fleet.cc 29.6 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"
T
tangwei12 已提交
21 22 23 24 25 26 27 28 29 30 31 32 33 34

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;
35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
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(
    const uint64_t src_table_id, const uint64_t dest_table_id,
    const std::vector<uint64_t>& feasign_list) {
  VLOG(0) << "CopyTableByFeasign support later";
  return 0;
}
T
tangwei12 已提交
54 55 56 57 58 59 60 61 62 63 64 65 66 67

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
68
  pserver_ptr_->_server_ptr->GetTable(table_id)->Load(path, meta);
T
tangwei12 已提交
69 70
}

71 72
void FleetWrapper::InitServer(
    const std::string& dist_desc,
T
tangwei12 已提交
73
    const std::vector<std::string>& host_sign_list, int index, int trainers,
74
    const std::vector<framework::ProgramDesc>& server_sub_program) {
T
tangwei12 已提交
75 76 77 78
  if (!is_initialized_) {
    VLOG(3) << "Going to init server";
    pserver_ptr_ = std::shared_ptr<paddle::distributed::PSCore>(
        new paddle::distributed::PSCore());
Z
zhaocaibei123 已提交
79 80
    pserver_ptr_->InitServer(dist_desc, &host_sign_list, host_sign_list.size(),
                             index, trainers, server_sub_program);
T
tangwei12 已提交
81 82 83 84 85 86
    is_initialized_ = true;
  } else {
    VLOG(3) << "Server can be initialized only once";
  }
}

87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105
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
106

107 108 109 110 111 112 113 114 115 116 117 118 119 120
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
121
      ps_env_.SetPsServers(&host_sign_list, servers);
122
      worker_ptr_ = std::shared_ptr<paddle::distributed::PSClient>(
Z
zhaocaibei123 已提交
123 124
          paddle::distributed::PSClientFactory::Create(ps_param));
      worker_ptr_->Configure(ps_param, dense_pull_regions, ps_env_, index);
125
    }
T
tangwei12 已提交
126
  } else {
127
    VLOG(3) << "Client can be initialized only once";
T
tangwei12 已提交
128 129 130 131 132
  }
}

void FleetWrapper::StopServer() {
  VLOG(3) << "Going to stop server";
Z
zhaocaibei123 已提交
133
  auto status = worker_ptr_->StopServer();
T
tangwei12 已提交
134 135 136 137 138
  status.wait();
}

void FleetWrapper::FinalizeWorker() {
  VLOG(3) << "Going to finalize worker";
Z
zhaocaibei123 已提交
139
  worker_ptr_->FinalizeWorker();
T
tangwei12 已提交
140 141 142 143 144 145 146 147 148 149
}

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
150
  auto ret = pserver_ptr_->RunServer(ip, port);
T
tangwei12 已提交
151 152 153 154 155
  return ret;
}

std::vector<uint64_t> FleetWrapper::GetClientsInfo() {
  VLOG(3) << "Going to get client info";
Z
zhaocaibei123 已提交
156
  std::vector<uint64_t> res = ps_env_.GetClientInfo();
157 158 159
  for (auto rr : res) {
    VLOG(2) << "FleetWrapper::GetClientInfo " << rr;
  }
Z
zhaocaibei123 已提交
160
  return res;
T
tangwei12 已提交
161 162
}

163 164
int FleetWrapper::SetClients(std::vector<uint64_t>& host_sign_list) {
  int node = host_sign_list.size();
Z
zhaocaibei123 已提交
165
  return ps_env_.SetPsClients(host_sign_list.data(), node);
166 167
}

T
tangwei12 已提交
168
void FleetWrapper::CreateClient2ClientConnection() {
Z
zhaocaibei123 已提交
169
  VLOG(1) << "Going to create client2client connection";
Z
zhaocaibei123 已提交
170 171 172
  worker_ptr_->CreateClient2ClientConnection(client2client_request_timeout_ms_,
                                             client2client_connect_timeout_ms_,
                                             client2client_max_retry_);
T
tangwei12 已提交
173 174
}

175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207
std::future<int32_t> FleetWrapper::PullSparseVarsAsync(
    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) {
  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;
Z
zhaocaibei123 已提交
208 209 210
  return pserver_ptr_->_worker_ptr->PullSparse(pull_result_ptr.data(), table_id,
                                               fea_keys->data(),
                                               fea_keys->size(), training);
211 212
}

T
tangwei12 已提交
213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255
void FleetWrapper::PullSparseVarsSync(
    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,
    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());
  }
256
  bool training = true;
Z
zhaocaibei123 已提交
257
  auto status = pserver_ptr_->_worker_ptr->PullSparse(
258 259
      pull_result_ptr.data(), table_id, fea_keys->data(), fea_keys->size(),
      training);
T
tangwei12 已提交
260 261 262 263 264 265 266 267 268 269 270 271
  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);
    }
  }
}

272 273 274
// is_training is true means training, false means inference, the behavior is
// different on pserver

T
tangwei12 已提交
275 276 277
void FleetWrapper::PullSparseToTensorSync(const uint64_t table_id, int fea_dim,
                                          uint64_t padding_id,
                                          platform::Place place,
278
                                          bool is_training,
T
tangwei12 已提交
279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314
                                          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) {
        memcpy(output_data + output_len, init_value.data(),
               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
315 316 317 318

  auto status =
      worker_ptr_->PullSparse(pull_result_ptr.data(), table_id, fea_keys.data(),
                              fea_keys.size(), is_training);
T
tangwei12 已提交
319 320 321 322 323 324 325 326 327 328 329 330
  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(
    const Scope& scope, const uint64_t tid,
    const std::vector<std::string>& var_names,
    std::vector<std::future<int32_t>>* pull_dense_status, bool in_cpu) {
Z
zhaocaibei123 已提交
331
  auto& regions = regions_[tid];
T
tangwei12 已提交
332 333 334 335 336 337 338 339 340 341 342 343 344
  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
345 346

  auto status = worker_ptr_->PullDense(regions.data(), regions.size(), tid);
T
tangwei12 已提交
347 348 349 350 351 352
  pull_dense_status->push_back(std::move(status));
}

void FleetWrapper::PullDenseVarsSync(
    const Scope& scope, const uint64_t tid,
    const std::vector<std::string>& var_names) {
Z
zhaocaibei123 已提交
353
  auto& regions = regions_[tid];
T
tangwei12 已提交
354 355 356 357 358
  regions.clear();
  regions.reserve(var_names.size());
  for (auto& t : var_names) {
    Variable* var = scope.FindVar(t);
    LoDTensor* tensor = var->GetMutable<LoDTensor>();
359 360 361 362 363
    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
364
  }
Z
zhaocaibei123 已提交
365
  auto status = worker_ptr_->PullDense(regions.data(), regions.size(), tid);
T
tangwei12 已提交
366 367 368 369 370 371 372 373 374 375 376 377
  status.wait();
}

void FleetWrapper::PushDenseParamSync(
    const Scope& scope, const uint64_t table_id,
    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>();
378 379 380 381 382
    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
383
  }
384
  auto push_status =
Z
zhaocaibei123 已提交
385
      worker_ptr_->PushDenseParam(regions.data(), regions.size(), table_id);
T
tangwei12 已提交
386 387 388 389 390 391 392 393 394 395 396 397 398 399
  push_status.wait();
  auto status = push_status.get();
  CHECK(status == 0) << "push dense param failed, status[" << status << "]";
}

void FleetWrapper::PushDenseVarsSync(
    Scope* scope, const uint64_t table_id,
    const std::vector<std::string>& var_names) {}

void FleetWrapper::PushDenseVarsAsync(
    const Scope& scope, const uint64_t table_id,
    const std::vector<std::string>& var_names,
    std::vector<std::future<int32_t>>* push_sparse_status, float scale_datanorm,
    int batch_size) {
Z
zhaocaibei123 已提交
400 401 402 403 404 405
  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>();
406
    int count = tensor->numel();
Z
zhaocaibei123 已提交
407
    float* g = tensor->mutable_data<float>(place);
408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423
    // 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
424 425 426 427 428 429 430
    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
431 432
  auto push_status =
      worker_ptr_->PushDense(regions.data(), regions.size(), table_id);
T
tangwei12 已提交
433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473
}

void FleetWrapper::PushSparseVarsAsync(
    const Scope& scope, const uint64_t table_id,
    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(
      communicator->Check(table_id), true,
      platform::errors::InvalidArgument(
          "can not find table: %s, please check your config", table_id));
  communicator->Send(varnames, scope);
}

void FleetWrapper::PushSparseVarsWithLabelAsync(
    const Scope& scope, const uint64_t table_id,
    const std::vector<uint64_t>& fea_keys, const std::vector<float>& fea_labels,
    const std::vector<std::string>& sparse_key_names,
    const std::vector<std::string>& sparse_grad_names, const int emb_dim,
    std::vector<std::vector<float>>* push_values,
    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) {
  // not support
  return;
}

void FleetWrapper::PushSparseFromTensorWithLabelAsync(
    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,
    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
474 475 476 477
void FleetWrapper::PushSparseFromTensorAsync(
    const uint64_t table_id, int fea_dim, uint64_t padding_id,
    platform::Place place, std::vector<const LoDTensor*>* inputs,
    const LoDTensor* shows, const LoDTensor* clks,
478
    std::vector<LoDTensor*>* outputs, bool use_cvm_op) {
Z
zhaocaibei123 已提交
479
  int batch_size = -1;
Z
zhaocaibei123 已提交
480
  bool batch_size_consist = true;
Z
zhaocaibei123 已提交
481 482 483 484 485
  for (auto* input : *inputs) {
    int cur_batch_size =
        input->lod().size() ? input->lod()[0].size() - 1 : input->dims()[0];
    if (batch_size == -1) {
      batch_size = cur_batch_size;
486
    } else if (batch_size != cur_batch_size) {
Z
zhaocaibei123 已提交
487 488 489
      // CHECK(batch_size == cur_batch_size);  // NOLINT
      batch_size_consist = false;
      break;
Z
zhaocaibei123 已提交
490 491 492 493 494 495 496 497 498 499 500
    }
  }
  CHECK(batch_size > 0);  // NOLINT

  int show_size =
      shows->lod().size() ? shows->lod()[0].size() - 1 : shows->dims()[0];
  CHECK(show_size == batch_size || show_size == 1);
  int clk_size =
      clks->lod().size() ? clks->lod()[0].size() - 1 : clks->dims()[0];
  CHECK(clk_size == batch_size || clk_size == 1);

501
  CHECK(outputs->size() == inputs->size());
Z
zhaocaibei123 已提交
502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517
  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>();
  const int64_t* show_tensor = shows->data<int64_t>();
  const int64_t* clk_tensor = clks->data<int64_t>();

  for (size_t index = 0; index < inputs->size(); ++index) {
518 519 520 521 522 523 524 525
    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);
526 527 528 529 530
      if (use_cvm_op) {
        g_mat.rightCols(fea_dim - 2) *= batch_size;
      } else {
        g_mat.rightCols(fea_dim) *= batch_size;
      }
531 532
    }

Z
zhaocaibei123 已提交
533 534 535
    const framework::LoDTensor* tensor = inputs->at(index);
    const int64_t* ids = tensor->data<int64_t>();
    size_t len = tensor->numel();
536
    output_len = 0;
Z
zhaocaibei123 已提交
537 538

    if (tensor->lod().size() > 0) {
Z
zhangchunle 已提交
539
      for (size_t i = 0; i < tensor->lod()[0].size() - 1; ++i) {
540
        for (size_t j = tensor->lod()[0][i]; j < tensor->lod()[0][i + 1];
Z
zhaocaibei123 已提交
541 542 543 544 545 546
             ++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);
547 548 549 550 551 552 553 554 555 556 557
          if (use_cvm_op) {
            push_values.emplace_back(fea_dim + 1);
            push_values.back()[0] = 2;  // TODO(zhaocaibei123): slot
            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
            // in
            // ctr_accessor.h
            push_values.back()[0] = 2;  // TODO(zhaocaibei123): slot
Z
zhangchunle 已提交
558 559 560 561 562 563
            push_values.back()[1] = (static_cast<int>(i) >= show_size
                                         ? 1
                                         : static_cast<float>(show_tensor[i]));
            push_values.back()[2] = (static_cast<int>(i) >= clk_size
                                         ? 0
                                         : static_cast<float>(clk_tensor[i]));
564 565 566 567 568 569 570
            float* data = push_values.back().data() + 3;
            memcpy(data, g + output_len, sizeof(float) * fea_dim);
          }
          ++input_idx;
        }
      }
    } else {
Z
zhangchunle 已提交
571
      for (size_t i = 0; i < len; ++i, output_len += fea_dim) {
572 573 574 575 576 577 578 579 580 581 582
        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);
          push_values.back()[0] = 2;  // TODO(zhaocaibei123): slot
          float* data = push_values.back().data() + 1;
          memcpy(data, g + output_len, sizeof(float) * fea_dim);
        } else {
Z
zhaocaibei123 已提交
583 584 585 586
          push_values.emplace_back(fea_dim + 3);
          // slot show clk grad... consistent with CtrCommonPushValue defined in
          // ctr_accessor.h
          push_values.back()[0] = 2;  // TODO(zhaocaibei123): slot
Z
zhangchunle 已提交
587 588 589 590 591 592
          push_values.back()[1] = (static_cast<int>(i) >= show_size
                                       ? 1
                                       : static_cast<float>(show_tensor[i]));
          push_values.back()[2] = (static_cast<int>(i) >= clk_size
                                       ? 0
                                       : static_cast<float>(clk_tensor[i]));
Z
zhaocaibei123 已提交
593
          float* data = push_values.back().data() + 3;
594
          memcpy(data, g + output_len, sizeof(float) * fea_dim);
Z
zhaocaibei123 已提交
595 596 597 598
        }
        ++input_idx;
      }
    }
Z
zhangchunle 已提交
599
    CHECK(static_cast<int64_t>(output_len) == g_tensor->numel());
Z
zhaocaibei123 已提交
600 601 602 603 604 605 606 607
  }

  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();
  }

Z
zhaocaibei123 已提交
608 609 610
  auto status = worker_ptr_->PushSparse(table_id, push_keys.data(),
                                        (const float**)push_g_vec.data(),
                                        push_keys.size());
Z
zhaocaibei123 已提交
611 612 613
}

void FleetWrapper::LoadModel(const std::string& path, const int mode) {
Z
zhaocaibei123 已提交
614
  auto ret = worker_ptr_->Load(path, std::to_string(mode));
T
tangwei12 已提交
615 616 617 618 619 620 621 622
  ret.wait();
  if (ret.get() != 0) {
    LOG(ERROR) << "load model from path:" << path << " failed";
  }
}

void FleetWrapper::LoadModelOneTable(const uint64_t table_id,
                                     const std::string& path, const int mode) {
Z
zhaocaibei123 已提交
623
  auto ret = worker_ptr_->Load(table_id, path, std::to_string(mode));
T
tangwei12 已提交
624 625 626 627 628 629 630 631
  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
632
  auto ret = worker_ptr_->Save(path, std::to_string(mode));
T
tangwei12 已提交
633 634 635 636 637 638 639 640 641
  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,
                                     const std::string& path, const int mode) {
Z
zhaocaibei123 已提交
642
  auto ret = worker_ptr_->Save(table_id, path, std::to_string(mode));
T
tangwei12 已提交
643 644 645 646 647 648 649
  ret.wait();
  if (ret.get() != 0) {
    LOG(ERROR) << "save model of table id: " << table_id
               << ", to path: " << path << " failed";
  }
}

650 651
void FleetWrapper::RecvAndSaveTable(const uint64_t table_id,
                                    const std::string& path) {
Z
zhaocaibei123 已提交
652
  auto ret = worker_ptr_->RecvAndSaveTable(table_id, path);
653 654 655 656 657 658
  if (ret != 0) {
    LOG(ERROR) << "save model of table id: " << table_id
               << ", to path: " << path << " failed";
  }
}

T
tangwei12 已提交
659
void FleetWrapper::PrintTableStat(const uint64_t table_id) {
Z
zhaocaibei123 已提交
660
  auto ret = worker_ptr_->PrintTableStat(table_id);
T
tangwei12 已提交
661 662 663 664 665 666 667
  ret.wait();
  int32_t err_code = ret.get();
  if (err_code == -1) {
    LOG(ERROR) << "print table stat failed";
  }
}

668
void FleetWrapper::ShrinkSparseTable(int table_id, int threshold) {
Z
zhaocaibei123 已提交
669
  auto ret = worker_ptr_->Shrink(table_id, std::to_string(threshold));
T
tangwei12 已提交
670
  ret.wait();
671 672 673 674
  int32_t err_code = ret.get();
  if (err_code == -1) {
    LOG(ERROR) << "shrink sparse table stat failed";
  }
T
tangwei12 已提交
675 676 677
}

void FleetWrapper::ClearModel() {
Z
zhaocaibei123 已提交
678
  auto ret = pserver_ptr_->_worker_ptr->Clear();
T
tangwei12 已提交
679 680 681 682
  ret.wait();
}

void FleetWrapper::ClearOneTable(const uint64_t table_id) {
Z
zhaocaibei123 已提交
683
  auto ret = pserver_ptr_->_worker_ptr->Clear(table_id);
T
tangwei12 已提交
684 685 686 687 688 689 690 691 692 693 694
  ret.wait();
}

void FleetWrapper::ShrinkDenseTable(int table_id, Scope* scope,
                                    std::vector<std::string> var_list,
                                    float decay, int emb_dim) {
  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";
695
      VLOG(3) << "prepare shrink dense batch_sum";
T
tangwei12 已提交
696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721
      LoDTensor* tensor = var->GetMutable<LoDTensor>();
      float* g = tensor->data<float>();

      // show_batch_sum += N * log(decay)
      std::string size_name = name;
      size_name.replace(size_name.find("batch_sum"), size_name.length(),
                        "batch_size");
      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
722
  auto push_status = pserver_ptr_->_worker_ptr->PushDenseParam(
T
tangwei12 已提交
723 724 725 726 727 728 729 730 731 732 733 734
      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() {
735 736 737 738
  if (worker_ptr_.get() == nullptr) {
    VLOG(0) << "worker_ptr null, do nothing";
    return;
  }
Z
zhaocaibei123 已提交
739
  auto ret = worker_ptr_->Flush();
T
tangwei12 已提交
740
  ret.wait();
741 742 743 744
  int32_t err_code = ret.get();
  if (err_code == -1) {
    LOG(ERROR) << "Client Flush failed";
  }
T
tangwei12 已提交
745 746 747 748
}

int FleetWrapper::RegisterClientToClientMsgHandler(int msg_type,
                                                   MsgHandlerFunc handler) {
749 750
  if (worker_ptr_.get() == nullptr) {
    VLOG(0) << "FleetWrapper::Client is null";
Z
zhaocaibei123 已提交
751 752
    return -1;
  } else {
Z
zhaocaibei123 已提交
753
    return worker_ptr_->RegisteClient2ClientMsgHandler(msg_type, handler);
Z
zhaocaibei123 已提交
754
  }
T
tangwei12 已提交
755 756 757 758
}

std::future<int32_t> FleetWrapper::SendClientToClientMsg(
    int msg_type, int to_client_id, const std::string& msg) {
Z
zhaocaibei123 已提交
759
  return worker_ptr_->SendClient2ClientMsg(msg_type, to_client_id, msg);
T
tangwei12 已提交
760 761
}

Z
zhaocaibei123 已提交
762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801
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;
}

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));
  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);
  }
}

int32_t FleetWrapper::SaveCache(int table_id, const std::string& path,
                                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;
}

T
tangwei12 已提交
802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834
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;
}

size_t FleetWrapper::GetAbsoluteSum(size_t start, size_t end, size_t level,
                                    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