mul_op.h 2.4 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.

   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
Y
Yu Yang 已提交
16

Q
qijun 已提交
17
#include "paddle/operators/math/math_function.h"
Q
qijun 已提交
18

D
dongzhihong 已提交
19 20
#include "paddle/framework/eigen.h"
#include "paddle/framework/op_registry.h"
21 22 23 24

namespace paddle {
namespace operators {

D
dongzhihong 已提交
25 26 27 28 29
using Tensor = framework::Tensor;
template <typename T, int MajorType = Eigen::RowMajor,
          typename IndexType = Eigen::DenseIndex>
using EigenMatrix = framework::EigenMatrix<T, MajorType, IndexType>;

Q
qijun 已提交
30
template <typename Place, typename T>
D
dongzhihong 已提交
31
class MulKernel : public framework::OpKernel {
32
 public:
D
dongzhihong 已提交
33
  void Compute(const framework::ExecutionContext& context) const override {
D
dongzhihong 已提交
34 35 36 37 38 39 40
    auto* X = context.Input<Tensor>("X");
    auto* Y = context.Input<Tensor>("Y");
    auto* Z = context.Output<Tensor>("Out");
    Z->mutable_data<T>(context.GetPlace());
    auto* device_context =
        const_cast<platform::DeviceContext*>(context.device_context_);
    math::matmul<Place, T>(*X, false, *Y, false, 1, Z, 0, device_context);
41 42
  }
};
D
dongzhihong 已提交
43

D
dongzhihong 已提交
44 45 46 47
template <typename Place, typename T>
class MulGradKernel : public framework::OpKernel {
 public:
  void Compute(const framework::ExecutionContext& ctx) const override {
D
dongzhihong 已提交
48 49 50
    auto* X = ctx.Input<Tensor>("X");
    auto* Y = ctx.Input<Tensor>("Y");
    auto* dOut = ctx.Input<Tensor>(framework::GradVarName("Out"));
D
dongzhihong 已提交
51

D
dongzhihong 已提交
52 53
    auto* dX = ctx.Output<Tensor>(framework::GradVarName("X"));
    auto* dY = ctx.Output<Tensor>(framework::GradVarName("Y"));
D
dongzhihong 已提交
54 55
    dX->mutable_data<T>(ctx.GetPlace());
    dY->mutable_data<T>(ctx.GetPlace());
D
dongzhihong 已提交
56 57
    auto* device_context =
        const_cast<platform::DeviceContext*>(ctx.device_context_);
D
dongzhihong 已提交
58
    // dX = dOut * Y'. dX: M x K, dOut : M x N, Y : K x N
D
dongzhihong 已提交
59
    math::matmul<Place, T>(*dOut, false, *Y, true, 1, dX, 0, device_context);
D
dongzhihong 已提交
60
    // dY = X' * dOut. dY: K x N, dOut : M x N, X : M x K
D
dongzhihong 已提交
61
    math::matmul<Place, T>(*X, true, *dOut, false, 1, dY, 0, device_context);
D
dongzhihong 已提交
62 63 64
  }
};

65 66
}  // namespace operators
}  // namespace paddle