MathFunctions.h 2.6 KB
Newer Older
1
/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.
Z
zhangjinchao01 已提交
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

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. */

#ifndef MATHFUNCTIONS_H_
#define MATHFUNCTIONS_H_

#ifdef PADDLE_USE_MKL
#include <mkl.h>
L
lzhao4ever 已提交
20
#include <mkl_lapacke.h>
L
liaogang 已提交
21 22
#endif

L
lzhao4ever 已提交
23 24
#ifdef PADDLE_USE_ATLAS
extern "C" {
L
liaogang 已提交
25
#include <cblas.h>
L
lzhao4ever 已提交
26 27
#include <clapack.h>
}
L
liaogang 已提交
28 29 30 31
#endif

#ifdef PADDLE_USE_OPENBLAS
#include <cblas.h>
L
lzhao4ever 已提交
32 33
#include <lapacke.h>
#endif
L
liaogang 已提交
34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50

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

#ifndef LAPACK_FOUND
extern "C" {
int LAPACKE_sgetrf(int matrix_layout, int m, int n,
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);
}
Z
zhangjinchao01 已提交
51 52
#endif

L
liaogang 已提交
53 54
#include <cmath>

Z
zhangjinchao01 已提交
55 56
namespace paddle {

57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84
template <class T>
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);

template <class T>
int getrf(const CBLAS_ORDER Order,
          const int M,
          const int N,
          T* A,
          const int lda,
          int* ipiv);

template <class T>
int getri(
    const CBLAS_ORDER Order, const int N, T* A, const int lda, const int* ipiv);

template <class T>
Z
zhangjinchao01 已提交
85 86
void axpy(const int n, const T alpha, const T* x, T* y);

87
template <class T>
Z
zhangjinchao01 已提交
88 89
T dotProduct(const int n, const T* x, const T* y);

90
template <class T>
Z
zhangjinchao01 已提交
91 92
void vExp(const int n, const T* a, T* r);

93
template <class T>
Z
zhangjinchao01 已提交
94 95
void vPow(const int n, const T* a, const T b, T* r);

96
template <class T>
Z
zhangjinchao01 已提交
97 98
void vLog(const int n, const T* a, T* r);

99
template <class T>
Z
zhangjinchao01 已提交
100 101
void vAdd(const int n, const T* a, const T* b, T* r);

102
template <class T>
Z
zhangjinchao01 已提交
103 104
void vInvSqrt(const int n, const T* a, T* r);

105
template <class T>
Z
zhangjinchao01 已提交
106 107
void vLog1p(const int n, const T* a, T* r);

108
template <class T>
Z
zhangjinchao01 已提交
109 110 111 112 113
void vTanh(const int n, const T* a, T* r);

}  // namespace paddle

#endif  // MATHFUNCTIONS_H_