paddle_pass_builder.h 4.4 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64
// 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.

#pragma once

#include <sstream>
#include <string>
#include <vector>

namespace paddle {
/*
 * This is a pass builder based on string. It is part of inference API.
 */
class PaddlePassBuilder {
 public:
  explicit PaddlePassBuilder(const std::vector<std::string> &passes)
      : passes_(passes) {}

  void AppendPass(const std::string &pass_type);

  void InsertPass(size_t idx, const std::string &pass_type);

  // Delete the `idx`-th pass.
  void DeletePass(size_t idx);

  // Delete all the passes that has type `pass_type`.
  void DeletePass(const std::string &pass_type);

  // Visualize the computation graph after each pass by generating a DOT
  // language file, one can draw them with the Graphviz toolkit.
  void TurnOnDebug();

  // Human-readible information.
  std::string DebugString();

  const std::vector<std::string> &AllPasses() const { return passes_; }

 protected:
  std::vector<std::string> passes_;
};

/*
 * Pass strategy to help control the IR passes.
 */
class PassStrategy : public PaddlePassBuilder {
 public:
  explicit PassStrategy(const std::vector<std::string> &passes)
      : PaddlePassBuilder(passes) {}

  // The MKLDNN control exists in both CPU and GPU mode, because there can be
  // still some CPU kernels running in CPU mode.
  virtual void EnableMKLDNN() = 0;

65 66
  bool use_gpu() const { return use_gpu_; }

67
  virtual ~PassStrategy() = default;
68 69 70

 protected:
  bool use_gpu_{false};
71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93
};

/*
 * The CPU passes controller, it is used in AnalysisPredictor with CPU mode.
 */
class CpuPassStrategy : public PassStrategy {
 public:
  CpuPassStrategy() : PassStrategy({}) {
    // NOTE the large fusions should be located in the front, so that they will
    // not be damaged by smaller ones.
    passes_.assign({
        "infer_clean_graph_pass",         //
        "attention_lstm_fuse_pass",       //
        "seqconv_eltadd_relu_fuse_pass",  //
        // "embedding_fc_lstm_fuse_pass", //
        "fc_lstm_fuse_pass",             //
        "mul_lstm_fuse_pass",            //
        "fc_gru_fuse_pass",              //
        "mul_gru_fuse_pass",             //
        "seq_concat_fc_fuse_pass",       //
        "fc_fuse_pass",                  //
        "conv_bn_fuse_pass",             //
        "conv_eltwiseadd_bn_fuse_pass",  //
94
        "is_test_pass",                  //
95
    });
96
    use_gpu_ = false;
97 98 99 100
  }

  virtual ~CpuPassStrategy() = default;

101
  void EnableMKLDNN() override {
102 103 104 105 106
// TODO(Superjomn) Consider the way to mix CPU with GPU.
#ifdef PADDLE_WITH_MKLDNN
    passes_.insert(passes_.begin(), "mkldnn_placement_pass");

    for (auto &pass :
107 108 109 110
         std::vector<std::string>({"depthwise_conv_mkldnn_pass",    //
                                   "conv_bias_mkldnn_fuse_pass",    //
                                   "conv3d_bias_mkldnn_fuse_pass",  //
                                   "conv_relu_mkldnn_fuse_pass",    //
111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126
                                   "conv_elementwise_add_mkldnn_fuse_pass"})) {
      passes_.push_back(pass);
    }
#endif
  }

  CpuPassStrategy(const CpuPassStrategy &other) : PassStrategy(other.passes_) {}
};

/*
 * The GPU passes strategy, it is used in
 */
class GpuPassStrategy : public PassStrategy {
 public:
  GpuPassStrategy() : PassStrategy({}) {
    passes_.assign({
N
nhzlx 已提交
127 128 129 130 131 132 133
        "infer_clean_graph_pass",                    //
        "conv_affine_channel_fuse_pass",             //
        "conv_eltwiseadd_affine_channel_fuse_pass",  //
        "conv_bn_fuse_pass",                         //
        "conv_elementwise_add_act_fuse_pass",        //
        "conv_elementwise_add2_act_fuse_pass",       //
        "conv_elementwise_add_fuse_pass",            //
134
    });
135 136

    use_gpu_ = true;
137 138 139
  }

  GpuPassStrategy(const GpuPassStrategy &other)
140 141 142
      : PassStrategy(other.AllPasses()) {
    use_gpu_ = true;
  }
143

144
  void EnableMKLDNN() override;
145 146 147 148 149

  virtual ~GpuPassStrategy() = default;
};

}  // namespace paddle