gru.h 3.3 KB
Newer Older
T
tensor-tang 已提交
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
/* 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 <string>
#include "glog/logging.h"
#include "paddle/fluid/operators/jit/gen/act.h"
#include "paddle/fluid/operators/jit/gen/jitcode.h"

namespace paddle {
namespace operators {
namespace jit {
namespace gen {

T
tensor-tang 已提交
27
class GRUJitCode : public VActFunc {
T
tensor-tang 已提交
28 29 30
 public:
  explicit GRUJitCode(int id, const gru_attr_t& attr, size_t code_size,
                      void* code_ptr = nullptr)
T
tensor-tang 已提交
31
      : VActFunc(code_size, code_ptr), id_(id), num_(attr.d) {
T
tensor-tang 已提交
32
    auto typeExchange = [](KernelType type) -> gen::operand_type {
T
tensor-tang 已提交
33
      if (type == KernelType::kVSigmoid) {
34
        return operand_type::SIGMOID;
T
tensor-tang 已提交
35
      } else if (type == KernelType::kVRelu) {
36
        return operand_type::RELU;
T
tensor-tang 已提交
37
      } else if (type == KernelType::kVTanh) {
38
        return operand_type::TANH;
T
tensor-tang 已提交
39
      } else if (type == KernelType::kVIdentity) {
40
        return operand_type::IDENTITY;
T
tensor-tang 已提交
41
      } else {
42 43
        PADDLE_THROW(platform::errors::Unimplemented(
            "Do not support jit::KernelType code: %d.", type));
T
tensor-tang 已提交
44
      }
45
      return operand_type::IDENTITY;
T
tensor-tang 已提交
46 47 48 49 50 51 52
    };
    act_gate_ = typeExchange(attr.act_gate);
    act_cand_ = typeExchange(attr.act_cand);

    this->genCode();
  }

T
tensor-tang 已提交
53
  std::string name() const override {
T
tensor-tang 已提交
54 55 56 57 58 59 60 61 62 63
    std::string base = "GRUJitCode";
    if (id_ == 0) {
      base += "_H1";
    } else if (id_ == 1) {
      base += "_HtPart1";
    } else if (id_ == 2) {
      base += "_HtPart2";
    }
    auto AddTypeStr = [&](operand_type type) {
      switch (type) {
64
        case operand_type::RELU:
T
tensor-tang 已提交
65 66
          base += "_Relu";
          break;
67
        case operand_type::EXP:
T
tensor-tang 已提交
68 69
          base += "_Exp";
          break;
70
        case operand_type::SIGMOID:
T
tensor-tang 已提交
71 72
          base += "_Sigmoid";
          break;
73
        case operand_type::TANH:
T
tensor-tang 已提交
74 75
          base += "_Tanh";
          break;
76
        case operand_type::IDENTITY:
T
tensor-tang 已提交
77 78 79 80 81 82 83 84
          base += "_Identity";
          break;
        default:
          break;
      }
    };
    AddTypeStr(act_gate_);
    AddTypeStr(act_cand_);
T
tensor-tang 已提交
85
    return base;
T
tensor-tang 已提交
86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114
  }
  void genCode() override;

 protected:
  int id_;
  int num_;
  operand_type act_gate_;
  operand_type act_cand_;
  reg64_t param1{abi_param1};
};

#define DECLARE_GRU_JITCODE(name, id)                                \
  class name##JitCode : public GRUJitCode {                          \
   public:                                                           \
    explicit name##JitCode(const gru_attr_t& attr, size_t code_size, \
                           void* code_ptr = nullptr)                 \
        : GRUJitCode(id, attr, code_size, code_ptr) {}               \
  };

DECLARE_GRU_JITCODE(GRUH1, 0);
DECLARE_GRU_JITCODE(GRUHtPart1, 1);
DECLARE_GRU_JITCODE(GRUHtPart2, 2);

#undef DECLARE_GRU_JITCODE

}  // namespace gen
}  // namespace jit
}  // namespace operators
}  // namespace paddle