-
Notifications
You must be signed in to change notification settings - Fork 5.6k
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge pull request #3549 from hedaoyuan/convolution
Use EigenBlasGemm improve convolution computing performance in ARMv7 environment.
- Loading branch information
Showing
9 changed files
with
263 additions
and
110 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,91 @@ | ||
/* 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. */ | ||
|
||
#include <glog/logging.h> | ||
#include "unsupported/Eigen/CXX11/Tensor" | ||
|
||
namespace paddle { | ||
|
||
template <class T> | ||
struct EigenBlasGemm { | ||
typedef Eigen::TensorMap<Eigen::Tensor<T, 2, Eigen::RowMajor, int>, | ||
Eigen::Aligned> | ||
Matrix; | ||
|
||
static void compute(const bool transA, | ||
const bool 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) { | ||
Eigen::array<int, 2> sizeA; | ||
if (transA) { | ||
sizeA[0] = K; | ||
sizeA[1] = M; | ||
CHECK_EQ(M, lda); | ||
} else { | ||
sizeA[0] = M; | ||
sizeA[1] = K; | ||
CHECK_EQ(K, lda); | ||
} | ||
Eigen::array<int, 2> sizeB; | ||
if (transB) { | ||
sizeB[0] = N; | ||
sizeB[1] = K; | ||
CHECK_EQ(K, ldb); | ||
} else { | ||
sizeB[0] = K; | ||
sizeB[1] = N; | ||
CHECK_EQ(N, ldb); | ||
} | ||
Eigen::array<int, 2> sizeC; | ||
sizeC[0] = M; | ||
sizeC[1] = N; | ||
CHECK_EQ(N, ldc); | ||
|
||
const Matrix a(const_cast<T*>(A), sizeA); | ||
const Matrix b(const_cast<T*>(B), sizeB); | ||
Matrix c(C, sizeC); | ||
|
||
typedef typename Eigen::Tensor<T, 2>::DimensionPair DimPair; | ||
Eigen::array<DimPair, 1> dims; | ||
dims[0] = DimPair(1, 0); | ||
dims[0].first = transA ? 0 : 1; | ||
dims[0].second = transB ? 1 : 0; | ||
|
||
Eigen::DefaultDevice device; | ||
if (alpha == T(1) && beta == T(0)) { | ||
c.device(device) = a.contract(b, dims); | ||
} else if (alpha == T(1) && beta == T(1)) { | ||
c.device(device) += a.contract(b, dims); | ||
} else { | ||
c.device(device) = alpha * a.contract(b, dims) + beta * c; | ||
} | ||
} | ||
}; | ||
|
||
#ifdef PADDLE_TYPE_DOUBLE | ||
template class EigenBlasGemm<double>; | ||
#else | ||
template class EigenBlasGemm<float>; | ||
#endif | ||
|
||
} // namespace paddle |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,90 @@ | ||
/* 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. */ | ||
|
||
#include "GemmFunctor.h" | ||
#include "paddle/math/MathFunctions.h" | ||
|
||
namespace paddle { | ||
|
||
template <class T> | ||
struct BlasGemm<DEVICE_TYPE_CPU, T> { | ||
static void compute(const bool transA, | ||
const bool 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) { | ||
#ifdef PADDLE_USE_EIGEN_FOR_BLAS | ||
EigenBlasGemm<T>::compute( | ||
transA, transB, M, N, K, alpha, A, lda, B, ldb, beta, C, ldc); | ||
#else | ||
gemm<T>(transA == false ? CblasNoTrans : CblasTrans, | ||
transB == false ? CblasNoTrans : CblasTrans, | ||
M, | ||
N, | ||
K, | ||
alpha, | ||
A, | ||
lda, | ||
B, | ||
ldb, | ||
beta, | ||
C, | ||
ldc); | ||
#endif | ||
} | ||
}; | ||
|
||
template <class T> | ||
struct BlasGemm<DEVICE_TYPE_GPU, T> { | ||
static void compute(const bool transA, | ||
const bool 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) { | ||
hl_matrix_mul((T*)A, | ||
transA == false ? HPPL_OP_N : HPPL_OP_T, | ||
(T*)B, | ||
transB == false ? HPPL_OP_N : HPPL_OP_T, | ||
C, | ||
M, | ||
N, | ||
K, | ||
alpha, | ||
beta, | ||
lda, | ||
ldb, | ||
ldc); | ||
} | ||
}; | ||
|
||
template class BlasGemm<DEVICE_TYPE_CPU, real>; | ||
template class BlasGemm<DEVICE_TYPE_GPU, real>; | ||
|
||
} // namespace paddle |
Oops, something went wrong.