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
Mr.Vain
Mace
提交
9b78e12e
Mace
项目概览
Mr.Vain
/
Mace
与 Fork 源项目一致
Fork自
Xiaomi / Mace
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
DevOps
流水线
流水线任务
计划
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
Mace
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
DevOps
DevOps
流水线
流水线任务
计划
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
流水线任务
提交
Issue看板
体验新版 GitCode,发现更多精彩内容 >>
提交
9b78e12e
编写于
10月 23, 2018
作者:
L
Liangliang He
浏览文件
操作
浏览文件
下载
差异文件
Merge branch 'fix_deconv_cpu_thread-safety' into 'master'
fix deconv cpu thread safty See merge request !844
上级
e8a777b4
7e2e776f
变更
3
隐藏空白更改
内联
并排
Showing
3 changed file
with
7 addition
and
7 deletion
+7
-7
mace/kernels/arm/deconv_2d_neon_3x3.cc
mace/kernels/arm/deconv_2d_neon_3x3.cc
+2
-2
mace/kernels/arm/deconv_2d_neon_4x4.cc
mace/kernels/arm/deconv_2d_neon_4x4.cc
+2
-2
mace/kernels/deconv_2d.h
mace/kernels/deconv_2d.h
+3
-3
未找到文件。
mace/kernels/arm/deconv_2d_neon_3x3.cc
浏览文件 @
9b78e12e
...
...
@@ -293,11 +293,11 @@ void Deconv2dNeonK3x3S2(const float *input,
const
index_t
outw
=
out_shape
[
3
];
const
index_t
out_img_size
=
outh
*
outw
;
#pragma omp parallel for collapse(
3
)
#pragma omp parallel for collapse(
2
)
for
(
index_t
b
=
0
;
b
<
out_shape
[
0
];
++
b
)
{
for
(
index_t
oc
=
0
;
oc
<
outch
;
++
oc
)
{
float
*
out_base
=
output
+
(
b
*
outch
+
oc
)
*
out_img_size
;
for
(
index_t
ic
=
0
;
ic
<
inch
;
++
ic
)
{
float
*
out_base
=
output
+
(
b
*
outch
+
oc
)
*
out_img_size
;
const
float
*
input_base
=
input
+
(
b
*
inch
+
ic
)
*
h
*
w
;
const
float
*
kernel_base
=
filter
+
(
oc
*
inch
+
ic
)
*
9
;
const
float
*
in
=
input_base
;
...
...
mace/kernels/arm/deconv_2d_neon_4x4.cc
浏览文件 @
9b78e12e
...
...
@@ -386,11 +386,11 @@ void Deconv2dNeonK4x4S2(const float *input,
const
index_t
outch
=
out_shape
[
1
];
const
index_t
out_img_size
=
outh
*
outw
;
#pragma omp parallel for collapse(
3
)
#pragma omp parallel for collapse(
2
)
for
(
int
b
=
0
;
b
<
out_shape
[
0
];
++
b
)
{
for
(
int
p
=
0
;
p
<
outch
;
p
++
)
{
float
*
out_base
=
output
+
(
b
*
outch
+
p
)
*
out_img_size
;
for
(
int
q
=
0
;
q
<
inch
;
q
++
)
{
float
*
out_base
=
output
+
(
b
*
outch
+
p
)
*
out_img_size
;
const
float
*
input_base
=
input
+
(
b
*
inch
+
q
)
*
h
*
w
;
const
float
*
kernel_base
=
filter
+
(
p
*
inch
+
q
)
*
16
;
const
float
*
in
=
input_base
;
...
...
mace/kernels/deconv_2d.h
浏览文件 @
9b78e12e
...
...
@@ -214,13 +214,13 @@ struct Deconv2dFunctor<DeviceType::CPU, float>: Deconv2dFunctorBase {
const
index_t
out_channels
=
out_shape
[
1
];
const
index_t
in_channels
=
in_shape
[
1
];
#pragma omp parallel for collapse(
4
)
#pragma omp parallel for collapse(
2
)
for
(
int
b
=
0
;
b
<
batch
;
++
b
)
{
for
(
int
oc
=
0
;
oc
<
out_channels
;
++
oc
)
{
float
*
out_base
=
output
+
(
b
*
out_channels
+
oc
)
*
out_img_size
;
for
(
int
i
=
0
;
i
<
in_height
;
++
i
)
{
for
(
int
j
=
0
;
j
<
in_width
;
++
j
)
{
float
*
out_base
=
output
+
(
b
*
out_channels
+
oc
)
*
out_img_size
;
const
index_t
out_offset
=
i
*
strides
[
0
]
*
out_width
+
j
*
strides
[
1
];
for
(
int
ic
=
0
;
ic
<
in_channels
;
++
ic
)
{
...
...
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