Skip to content
体验新版
项目
组织
正在加载...
登录
切换导航
打开侧边栏
飞不过沧海
Paddle
提交
9e7da679
P
Paddle
项目概览
飞不过沧海
/
Paddle
与 Fork 源项目一致
Fork自
PaddlePaddle / Paddle
通知
1
Star
1
Fork
0
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
0
列表
看板
标记
里程碑
合并请求
0
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
P
Paddle
项目概览
项目概览
详情
发布
仓库
仓库
文件
提交
分支
标签
贡献者
分支图
比较
Issue
0
Issue
0
列表
看板
标记
里程碑
合并请求
0
合并请求
0
Pages
分析
分析
仓库分析
DevOps
Wiki
0
Wiki
成员
成员
收起侧边栏
关闭侧边栏
动态
分支图
创建新Issue
提交
Issue看板
体验新版 GitCode,发现更多精彩内容 >>
提交
9e7da679
编写于
9月 15, 2017
作者:
H
hedaoyuan
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
Add a neon::relu.
上级
c86e7e2a
变更
3
显示空白变更内容
内联
并排
Showing
3 changed file
with
86 addition
and
0 deletion
+86
-0
paddle/math/BaseMatrix.cu
paddle/math/BaseMatrix.cu
+8
-0
paddle/math/NEONFunctions.cpp
paddle/math/NEONFunctions.cpp
+55
-0
paddle/math/NEONFunctions.h
paddle/math/NEONFunctions.h
+23
-0
未找到文件。
paddle/math/BaseMatrix.cu
浏览文件 @
9e7da679
...
@@ -17,6 +17,7 @@ limitations under the License. */
...
@@ -17,6 +17,7 @@ limitations under the License. */
#include <cmath>
#include <cmath>
#include "BaseMatrix.h"
#include "BaseMatrix.h"
#include "MathFunctions.h"
#include "MathFunctions.h"
#include "NEONFunctions.h"
#include "SIMDFunctions.h"
#include "SIMDFunctions.h"
#include "hl_matrix_apply.cuh"
#include "hl_matrix_apply.cuh"
#include "hl_matrix_base.cuh"
#include "hl_matrix_base.cuh"
...
@@ -666,6 +667,13 @@ void BaseMatrixT<T>::relu(BaseMatrixT& b) {
...
@@ -666,6 +667,13 @@ void BaseMatrixT<T>::relu(BaseMatrixT& b) {
applyBinary
(
binary
::
Relu
<
T
>
(),
b
);
applyBinary
(
binary
::
Relu
<
T
>
(),
b
);
}
}
#if defined(__ARM_NEON__) || defined(__ARM_NEON)
template
<
>
void
BaseMatrixT
<
float
>::
relu
(
BaseMatrixT
&
b
)
{
neon
::
relu
(
data_
,
b
.
data_
,
height_
*
width_
);
}
#endif
DEFINE_MATRIX_BINARY_OP
(
ReluDerivative
,
a
*=
(
b
>
0.0
f
?
1.0
f
:
0.0
f
));
DEFINE_MATRIX_BINARY_OP
(
ReluDerivative
,
a
*=
(
b
>
0.0
f
?
1.0
f
:
0.0
f
));
template
<
class
T
>
template
<
class
T
>
void
BaseMatrixT
<
T
>::
reluDerivative
(
BaseMatrixT
&
b
)
{
void
BaseMatrixT
<
T
>::
reluDerivative
(
BaseMatrixT
&
b
)
{
...
...
paddle/math/NEONFunctions.cpp
0 → 100644
浏览文件 @
9e7da679
/* 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. */
#if defined(__ARM_NEON__) || defined(__ARM_NEON)
#include "NEONFunctions.h"
#include <arm_neon.h>
namespace
paddle
{
namespace
neon
{
// b[i] = a[i] > 0.0f ? a[i] : 0.0f
void
relu
(
const
float
*
a
,
float
*
b
,
int
len
)
{
int
offset
=
len
%
16
;
float32x4_t
ma0
,
ma1
,
ma2
,
ma3
;
float32x4_t
mb0
,
mb1
,
mb2
,
mb3
;
float32x4_t
zero
=
vdupq_n_f32
(
0.
f
);
for
(
int
k
=
0
;
k
<
len
/
16
;
k
++
,
a
+=
16
,
b
+=
16
)
{
ma0
=
vld1q_f32
(
a
);
ma1
=
vld1q_f32
(
a
+
4
);
ma2
=
vld1q_f32
(
a
+
8
);
ma3
=
vld1q_f32
(
a
+
12
);
mb0
=
vmaxq_f32
(
ma0
,
zero
);
mb1
=
vmaxq_f32
(
ma1
,
zero
);
mb2
=
vmaxq_f32
(
ma2
,
zero
);
mb3
=
vmaxq_f32
(
ma3
,
zero
);
vst1q_f32
(
b
,
mb0
);
vst1q_f32
(
b
+
4
,
mb1
);
vst1q_f32
(
b
+
8
,
mb2
);
vst1q_f32
(
b
+
12
,
mb3
);
}
for
(
int
i
=
0
;
i
<
offset
;
i
++
)
{
b
[
i
]
=
a
[
i
]
>
0.0
f
?
a
[
i
]
:
0.0
f
;
}
}
}
// namespace neon
}
// namespace paddle
#endif
paddle/math/NEONFunctions.h
0 → 100644
浏览文件 @
9e7da679
/* 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
namespace
paddle
{
namespace
neon
{
void
relu
(
const
float
*
a
,
float
*
b
,
int
len
);
}
// namespace neon
}
// namespace paddle
编辑
预览
Markdown
is supported
0%
请重试
或
添加新附件
.
添加附件
取消
You are about to add
0
people
to the discussion. Proceed with caution.
先完成此消息的编辑!
取消
想要评论请
注册
或
登录