downpour_worker.cc

/* Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License. */

#include "paddle/fluid/framework/device_worker.h"
#include "paddle/fluid/platform/cpu_helper.h"

namespace paddle {
namespace framework {

void DownpourWorker::Initilize(const TrainerDesc& desc) {
  param_ = desc.downpour_param();

  for (size_t i = 0; i < param_.sparse_table_size(); ++i) {
    uint64_t table_id =
        static_cast<uint64_t>(param_.sparse_table(i).table_id());
    TableParameter table = param_.sparse_table(i);
    sparse_key_names_[table_id].resize(table.sparse_key_name_size());
    for (size_t j = 0; j < table.sparse_key_name_size(); ++j) {
      sparse_key_names_[table_id][j] = table.sparse_key_name(j);
    }
    sparse_value_names_[table_id].resize(table.sparse_value_name_size());
    for (size_t j = 0; j < table.sparse_value_name_size(); ++j) {
      sparse_value_names_[table_id][j] = table.sparse_value_name(j);
    }
    sparse_grad_names_[table_id].resize(table.sparse_grad_name_size());
    for (size_t j = 0; j < table.sparse_grad_name_size(); ++j) {
      sparse_grad_names_[table_id][j] = table.sparse_grad_name(j);
    }
  }

  for (size_t i = 0; i < param_.dense_table_size(); ++i) {
    uint64_t table_id = static_cast<uint64_t>(param_.dense_table(i).table_id());
    auto table = param_.dense_table(i);
    dense_value_names_[table_id].resize(table.dense_value_name_size());
    for (size_t j = 0; j < table.dense_value_name_size(); ++j) {
      dense_value_names_[table_id][j] = table.dense_value_name(j);
    }
    dense_grad_names_[table_id].resize(table.dense_grad_name_size());
    for (size_t j = 0; j < table.dense_grad_name_size(); ++j) {
      dense_grad_names_[table_id][j] = table.dense_grad_name(j);
    }
  }

  skip_ops_.resize(param_.skip_ops_size());
  for (size_t i = 0; i < param_.skip_ops_size(); ++i) {
    skip_ops_[i] = param_.skip_ops(i);
  }

  label_var_name_ = param_.label_var_name();
}

void DownpourWorker::CollectLabelInfo(size_t table_id) {
  auto& feature = features_[table_id];
  auto& feature_label = feature_labels_[table_id];
  feature_label.resize(feature.size());
  Variable* var = thread_scope_->FindVar(label_var_name_);
  LoDTensor* tensor = var->GetMutable<LoDTensor>();
  int64_t* label_ptr = tensor->data<int64_t>();

  int global_index = 0;
  for (size_t i = 0; i < sparse_key_names_[table_id].size(); ++i) {
    Variable* fea_var = thread_scope_->FindVar(sparse_key_names_[table_id][i]);
    LoDTensor* tensor = fea_var->GetMutable<LoDTensor>();
    int64_t* ids = tensor->data<int64_t>();
    int fea_idx = 0;
    // tensor->lod()[0].size() == batch_size + 1
    for (auto ins_idx = 0u; ins_idx < tensor->lod()[0].size() - 1; ++ins_idx) {
      for (; fea_idx < tensor->lod()[0][ins_idx]; ++fea_idx) {
        // should be skipped feasign defined in protobuf
        if (ids[fea_idx] == 0u) {
          continue;
        }
        feature_label[global_index++] = static_cast<float>(label_ptr[ins_idx]);
      }
    }
  }
  CHECK(global_index == feature.size())
      << "expect fea info size:" << feature.size() << " real:" << global_index;
}

void DownpourWorker::FillSparseValue(size_t table_idx) {
  auto table = param_.sparse_table(table_idx);

  uint64_t table_id =
      static_cast<uint64_t>(param_.sparse_table(table_idx).table_id());
  auto& fea_value = feature_values_[table_id];
  auto fea_idx = 0u;

  std::vector<float> init_value(table.emb_dim());
  for (size_t i = 0; i < sparse_key_names_[table_id].size(); ++i) {
    std::string slot_name = sparse_key_names_[table_id][i];
    std::string emb_slot_name = sparse_value_names_[table_id][i];
    Variable* var = thread_scope_->FindVar(slot_name);
    LoDTensor* tensor = var->GetMutable<LoDTensor>();
    int64_t* ids = tensor->data<int64_t>();
    int len = tensor->numel();
    Variable* var_emb = thread_scope_->FindVar(emb_slot_name);
    LoDTensor* tensor_emb = var_emb->GetMutable<LoDTensor>();
    float* ptr = tensor_emb->mutable_data<float>({len, table.emb_dim()},
                                                 platform::CPUPlace());
    memset(ptr, 0, sizeof(float) * len * table.emb_dim());
    auto& tensor_lod = tensor->lod()[0];
    LoD data_lod{tensor_lod};
    tensor_emb->set_lod(data_lod);
    for (auto index = 0u; index < len; ++index) {
      if (ids[index] == 0u) {
        memcpy(ptr + table.emb_dim() * index, init_value.data() + 2,
               sizeof(float) * table.emb_dim());
        continue;
      }
      memcpy(ptr + table.emb_dim() * index, fea_value[fea_idx].data() + 2,
             sizeof(float) * table.emb_dim());
      fea_idx++;
    }
  }
}

void DownpourWorker::TrainFiles() {
  platform::SetNumThreads(1);
  thread_reader_->Start();
  int batch_cnt = 0;
  int cur_batch;
  while ((cur_batch = thread_reader_->Next()) > 0) {
    // pull sparse here
    for (size_t i = 0; i < param_.sparse_table_size(); ++i) {
      uint64_t tid = static_cast<uint64_t>(param_.sparse_table(i).table_id());
      fleet_ptr_->PullSparseVarsSync(
          *thread_scope_, tid, sparse_key_names_[tid], &features_[tid],
          &feature_values_[tid], param_.sparse_table(i).fea_dim());
      CollectLabelInfo(i);
      FillSparseValue(i);
    }

    // do computation here
    for (auto& op : ops_) {
      op->Run(*thread_scope_, place_);
    }

    // push gradients here
    for (size_t i = 0; i < param_.sparse_table_size(); ++i) {
      uint64_t tid = static_cast<uint64_t>(param_.sparse_table(i).table_id());
      fleet_ptr_->PushSparseVarsWithLabelAsync(
          *thread_scope_, tid, features_[tid], feature_labels_[tid],
          sparse_key_names_[tid], sparse_grad_names_[tid],
          param_.sparse_table(i).emb_dim(), &feature_grads_[tid],
          &push_sparse_status_);
    }

    for (size_t i = 0; i < param_.dense_table_size(); ++i) {
      uint64_t tid = static_cast<uint64_t>(param_.dense_table(i).table_id());
      fleet_ptr_->PushDenseVarsAsync(
          *thread_scope_, tid, dense_grad_names_[tid], &push_sparse_status_);
    }

    // the following code should be more precise and clean
    // TODO(guru4elephant)
    int32_t tmp_push_dense_wait_times = -1;
    int32_t tmp_push_sparse_wait_times = -1;
    static uint32_t push_dense_wait_times =
        static_cast<uint32_t>(tmp_push_dense_wait_times);
    static uint32_t push_sparse_wait_times =
        static_cast<uint32_t>(tmp_push_sparse_wait_times);

    if (push_dense_status_.size() >= push_dense_wait_times) {
      for (auto& t : push_dense_status_) {
        t.wait();
      }
      push_dense_status_.resize(0);
    }

    if (tmp_push_dense_wait_times == -1) {
      push_dense_status_.resize(0);
    }

    if (push_sparse_status_.size() >= push_sparse_wait_times) {
      for (auto& t : push_sparse_status_) {
        t.wait();
      }
      push_sparse_status_.resize(0);
    }

    if (tmp_push_sparse_wait_times == -1) {
      push_sparse_status_.resize(0);
    }

    for (size_t i = 0; i < param_.dense_table_size(); ++i) {
      uint64_t tid = static_cast<uint64_t>(param_.dense_table(i).table_id());
      pull_dense_worker_->IncreaseThreadVersion(thread_id_, tid);
    }
    thread_scope_->DropKids();
    ++batch_cnt;
  }
}

}  // end namespace framework
}  // end namespace paddle