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
PaddlePaddle
Paddle
提交
201e7157
P
Paddle
项目概览
PaddlePaddle
/
Paddle
1 年多 前同步成功
通知
2302
Star
20931
Fork
5422
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
1423
列表
看板
标记
里程碑
合并请求
543
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
1,423
Issue
1,423
列表
看板
标记
里程碑
合并请求
543
合并请求
543
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
提交
201e7157
编写于
7月 28, 2017
作者:
L
liaogang
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
ENH: Add comments for memory and memcpy
上级
2200ff5e
变更
2
隐藏空白更改
内联
并排
Showing
2 changed file
with
63 addition
and
3 deletion
+63
-3
paddle/memory/memcpy.h
paddle/memory/memcpy.h
+26
-0
paddle/memory/memory.h
paddle/memory/memory.h
+37
-3
未找到文件。
paddle/memory/memcpy.h
浏览文件 @
201e7157
...
...
@@ -20,13 +20,39 @@ limitations under the License. */
namespace
paddle
{
namespace
memory
{
/**
* \brief Copy memory from one place to another place.
*
* \param[in] DstPlace Destination allocation place (CPU).
* \param[in] dst Destination memory address.
* \param[in] SrcPlace Source allocation place (CPU).
* \param[in] src Source memory address.
* \param[in] num memory size in bytes to copy.
*
*/
template
<
typename
DstPlace
,
typename
SrcPlace
>
void
Copy
(
DstPlace
,
void
*
dst
,
SrcPlace
,
const
void
*
src
,
size_t
num
);
#ifndef PADDLE_ONLY_CPU
/**
* \brief Copy memory from one place to another place.
*
* \param[in] DstPlace Destination allocation place (CPU or GPU).
* \param[in] dst Destination memory address.
* \param[in] SrcPlace Source allocation place (CPU or GPU).
* \param[in] src Source memory address.
* \param[in] num memory size in bytes to copy.
* \param[in] stream CUDA stream.
*
* \note For GPU memory copy, CUDA stream need to be specified
* for asynchronously memory copy.
*
*/
template
<
typename
DstPlace
,
typename
SrcPlace
>
void
Copy
(
DstPlace
,
void
*
dst
,
SrcPlace
,
const
void
*
src
,
size_t
num
,
cudaStream_t
stream
);
#endif // PADDLE_ONLY_CPU
}
// namespace memory
...
...
paddle/memory/memory.h
浏览文件 @
201e7157
...
...
@@ -20,15 +20,49 @@ limitations under the License. */
namespace
paddle
{
namespace
memory
{
/**
* \brief Allocate memory block in one place.
*
* \param[in] place Allocation place (CPU or GPU).
* \param[in] size Allocation size.
*
* \return Allocated memory block address.
*
* \note If return nullptr, it indicates memory allocation failed
* because insufficient memory in current system. When Alloc
* function is invoked, you must check the returned memory
* address is valid or not.
*/
template
<
typename
Place
>
void
*
Alloc
(
Place
,
size_t
);
void
*
Alloc
(
Place
place
,
size_t
size
);
/**
* \brief Free memory block in one place.
*
* \param[in] place Allocation place (CPU or GPU).
* \param[in] ptr Memory block address to free.
*
*/
template
<
typename
Place
>
void
Free
(
Place
,
void
*
);
void
Free
(
Place
place
,
void
*
ptr
);
/**
* \brief Total size of used memory in one place.
*
* \param[in] place Allocation place (CPU or GPU).
*
*/
template
<
typename
Place
>
size_t
Used
(
Place
);
size_t
Used
(
Place
place
);
/**
* \brief Free memory block in one place.
*
* \note In some cases, custom deleter is used to
* deallocate the memory automatically for
* std::unique_ptr<T> in tensor.h.
*
*/
template
<
typename
T
,
typename
Place
>
class
PODDeleter
{
static_assert
(
std
::
is_pod
<
T
>::
value
,
"T must be POD"
);
...
...
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