math_function.h 2.2 KB
Newer Older
Q
qijun 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 27 28 29 30 31 32 33 34 35 36 37 38 39
/* 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

#ifdef PADDLE_USE_MKLML
#include <mkl_cblas.h>
#include <mkl_lapacke.h>
#include <mkl_vml_functions.h>
#endif

#ifdef PADDLE_USE_MKL
#include <mkl.h>
#include <mkl_lapacke.h>
#endif

#ifdef PADDLE_USE_ATLAS
extern "C" {
#include <cblas.h>
#include <clapack.h>
}
#endif

#ifdef PADDLE_USE_OPENBLAS
#include <cblas.h>
#include <lapacke.h>
#endif

Q
qijun 已提交
40 41 42
#ifndef LAPACK_FOUND
extern "C" {
#include <cblas.h>
Q
qijun 已提交
43 44 45 46 47 48 49 50
int LAPACKE_sgetrf(int matrix_layout, int m, int n, float* a, int lda,
                   int* ipiv);
int LAPACKE_dgetrf(int matrix_layout, int m, int n, double* a, int lda,
                   int* ipiv);
int LAPACKE_sgetri(int matrix_layout, int n, float* a, int lda,
                   const int* ipiv);
int LAPACKE_dgetri(int matrix_layout, int n, double* a, int lda,
                   const int* ipiv);
Q
qijun 已提交
51 52 53
}
#endif

Q
qijun 已提交
54
#include <cmath>
Q
qijun 已提交
55
#include "paddle/framework/tensor.h"
Q
qijun 已提交
56
#include "paddle/platform/device_context.h"
Q
qijun 已提交
57
#include "paddle/platform/enforce.h"
Q
qijun 已提交
58 59 60 61 62 63

namespace paddle {
namespace operators {
namespace math {

template <typename Place, typename T>
Q
qijun 已提交
64 65 66 67
void gemm(const CBLAS_TRANSPOSE transA, const CBLAS_TRANSPOSE transB,
          const int M, const int N, const int K, const T alpha, const T* A,
          const int lda, const T* B, const int ldb, const T beta, T* C,
          const int ldc, platform::DeviceContext* context);
Q
qijun 已提交
68

Q
qijun 已提交
69
// matrix multiply with continuous memory
Q
qijun 已提交
70
template <typename Place, typename T>
Q
qijun 已提交
71 72 73
void matmul(const framework::Tensor& in1, bool in1_T,
            const framework::Tensor& in2, bool in2_T, float alpha,
            framework::Tensor* out, float beta,
Q
qijun 已提交
74 75
            platform::DeviceContext* context);

Q
qijun 已提交
76 77 78
}  // namespace math
}  // namespace operators
}  // namespace paddle