交
600 601 602 603 604 605
             ++j, output_len += fea_dim) {
          uint64_t real_id = static_cast<uint64_t>(ids[j]);
          if (real_id == padding_id) {
            continue;
          }
          push_keys.emplace_back(real_id);
606 607
          if (use_cvm_op) {
            push_values.emplace_back(fea_dim + 1);
608
            push_values.back()[0] = static_cast<float>(slots[index]);
609 610 611 612 613
            float* data = push_values.back().data() + 1;
            memcpy(data, g + output_len, sizeof(float) * fea_dim);
          } else {
            push_values.emplace_back(fea_dim + 3);
            // slot show clk grad... consistent with CtrCommonPushValue defined
614 615
            // in ctr_accessor.h
            push_values.back()[0] = static_cast<float>(slots[index]);
D
danleifeng 已提交
616 617 618 619
            push_values.back()[1] =
                (i >= show_size ? 1 : static_cast<float>(show_tensor[i]));
            push_values.back()[2] =
                (i >= clk_size ? 0 : static_cast<float>(clk_tensor[i]));
620 621 622 623 624 625 626
            float* data = push_values.back().data() + 3;
            memcpy(data, g + output_len, sizeof(float) * fea_dim);
          }
          ++input_idx;
        }
      }
    } else {
Z
zhangchunle 已提交
627
      for (size_t i = 0; i < len; ++i, output_len += fea_dim) {
628 629 630 631 632 633 634
        uint64_t real_id = static_cast<uint64_t>(ids[i]);
        if (real_id == padding_id) {
          continue;
        }
        push_keys.emplace_back(real_id);
        if (use_cvm_op) {
          push_values.emplace_back(fea_dim + 1);
635
          push_values.back()[0] = static_cast<float>(slots[index]);
636 637 638
          float* data = push_values.back().data() + 1;
          memcpy(data, g + output_len, sizeof(float) * fea_dim);
        } else {
Z
zhaocaibei123 已提交
639 640 641
          push_values.emplace_back(fea_dim + 3);
          // slot show clk grad... consistent with CtrCommonPushValue defined in
          // ctr_accessor.h
642 643 644
          push_values.back()[0] = static_cast<float>(slots[index]);
          push_values.back()[1] = (i >= show_size ? 1 : show_tensor[i]);
          push_values.back()[2] = (i >= clk_size ? 0 : clk_tensor[i]);
Z
zhaocaibei123 已提交
645
          float* data = push_values.back().data() + 3;
646
          memcpy(data, g + output_len, sizeof(float) * fea_dim);
Z
zhaocaibei123 已提交
647 648 649 650
        }
        ++input_idx;
      }
    }
Z
zhangchunle 已提交
651
    CHECK(static_cast<int64_t>(output_len) == g_tensor->numel());
Z
zhaocaibei123 已提交
652 653 654 655 656 657 658 659
  }

  std::vector<float*> push_g_vec(input_idx, nullptr);

  for (auto i = 0u; i < push_keys.size(); ++i) {
    push_g_vec[i] = push_values.at(i).data();
  }

660 661
  auto status = worker_ptr_->PushSparse(table_id,
                                        push_keys.data(),
Z
zhaocaibei123 已提交
662 663
                                        (const float**)push_g_vec.data(),
                                        push_keys.size());
Z
zhaocaibei123 已提交
664 665 666
}

void FleetWrapper::LoadModel(const std::string& path, const int mode) {
Z
zhaocaibei123 已提交
667
  auto ret = worker_ptr_->Load(path, std::to_string(mode));
T
tangwei12 已提交
668 669 670 671 672 673 674
  ret.wait();
  if (ret.get() != 0) {
    LOG(ERROR) << "load model from path:" << path << " failed";
  }
}

void FleetWrapper::LoadModelOneTable(const uint64_t table_id,
675 676
                                     const std::string& path,
                                     const int mode) {
Z
zhaocaibei123 已提交
677
  auto ret = worker_ptr_->Load(table_id, path, std::to_string(mode));
T
tangwei12 已提交
678 679 680 681 682 683 684 685
  ret.wait();
  if (ret.get() != 0) {
    LOG(ERROR) << "load model of table id: " << table_id
               << ", from path: " << path << " failed";
  }
}

void FleetWrapper::SaveModel(const std::string& path, const int mode) {
Z
zhaocaibei123 已提交
686
  auto ret = worker_ptr_->Save(path, std::to_string(mode));
T
tangwei12 已提交
687 688 689 690 691 692 693 694
  ret.wait();
  int32_t feasign_cnt = ret.get();
  if (feasign_cnt == -1) {
    LOG(ERROR) << "save model failed";
  }
}

void FleetWrapper::SaveModelOneTable(const uint64_t table_id,
695 696
                                     const std::string& path,
                                     const int mode) {
Z
zhaocaibei123 已提交
697
  auto ret = worker_ptr_->Save(table_id, path, std::to_string(mode));
T
tangwei12 已提交
698 699 700 701 702 703 704
  ret.wait();
  if (ret.get() != 0) {
    LOG(ERROR) << "save model of table id: " << table_id
               << ", to path: " << path << " failed";
  }
}

705 706
void FleetWrapper::RecvAndSaveTable(const uint64_t table_id,
                                    const std::string& path) {
Z
zhaocaibei123 已提交
707
  auto ret = worker_ptr_->RecvAndSaveTable(table_id, path);
708 709 710 711 712 713
  if (ret != 0) {
    LOG(ERROR) << "save model of table id: " << table_id
               << ", to path: " << path << " failed";
  }
}

T
tangwei12 已提交
714
void FleetWrapper::PrintTableStat(const uint64_t table_id) {
Z
zhaocaibei123 已提交
715
  auto ret = worker_ptr_->PrintTableStat(table_id);
T
tangwei12 已提交
716 717 718 719 720 721 722
  ret.wait();
  int32_t err_code = ret.get();
  if (err_code == -1) {
    LOG(ERROR) << "print table stat failed";
  }
}

723
void FleetWrapper::ShrinkSparseTable(int table_id, int threshold) {
Z
zhaocaibei123 已提交
724
  auto ret = worker_ptr_->Shrink(table_id, std::to_string(threshold));
T
tangwei12 已提交
725
  ret.wait();
726 727 728 729
  int32_t err_code = ret.get();
  if (err_code == -1) {
    LOG(ERROR) << "shrink sparse table stat failed";
  }
T
tangwei12 已提交
730 731 732
}

void FleetWrapper::ClearModel() {
Z
zhaocaibei123 已提交
733
  auto ret = pserver_ptr_->_worker_ptr->Clear();
T
tangwei12 已提交
734 735 736 737
  ret.wait();
}

void FleetWrapper::ClearOneTable(const uint64_t table_id) {
Z
zhaocaibei123 已提交
738
  auto ret = pserver_ptr_->_worker_ptr->Clear(table_id);
T
tangwei12 已提交
739 740 741
  ret.wait();
}

742 743
void FleetWrapper::ShrinkDenseTable(int table_id,
                                    Scope* scope,
T
tangwei12 已提交
744
                                    std::vector<std::string> var_list,
745 746
                                    float decay,
                                    int emb_dim) {
T
tangwei12 已提交
747 748 749 750 751
  std::vector<paddle::distributed::Region> regions;
  for (std::string& name : var_list) {
    if (name.find("batch_sum") != std::string::npos) {
      Variable* var = scope->FindVar(name);
      CHECK(var != nullptr) << "var[" << name << "] not found";
752
      VLOG(3) << "prepare shrink dense batch_sum";
T
tangwei12 已提交
753 754 755 756 757
      LoDTensor* tensor = var->GetMutable<LoDTensor>();
      float* g = tensor->data<float>();

      // show_batch_sum += N * log(decay)
      std::string size_name = name;
758 759
      size_name.replace(
          size_name.find("batch_sum"), size_name.length(), "batch_size");
T
tangwei12 已提交
760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778
      Variable* var_size = scope->FindVar(size_name);
      CHECK(var_size != nullptr) << "var[" << size_name << "] not found";
      VLOG(3) << "shrink dense batch_sum: " << name << ", " << size_name;
      float* g_size = var_size->GetMutable<LoDTensor>()->data<float>();

      for (int k = 0; k < tensor->numel(); k += emb_dim) {
        g[k] = g[k] + g_size[k] * log(decay);
      }
      paddle::distributed::Region reg(g, tensor->numel());
      regions.emplace_back(std::move(reg));
    } else {
      Variable* var = scope->FindVar(name);
      CHECK(var != nullptr) << "var[" << name << "] not found";
      LoDTensor* tensor = var->GetMutable<LoDTensor>();
      float* g = tensor->data<float>();
      paddle::distributed::Region reg(g, tensor->numel());
      regions.emplace_back(std::move(reg));
    }
  }
Z
zhaocaibei123 已提交
779
  auto push_status = pserver_ptr_->_worker_ptr->PushDenseParam(
T
tangwei12 已提交
780 781 782 783 784 785 786 787 788 789 790 791
      regions.data(), regions.size(), table_id);
  push_status.wait();
  auto status = push_status.get();
  if (status != 0) {
    // PADDLE_THORW(platform::errors::Fatal(
    //    "push shrink dense param failed, status is [%d].", status));
    sleep(sleep_seconds_before_fail_exit_);
    exit(-1);
  }
}

void FleetWrapper::ClientFlush() {
792 793 794 795
  if (worker_ptr_.get() == nullptr) {
    VLOG(0) << "worker_ptr null, do nothing";
    return;
  }
Z
zhaocaibei123 已提交
796
  auto ret = worker_ptr_->Flush();
T
tangwei12 已提交
797
  ret.wait();
798 799 800 801
  int32_t err_code = ret.get();
  if (err_code == -1) {
    LOG(ERROR) << "Client Flush failed";
  }
T
tangwei12 已提交
802 803 804 805
}

int FleetWrapper::RegisterClientToClientMsgHandler(int msg_type,
                                                   MsgHandlerFunc handler) {
806 807
  if (worker_ptr_.get() == nullptr) {
    VLOG(0) << "FleetWrapper::Client is null";
Z
zhaocaibei123 已提交
808 809
    return -1;
  } else {
Z
zhaocaibei123 已提交
810
    return worker_ptr_->RegisteClient2ClientMsgHandler(msg_type, handler);
Z
zhaocaibei123 已提交
811
  }
T
tangwei12 已提交
812 813 814 815
}

std::future<int32_t> FleetWrapper::SendClientToClientMsg(
    int msg_type, int to_client_id, const std::string& msg) {
Z
zhaocaibei123 已提交
816
  return worker_ptr_->SendClient2ClientMsg(msg_type, to_client_id, msg);
T
tangwei12 已提交
817 818
}

Z
zhaocaibei123 已提交
819 820 821 822 823 824 825 826 827 828 829 830 831 832
double FleetWrapper::GetCacheThreshold(int table_id) {
  double cache_threshold = 0.0;
  auto ret = worker_ptr_->Flush();
  ret.wait();
  ret = worker_ptr_->GetCacheThreshold(table_id, cache_threshold);
  ret.wait();
  if (cache_threshold < 0) {
    LOG(ERROR) << "get cache threshold failed";
    sleep(sleep_seconds_before_fail_exit_);
    exit(-1);
  }
  return cache_threshold;
}

833 834 835 836 837 838
void FleetWrapper::CacheShuffle(int table_id,
                                const std::string& path,
                                const int mode,
                                const double cache_threshold) {
  auto ret = worker_ptr_->CacheShuffle(
      table_id, path, std::to_string(mode), std::to_string(cache_threshold));
Z
zhaocaibei123 已提交
839 840 841 842 843 844 845 846 847
  ret.wait();
  int32_t feasign_cnt = ret.get();
  if (feasign_cnt == -1) {
    LOG(ERROR) << "cache shuffle failed";
    sleep(sleep_seconds_before_fail_exit_);
    exit(-1);
  }
}

848 849
int32_t FleetWrapper::SaveCache(int table_id,
                                const std::string& path,
Z
zhaocaibei123 已提交
850 851 852 853 854 855 856 857 858 859 860 861
                                const int mode) {
  auto ret = worker_ptr_->SaveCache(table_id, path, std::to_string(mode));
  ret.wait();
  int32_t feasign_cnt = ret.get();
  if (feasign_cnt == -1) {
    LOG(ERROR) << "table save cache failed";
    sleep(sleep_seconds_before_fail_exit_);
    exit(-1);
  }
  return feasign_cnt;
}

Z
zhaocaibei123 已提交
862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879
void FleetWrapper::Revert() {
  auto ret = worker_ptr_->Revert();
  ret.wait();
  if (ret.get() == -1) {
    LOG(ERROR) << "table revert failed";
    exit(-1);
  }
}

void FleetWrapper::CheckSavePrePatchDone() {
  auto ret = worker_ptr_->CheckSavePrePatchDone();
  ret.wait();
  if (ret.get() == -1) {
    LOG(ERROR) << "table revert failed";
    exit(-1);
  }
}

T
tangwei12 已提交
880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896
std::default_random_engine& FleetWrapper::LocalRandomEngine() {
  struct engine_wrapper_t {
    std::default_random_engine engine;

    engine_wrapper_t() {
      struct timespec tp;
      clock_gettime(CLOCK_REALTIME, &tp);
      double cur_time = tp.tv_sec + tp.tv_nsec * 1e-9;
      static std::atomic<uint64_t> x(0);
      std::seed_seq sseq = {x++, x++, x++, (uint64_t)(cur_time * 1000)};
      engine.seed(sseq);
    }
  };
  thread_local engine_wrapper_t r;
  return r.engine;
}

897 898 899
size_t FleetWrapper::GetAbsoluteSum(size_t start,
                                    size_t end,
                                    size_t level,
T
tangwei12 已提交
900 901 902 903 904 905 906 907 908 909 910 911 912 913 914
                                    const framework::LoD& lod) {
  if (level >= lod.size() - 1) {
    return end - start;
  }
  size_t ret = 0;
  for (size_t i = start; i < end - 1; ++i) {
    size_t pos1 = lod[level][i];
    size_t pos2 = lod[level][i + 1];
    ret += GetAbsoluteSum(pos1, pos2, level + 1, lod);
  }
  return ret;
}

}  // end namespace distributed
}  // end namespace paddle