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
OPTHREE
Paddle
提交
7bb18433
P
Paddle
项目概览
OPTHREE
/
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
7bb18433
编写于
3月 31, 2018
作者:
F
fengjiayi
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
refine code
上级
53fa7cb9
变更
1
隐藏空白更改
内联
并排
Showing
1 changed file
with
44 addition
and
44 deletion
+44
-44
paddle/fluid/operators/reader/create_double_buffer_reader_op.cc
.../fluid/operators/reader/create_double_buffer_reader_op.cc
+44
-44
未找到文件。
paddle/fluid/operators/reader/create_double_buffer_reader_op.cc
浏览文件 @
7bb18433
...
...
@@ -20,7 +20,8 @@ namespace paddle {
namespace
operators
{
namespace
reader
{
static
constexpr
size_t
kDoubleBufferSize
=
2
;
static
constexpr
size_t
kChannelSize
=
2
;
static
constexpr
size_t
kCacheSize
=
4
;
// kChannelSize + 2
class
DoubleBufferReader
:
public
framework
::
DecoratedReader
{
public:
...
...
@@ -34,33 +35,36 @@ class DoubleBufferReader : public framework::DecoratedReader {
explicit
DoubleBufferReader
(
ReaderBase
*
reader
,
platform
::
Place
target_place
=
platform
::
CPUPlace
())
:
DecoratedReader
(
reader
),
place_
(
target_place
)
{
for
(
size_t
i
=
0
;
i
<
kDoubleBufferSize
;
++
i
)
{
if
(
platform
::
is_gpu_place
(
place_
))
{
#ifdef PADDLE_WITH_CUDA
for
(
size_t
i
=
0
;
i
<
kChannelSize
+
2
;
++
i
)
{
if
(
platform
::
is_gpu_place
(
place_
))
{
ctxs_
.
emplace_back
(
new
platform
::
CUDADeviceContext
(
boost
::
get
<
platform
::
CUDAPlace
>
(
place_
)));
#endif
}
}
start_thread
();
}
void
start_thread
()
{
buffer_
=
framework
::
MakeChannel
<
Item
>
(
kDoubleBufferSize
);
prefetcher_
=
std
::
thread
([
this
]
{
PrefetchThreadFunc
();
});
#endif
StartPrefetcher
();
}
bool
HasNext
()
const
override
;
void
ReadNext
(
std
::
vector
<
framework
::
LoDTensor
>*
out
)
override
;
void
ReInit
()
override
;
~
DoubleBufferReader
()
{
void
StartPrefetcher
()
{
buffer_
=
framework
::
MakeChannel
<
Item
>
(
kChannelSize
);
prefetcher_
=
std
::
thread
([
this
]
{
PrefetchThreadFunc
();
});
}
void
EndPrefetcher
()
{
buffer_
->
Close
();
prefetcher_
.
join
();
if
(
prefecther_
.
joinable
())
{
prefetcher_
.
join
();
}
delete
buffer_
;
buffer_
=
nullptr
;
}
bool
HasNext
()
const
override
;
~
DoubleBufferReader
()
{
EndPrefetcher
();
}
private:
void
PrefetchThreadFunc
();
...
...
@@ -123,6 +127,15 @@ class CreateDoubleBufferReaderOpMaker : public DecoratedReaderMakerBase {
}
};
bool
DoubleBufferReader
::
HasNext
()
const
{
if
(
local_buffer_
.
payloads_
.
empty
())
{
bool
ok
=
buffer_
->
Receive
(
&
local_buffer_
);
return
ok
;
}
else
{
return
true
;
}
}
void
DoubleBufferReader
::
ReadNext
(
std
::
vector
<
framework
::
LoDTensor
>*
out
)
{
if
(
!
HasNext
())
{
PADDLE_THROW
(
"There is no next data!"
);
...
...
@@ -137,40 +150,36 @@ void DoubleBufferReader::ReadNext(std::vector<framework::LoDTensor>* out) {
void
DoubleBufferReader
::
ReInit
()
{
reader_
->
ReInit
();
buffer_
->
Close
();
prefetcher_
.
join
();
delete
buffer_
;
start_thread
();
EndPrefetcher
();
StartPrefetcher
();
}
void
DoubleBufferReader
::
PrefetchThreadFunc
()
{
VLOG
(
5
)
<<
"A new prefetch thread starts."
;
size_t
gpu_ctx_offset
=
0
;
std
::
vector
<
std
::
vector
<
framework
::
LoDTensor
>>
cpu_tensor_cache
(
4
);
std
::
vector
<
std
::
vector
<
framework
::
LoDTensor
>>
gpu_tensor_cache
(
4
);
size_t
tensor_cache_id
=
0
;
std
::
vector
<
std
::
vector
<
framework
::
LoDTensor
>>
cpu_tensor_cache
(
kCacheSize
);
std
::
vector
<
std
::
vector
<
framework
::
LoDTensor
>>
gpu_tensor_cache
(
kCacheSize
);
size_t
cached_tensor_id
=
0
;
while
(
reader_
->
HasNext
())
{
Item
batch
;
reader_
->
ReadNext
(
&
batch
.
payloads_
);
auto
&
cpu_batch
=
cpu_tensor_cache
[
cached_tensor_id
];
reader_
->
ReadNext
(
&
cpu_batch
);
if
(
platform
::
is_gpu_place
(
place_
))
{
tensor_cache_id
%=
4
;
auto
&
gpu_batch
=
gpu_tensor_cache
[
tensor_cache_id
];
auto
&
cpu_batch
=
cpu_tensor_cache
[
tensor_cache_id
];
cpu_batch
=
batch
.
payloads_
;
++
tensor_cache_id
;
auto
&
gpu_ctx
=
this
->
ctxs_
[
gpu_ctx_offset
++
];
gpu_ctx_offset
%=
this
->
ctxs_
.
size
();
gpu_batch
.
resize
(
batch
.
payloads_
.
size
());
auto
&
gpu_batch
=
gpu_tensor_cache
[
cached_tensor_id
];
auto
*
gpu_ctx
=
ctxs_
[
cached_tensor_id
].
get
();
gpu_batch
.
resize
(
cpu_batch
.
size
());
for
(
size_t
i
=
0
;
i
<
cpu_batch
.
size
();
++
i
)
{
framework
::
TensorCopy
(
cpu_batch
[
i
],
place_
,
*
gpu_ctx
,
&
gpu_batch
[
i
]);
gpu_batch
[
i
].
set_lod
(
batch
.
payloads_
[
i
].
lod
());
}
batch
.
ctx_
=
gpu_ctx
.
get
();
batch
.
payloads_
=
gpu_batch
;
batch
.
payload_
=
gpu_batch
;
batch
.
ctx_
=
gpu_ctx
;
}
else
{
// CPUPlace
batch
.
payload_
=
cpu_batch
;
}
++
cached_tensor_id
;
cached_tensor_id
%=
kCacheSize
;
try
{
buffer_
->
Send
(
&
batch
);
...
...
@@ -184,15 +193,6 @@ void DoubleBufferReader::PrefetchThreadFunc() {
VLOG
(
5
)
<<
"Prefetch thread terminates."
;
}
bool
DoubleBufferReader
::
HasNext
()
const
{
if
(
local_buffer_
.
payloads_
.
empty
())
{
bool
ok
=
buffer_
->
Receive
(
&
local_buffer_
);
return
ok
;
}
else
{
return
true
;
}
}
}
// namespace reader
}
// namespace operators
}
// 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