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
兽拳
Paddle
提交
2ab12ca2
P
Paddle
项目概览
兽拳
/
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
1
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
1
Issue
1
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
2ab12ca2
编写于
4月 17, 2018
作者:
Y
Yu Yang
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
Add comments and clean code
上级
a822f8dd
变更
2
隐藏空白更改
内联
并排
Showing
2 changed file
with
13 addition
and
2 deletion
+13
-2
paddle/fluid/pybind/pybind.cc
paddle/fluid/pybind/pybind.cc
+4
-2
paddle/fluid/pybind/tensor_py.h
paddle/fluid/pybind/tensor_py.h
+9
-0
未找到文件。
paddle/fluid/pybind/pybind.cc
浏览文件 @
2ab12ca2
...
...
@@ -505,13 +505,15 @@ All parameter, weight, gradient are variables in Paddle.
scope
,
local_scopes
,
allow_op_delay
);
})
.
def
(
"bcast_params"
,
&
ParallelExecutor
::
BCastParamsToGPUs
)
// NOTE: even we return a vec<Scope*>* to Python use reference policy.
// We still cannot get local_scope from this vector, since the element
// of vec<Scope*> will be freed by Python GC. We can only return Scope*
// one by one and mark them as reference.
.
def
(
"local_scopes"
,
[](
ParallelExecutor
&
self
)
->
std
::
vector
<
Scope
*>
*
{
return
&
self
.
GetLocalScopes
();
},
py
::
return_value_policy
::
reference
)
.
def
(
"local_scopes_len"
,
[](
ParallelExecutor
&
self
)
{
return
self
.
GetLocalScopes
().
size
();
})
.
def
(
"local_scope"
,
[](
ParallelExecutor
&
self
,
size_t
i
)
{
return
self
.
GetLocalScopes
()[
i
];
},
py
::
return_value_policy
::
reference
)
...
...
paddle/fluid/pybind/tensor_py.h
浏览文件 @
2ab12ca2
...
...
@@ -190,6 +190,10 @@ void PyCUDATensorSetFromArray(
static_cast
<
const
platform
::
CUDADeviceContext
*>
(
pool
.
Get
(
place
));
paddle
::
platform
::
GpuMemcpyAsync
(
dst
,
array
.
data
(),
sizeof
(
T
)
*
array
.
size
(),
cudaMemcpyHostToDevice
,
dev_ctx
->
stream
());
// NOTE: For safety, here wait the copy complete.
// It because the CPU array.data() could be destroyed after this method.
// If we make this method async, it could be copied data from a memory buffer
// that has been freed.
dev_ctx
->
Wait
();
}
...
...
@@ -217,6 +221,11 @@ void PyCUDATensorSetFromArray(
paddle
::
platform
::
GpuMemcpyAsync
(
dst
,
array
.
data
(),
sizeof
(
uint16_t
)
*
array
.
size
(),
cudaMemcpyHostToDevice
,
dev_ctx
->
stream
());
// NOTE: For safety, here wait the copy complete.
// It because the CPU array.data() could be destroyed after this method.
// If we make this method async, it could be copied data from a memory buffer
// that has been freed.
dev_ctx
->
Wait
();
}
template
<
typename
T
>
...
...
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