MathFunctions.h 3.0 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

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_

T
tensor-tang 已提交
18
#ifdef PADDLE_USE_MKLML
19 20 21 22 23
#include <mkl_cblas.h>
#include <mkl_lapacke.h>
#include <mkl_vml_functions.h>
#endif

24
#if defined(PADDLE_USE_ATLAS) || defined(PADDLE_USE_VECLIB)
L
lzhao4ever 已提交
25
extern "C" {
L
liaogang 已提交
26
#include <cblas.h>
L
lzhao4ever 已提交
27 28
#include <clapack.h>
}
L
liaogang 已提交
29 30 31 32
#endif

#ifdef PADDLE_USE_OPENBLAS
#include <cblas.h>
L
lzhao4ever 已提交
33 34
#include <lapacke.h>
#endif
L
liaogang 已提交
35 36 37

#ifndef LAPACK_FOUND
extern "C" {
38
#ifndef PADDLE_USE_EIGEN_FOR_BLAS
L
liaogang 已提交
39
#include <cblas.h>
40 41 42 43 44 45
#else
typedef enum CBLAS_ORDER {
  CblasRowMajor = 101,
  CblasColMajor = 102
} CBLAS_ORDER;
#endif
L
liaogang 已提交
46 47 48 49 50 51 52 53
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);
L
liaogang 已提交
54
}
Z
zhangjinchao01 已提交
55 56
#endif

L
liaogang 已提交
57 58
#include <cmath>

Z
zhangjinchao01 已提交
59 60
namespace paddle {

61
#ifndef PADDLE_USE_EIGEN_FOR_BLAS
62 63 64 65 66 67 68 69 70 71 72 73 74 75
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);
76
#endif
77 78 79 80 81 82 83 84 85 86 87 88 89 90

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>
91 92 93 94 95 96
void axpy(const int n, const T alpha, const T* x, T* y) {
  /// y = y + alpha * x
  for (int i = 0; i < n; i++) {
    y[i] = y[i] + alpha * x[i];
  }
}
Z
zhangjinchao01 已提交
97

98
template <class T>
99 100 101 102 103
T dotProduct(const int n, const T* x, const T* y) {
  T result = static_cast<T>(0);
  for (int i = 0; i < n; i++) {
    result += x[i] * y[i];
  }
L
Liu Yiqun 已提交
104
  return result;
105
}
Z
zhangjinchao01 已提交
106

107
template <class T>
Z
zhangjinchao01 已提交
108 109
void vExp(const int n, const T* a, T* r);

110
template <class T>
Z
zhangjinchao01 已提交
111 112
void vPow(const int n, const T* a, const T b, T* r);

113
template <class T>
Z
zhangjinchao01 已提交
114 115
void vLog(const int n, const T* a, T* r);

116
template <class T>
Z
zhangjinchao01 已提交
117 118
void vAdd(const int n, const T* a, const T* b, T* r);

119
template <class T>
Z
zhangjinchao01 已提交
120 121
void vInvSqrt(const int n, const T* a, T* r);

122
template <class T>
Z
zhangjinchao01 已提交
123 124
void vLog1p(const int n, const T* a, T* r);

125
template <class T>
Z
zhangjinchao01 已提交
126 127 128 129 130
void vTanh(const int n, const T* a, T* r);

}  // namespace paddle

#endif  // MATHFUNCTIONS_H_