fleet.cc 29.3 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 <google/protobuf/text_format.h>

17 18
#include "paddle/fluid/distributed/ps/service/communicator/communicator.h"
#include "paddle/fluid/distributed/ps/table/table.h"
19
#include "paddle/fluid/distributed/ps/wrapper/fleet.h"
T
tangwei12 已提交
20 21 22 23 24 25 26 27 28 29 30 31 32 33

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;
34 35 36 37 38 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(
    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
53 54 55 56 57 58 59 60 61 62 63 64 65 66

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
67
  pserver_ptr_->_server_ptr->GetTable(table_id)->Load(path, meta);
T
tangwei12 已提交
68 69
}

70 71
void FleetWrapper::InitServer(
    const std::string& dist_desc,
T
tangwei12 已提交
72
    const std::vector<std::string>& host_sign_list, int index, int trainers,
73
    const std::vector<framework::ProgramDesc>& server_sub_program) {
T
tangwei12 已提交
74 75 76 77
  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
78 79
    pserver_ptr_->InitServer(dist_desc, &host_sign_list, host_sign_list.size(),
                             index, trainers, server_sub_program);
T
tangwei12 已提交
80 81 82 83 84 85
    is_initialized_ = true;
  } else {
    VLOG(3) << "Server can be initialized only once";
  }
}

86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104
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
105

106 107 108 109 110 111 112 113 114 115 116 117 118 119
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
120
      ps_env_.SetPsServers(&host_sign_list, servers);
121
      worker_ptr_ = std::shared_ptr<paddle::distributed::PSClient>(
Z
zhaocaibei123 已提交
122 123
          paddle::distributed::PSClientFactory::Create(ps_param));
      worker_ptr_->Configure(ps_param, dense_pull_regions, ps_env_, index);
124
    }
T
tangwei12 已提交
125
  } else {
126
    VLOG(3) << "Client can be initialized only once";
T
tangwei12 已提交
127 128 129 130 131
  }
}

void FleetWrapper::StopServer() {
  VLOG(3) << "Going to stop server";
Z
zhaocaibei123 已提交
132
  auto status = worker_ptr_->StopServer();
T
tangwei12 已提交
133 134 135 136 137
  status.wait();
}

void FleetWrapper::FinalizeWorker() {
  VLOG(3) << "Going to finalize worker";
Z
zhaocaibei123 已提交
138
  worker_ptr_->FinalizeWorker();
T
tangwei12 已提交
139 140 141 142 143 144 145 146 147 148
}

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
149
  auto ret = pserver_ptr_->RunServer(ip, port);
T
tangwei12 已提交
150 151 152 153 154
  return ret;
}

std::vector<uint64_t> FleetWrapper::GetClientsInfo() {
  VLOG(3) << "Going to get client info";
Z
zhaocaibei123 已提交
155
  std::vector<uint64_t> res = ps_env_.GetClientInfo();
156 157 158
  for (auto rr : res) {
    VLOG(2) << "FleetWrapper::GetClientInfo " << rr;
  }
Z
zhaocaibei123 已提交
159
  return res;
T
tangwei12 已提交
160 161
}

162 163
int FleetWrapper::SetClients(std::vector<uint64_t>& host_sign_list) {
  int node = host_sign_list.size();
Z
zhaocaibei123 已提交
164
  return ps_env_.SetPsClients(host_sign_list.data(), node);
165 166
}

T
tangwei12 已提交
167
void FleetWrapper::CreateClient2ClientConnection() {
Z
zhaocaibei123 已提交
168
  VLOG(1) << "Going to create client2client connection";
Z
zhaocaibei123 已提交
169 170 171
  worker_ptr_->CreateClient2ClientConnection(client2client_request_timeout_ms_,
                                             client2client_connect_timeout_ms_,
                                             client2client_max_retry_);
T
tangwei12 已提交
172 173
}

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
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
207 208 209
  return pserver_ptr_->_worker_ptr->PullSparse(pull_result_ptr.data(), table_id,
                                               fea_keys->data(),
                                               fea_keys->size(), training);
210 211
}

T
tangwei12 已提交
212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254
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());
  }
255
  bool training = true;
Z
zhaocaibei123 已提交
256
  auto status = pserver_ptr_->_worker_ptr->PullSparse(
257 258
      pull_result_ptr.data(), table_id, fea_keys->data(), fea_keys->size(),
      training);
T
tangwei12 已提交
259 260 261 262 263 264 265 266 267 268 269 270
  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);
    }
  }
}

271 272 273
// is_training is true means training, false means inference, the behavior is
// different on pserver

T
tangwei12 已提交
274 275 276
void FleetWrapper::PullSparseToTensorSync(const uint64_t table_id, int fea_dim,
                                          uint64_t padding_id,
                                          platform::Place place,
277
                                          bool is_training,
T
tangwei12 已提交
278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313
                                          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
314 315 316 317

  auto status =
      worker_ptr_->PullSparse(pull_result_ptr.data(), table_id, fea_keys.data(),
                              fea_keys.size(), is_training);
T
tangwei12 已提交
318 319 320 321 322 323 324 325 326 327 328 329
  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
330
  auto& regions = regions_[tid];
T
tangwei12 已提交
331 332 333 334 335 336 337 338 339 340 341 342 343
  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
344 345

  auto status = worker_ptr_->PullDense(regions.data(), regions.size(), tid);
T
tangwei12 已提交
346 347 348 349 350 351
  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
352
  auto& regions = regions_[tid];
T
tangwei12 已提交
353 354 355 356 357
  regions.clear();
  regions.reserve(var_names.size());
  for (auto& t : var_names) {
    Variable* var = scope.FindVar(t);
    LoDTensor* tensor = var->GetMutable<LoDTensor>();
358 359 360 361 362
    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
363
  }
Z
zhaocaibei123 已提交
364
  auto status = worker_ptr_->PullDense(regions.data(), regions.size(), tid);
T
tangwei12 已提交
365 366 367 368 369 370 371 372 373 374 375 376
  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>();
377 378 379 380 381
    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
382
  }
383
  auto push_status =
Z
zhaocaibei123 已提交
384
      worker_ptr_->PushDenseParam(regions.data(), regions.size(), table_id);
T
tangwei12 已提交
385 386 387 388 389 390 391 392 393 394 395 396 397 398
  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
399 400 401 402 403 404
  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>();
405
    int count = tensor->numel();
Z
zhaocaibei123 已提交
406
    float* g = tensor->mutable_data<float>(place);
407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422
    // 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
423 424 425 426 427 428 429
    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
430 431
  auto push_status =
      worker_ptr_->PushDense(regions.data(), regions.size(), table_id);
T
tangwei12 已提交
432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472
}

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
473 474 475 476
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,
477
    std::vector<LoDTensor*>* outputs, bool use_cvm_op) {
Z
zhaocaibei123 已提交
478
  int batch_size = -1;
Z
zhaocaibei123 已提交
479
  bool batch_size_consist = true;
Z
zhaocaibei123 已提交
480 481 482 483 484
  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;
485
    } else if (batch_size != cur_batch_size) {
Z
zhaocaibei123 已提交
486 487 488
      // CHECK(batch_size == cur_batch_size);  // NOLINT
      batch_size_consist = false;
      break;
Z
zhaocaibei123 已提交
489 490 491 492 493 494 495 496 497 498 499
    }
  }
  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);

500
  CHECK(outputs->size() == inputs->size());
Z
zhaocaibei123 已提交
501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516
  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) {
517 518 519 520 521 522 523 524
    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);
525 526 527 528 529
      if (use_cvm_op) {
        g_mat.rightCols(fea_dim - 2) *= batch_size;
      } else {
        g_mat.rightCols(fea_dim) *= batch_size;
      }
530 531
    }

Z
zhaocaibei123 已提交
532 533 534
    const framework::LoDTensor* tensor = inputs->at(index);
    const int64_t* ids = tensor->data<int64_t>();
    size_t len = tensor->numel();
535
    output_len = 0;
Z
zhaocaibei123 已提交
536 537 538 539 540 541 542 543 544 545

    if (tensor->lod().size() > 0) {
      for (size_t i = 0; i < tensor->lod()[0].size() - 1; ++i) {
        for (int j = tensor->lod()[0][i]; j < tensor->lod()[0][i + 1];
             ++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);
546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579
          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
            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]));
            float* data = push_values.back().data() + 3;
            memcpy(data, g + output_len, sizeof(float) * fea_dim);
          }
          ++input_idx;
        }
      }
    } else {
      for (size_t i = 0; i < len; ++i, output_len += fea_dim) {
        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
580 581 582 583 584 585 586 587 588
          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
          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]));
          float* data = push_values.back().data() + 3;
589
          memcpy(data, g + output_len, sizeof(float) * fea_dim);
Z
zhaocaibei123 已提交
590 591 592 593
        }
        ++input_idx;
      }
    }
594
    CHECK(output_len == g_tensor->numel());
Z
zhaocaibei123 已提交
595 596 597 598 599 600 601 602
  }

  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
603 604 605
  auto status = worker_ptr_->PushSparse(table_id, push_keys.data(),
                                        (const float**)push_g_vec.data(),
                                        push_keys.size());
Z
zhaocaibei123 已提交
606 607 608
}

void FleetWrapper::LoadModel(const std::string& path, const int mode) {
Z
zhaocaibei123 已提交
609
  auto ret = worker_ptr_->Load(path, std::to_string(mode));
T
tangwei12 已提交
610 611 612 613 614 615 616 617
  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
618
  auto ret = worker_ptr_->Load(table_id, path, std::to_string(mode));
T
tangwei12 已提交
619 620 621 622 623 624 625 626
  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
627
  auto ret = worker_ptr_->Save(path, std::to_string(mode));
T
tangwei12 已提交
628 629 630 631 632 633 634 635 636
  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
637
  auto ret = worker_ptr_->Save(table_id, path, std::to_string(mode));
T
tangwei12 已提交
638 639 640 641 642 643 644
  ret.wait();
  if (ret.get() != 0) {
    LOG(ERROR) << "save model of table id: " << table_id
               << ", to path: " << path << " failed";
  }
}

645 646
void FleetWrapper::RecvAndSaveTable(const uint64_t table_id,
                                    const std::string& path) {
Z
zhaocaibei123 已提交
647
  auto ret = worker_ptr_->RecvAndSaveTable(table_id, path);
648 649 650 651 652 653
  if (ret != 0) {
    LOG(ERROR) << "save model of table id: " << table_id
               << ", to path: " << path << " failed";
  }
}

T
tangwei12 已提交
654
void FleetWrapper::PrintTableStat(const uint64_t table_id) {
Z
zhaocaibei123 已提交
655
  auto ret = worker_ptr_->PrintTableStat(table_id);
T
tangwei12 已提交
656 657 658 659 660 661 662
  ret.wait();
  int32_t err_code = ret.get();
  if (err_code == -1) {
    LOG(ERROR) << "print table stat failed";
  }
}

663
void FleetWrapper::ShrinkSparseTable(int table_id, int threshold) {
Z
zhaocaibei123 已提交
664
  auto ret = worker_ptr_->Shrink(table_id, std::to_string(threshold));
T
tangwei12 已提交
665
  ret.wait();
666 667 668 669
  int32_t err_code = ret.get();
  if (err_code == -1) {
    LOG(ERROR) << "shrink sparse table stat failed";
  }
T
tangwei12 已提交
670 671 672
}

void FleetWrapper::ClearModel() {
Z
zhaocaibei123 已提交
673
  auto ret = pserver_ptr_->_worker_ptr->Clear();
T
tangwei12 已提交
674 675 676 677
  ret.wait();
}

void FleetWrapper::ClearOneTable(const uint64_t table_id) {
Z
zhaocaibei123 已提交
678
  auto ret = pserver_ptr_->_worker_ptr->Clear(table_id);
T
tangwei12 已提交
679 680 681 682 683 684 685 686 687 688 689
  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";
690
      VLOG(3) << "prepare shrink dense batch_sum";
T
tangwei12 已提交
691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716
      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
717
  auto push_status = pserver_ptr_->_worker_ptr->PushDenseParam(
T
tangwei12 已提交
718 719 720 721 722 723 724 725 726 727 728 729
      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() {
730 731 732 733
  if (worker_ptr_.get() == nullptr) {
    VLOG(0) << "worker_ptr null, do nothing";
    return;
  }
Z
zhaocaibei123 已提交
734
  auto ret = worker_ptr_->Flush();
T
tangwei12 已提交
735
  ret.wait();
736 737 738 739
  int32_t err_code = ret.get();
  if (err_code == -1) {
    LOG(ERROR) << "Client Flush failed";
  }
T
tangwei12 已提交
740 741 742 743
}

int FleetWrapper::RegisterClientToClientMsgHandler(int msg_type,
                                                   MsgHandlerFunc handler) {
744 745
  if (worker_ptr_.get() == nullptr) {
    VLOG(0) << "FleetWrapper::Client is null";
Z
zhaocaibei123 已提交
746 747
    return -1;
  } else {
Z
zhaocaibei123 已提交
748
    return worker_ptr_->RegisteClient2ClientMsgHandler(msg_type, handler);
Z
zhaocaibei123 已提交
749
  }
T
tangwei12 已提交
750 751 752 753
}

std::future<int32_t> FleetWrapper::SendClientToClientMsg(
    int msg_type, int to_client_id, const std::string& msg) {
Z
zhaocaibei123 已提交
754
  return worker_ptr_->SendClient2ClientMsg(msg_type, to_client_id, msg);
T
tangwei12 已提交
755 756
}

Z
zhaocaibei123 已提交
757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796
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
797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829
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