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
PaddleDetection
提交
0d491b67
P
PaddleDetection
项目概览
PaddlePaddle
/
PaddleDetection
大约 1 年 前同步成功
通知
695
Star
11112
Fork
2696
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
184
列表
看板
标记
里程碑
合并请求
40
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
P
PaddleDetection
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
184
Issue
184
列表
看板
标记
里程碑
合并请求
40
合并请求
40
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
0d491b67
编写于
4月 26, 2018
作者:
Q
qiaolongfei
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
use-multi-thread-todo-update
上级
0264ec39
变更
2
隐藏空白更改
内联
并排
Showing
2 changed file
with
40 addition
and
32 deletion
+40
-32
paddle/fluid/operators/listen_and_serv_op.cc
paddle/fluid/operators/listen_and_serv_op.cc
+39
-29
paddle/fluid/operators/listen_and_serv_op.h
paddle/fluid/operators/listen_and_serv_op.h
+1
-3
未找到文件。
paddle/fluid/operators/listen_and_serv_op.cc
浏览文件 @
0d491b67
...
...
@@ -45,20 +45,6 @@ static void split(const std::string &str, char sep,
}
}
static
void
AsyncExecuteBlock
(
framework
::
Executor
*
executor
,
framework
::
ExecutorPrepareContext
*
prepared
,
framework
::
Scope
*
scope
)
{
std
::
future
<
void
>
future
=
framework
::
Async
([
&
executor
,
&
prepared
,
&
scope
]()
{
try
{
executor
->
RunPreparedContext
(
prepared
,
scope
,
false
,
false
);
}
catch
(
std
::
exception
&
e
)
{
LOG
(
ERROR
)
<<
"run sub program error "
<<
e
.
what
();
}
});
// TODO(qiao) maybe we can remove this
future
.
wait
();
}
static
void
ParallelExecuteBlocks
(
const
std
::
vector
<
size_t
>
&
parallel_blkids
,
framework
::
Executor
*
executor
,
const
std
::
vector
<
std
::
shared_ptr
<
framework
::
ExecutorPrepareContext
>>
...
...
@@ -201,14 +187,35 @@ void ListenAndServOp::RunSyncLoop(framework::Executor *executor,
}
// while(true)
}
static
void
AsyncUpdateThread
(
const
bool
&
exit_flag
,
const
std
::
shared_ptr
<
detail
::
ReceivedQueue
>
&
queue
,
framework
::
Executor
*
executor
,
framework
::
ExecutorPrepareContext
*
prepared
)
{
while
(
!
exit_flag
)
{
const
detail
::
ReceivedMessage
v
=
queue
->
Pop
();
auto
recv_var_name
=
v
.
first
;
auto
var
=
v
.
second
->
GetVar
();
if
(
var
==
nullptr
)
{
LOG
(
ERROR
)
<<
"Can not find server side var: "
<<
recv_var_name
;
PADDLE_THROW
(
"Can not find server side var"
);
}
try
{
executor
->
RunPreparedContext
(
prepared
,
v
.
second
->
GetMutableLocalScope
(),
false
,
false
);
}
catch
(
std
::
exception
&
e
)
{
LOG
(
ERROR
)
<<
"run sub program error "
<<
e
.
what
();
}
}
}
void
ListenAndServOp
::
RunAsyncLoop
(
framework
::
Executor
*
executor
,
framework
::
ProgramDesc
*
program
,
framework
::
Scope
*
recv_scope
,
framework
::
BlockDesc
*
prefetch_block
)
const
{
framework
::
ProgramDesc
*
program
)
const
{
VLOG
(
3
)
<<
"RunAsyncLoop in"
;
// grad name to block id
std
::
unordered_map
<
std
::
string
,
int32_t
>
grad_to_block_id
;
std
::
unordered_map
<
int32_t
,
std
::
string
>
id_to_grad
;
std
::
unordered_map
<
std
::
string
,
std
::
shared_ptr
<
detail
::
ReceivedQueue
>>
grad_to_queue
;
auto
grad_to_block_id_str
=
Attr
<
std
::
vector
<
std
::
string
>>
(
"grad_to_block_id"
);
...
...
@@ -220,6 +227,7 @@ void ListenAndServOp::RunAsyncLoop(framework::Executor *executor,
PADDLE_ENFORCE_EQ
(
grad_to_block_id
.
count
(
pieces
[
0
]),
0
);
int
block_id
=
std
::
stoi
(
pieces
[
1
]);
grad_to_block_id
[
pieces
[
0
]]
=
block_id
;
grad_to_queue
[
pieces
[
0
]]
=
std
::
make_shared
<
detail
::
ReceivedQueue
>
();
id_to_grad
[
block_id
]
=
pieces
[
0
];
}
size_t
num_blocks
=
program
->
Size
();
...
...
@@ -240,6 +248,18 @@ void ListenAndServOp::RunAsyncLoop(framework::Executor *executor,
VLOG
(
3
)
<<
"RunAsyncLoop into while"
;
bool
exit_flag
=
false
;
VLOG
(
3
)
<<
"start async optimize threads"
;
std
::
vector
<
std
::
future
<
void
>>
fs
;
for
(
auto
iter
=
grad_to_queue
.
begin
();
iter
!=
grad_to_queue
.
end
();
iter
++
)
{
std
::
string
grad_name
=
iter
->
first
;
fs
.
push_back
(
framework
::
Async
([
grad_name
,
&
exit_flag
,
&
executor
,
&
grad_to_queue
,
&
grad_to_prepared_block
]()
{
AsyncUpdateThread
(
exit_flag
,
grad_to_queue
[
grad_name
],
executor
,
grad_to_prepared_block
[
grad_name
].
get
());
}));
}
while
(
!
exit_flag
)
{
const
detail
::
ReceivedMessage
v
=
rpc_service_
->
Get
();
auto
recv_var_name
=
v
.
first
;
...
...
@@ -249,17 +269,7 @@ void ListenAndServOp::RunAsyncLoop(framework::Executor *executor,
break
;
}
else
{
VLOG
(
3
)
<<
"received grad: "
<<
recv_var_name
;
auto
var
=
v
.
second
->
GetVar
();
if
(
var
==
nullptr
)
{
LOG
(
ERROR
)
<<
"Can not find server side var: "
<<
recv_var_name
;
PADDLE_THROW
(
"Can not find server side var"
);
}
AsyncExecuteBlock
(
executor
,
grad_to_prepared_block
[
recv_var_name
].
get
(),
v
.
second
->
GetMutableLocalScope
());
// TODO(qiao): explain why
if
(
var
->
IsType
<
framework
::
SelectedRows
>
())
{
var
->
GetMutable
<
framework
::
SelectedRows
>
()
->
mutable_rows
()
->
clear
();
}
grad_to_queue
[
recv_var_name
]
->
Push
(
v
);
}
if
(
exit_flag
)
{
...
...
@@ -308,7 +318,7 @@ void ListenAndServOp::RunImpl(const framework::Scope &scope,
if
(
sync_mode
)
{
RunSyncLoop
(
&
executor
,
program
,
&
recv_scope
,
prefetch_block
);
}
else
{
RunAsyncLoop
(
&
executor
,
program
,
&
recv_scope
,
prefetch_block
);
RunAsyncLoop
(
&
executor
,
program
);
}
}
...
...
paddle/fluid/operators/listen_and_serv_op.h
浏览文件 @
0d491b67
...
...
@@ -47,9 +47,7 @@ class ListenAndServOp : public framework::OperatorBase {
framework
::
BlockDesc
*
prefetch_block
)
const
;
void
RunAsyncLoop
(
framework
::
Executor
*
executor
,
framework
::
ProgramDesc
*
program
,
framework
::
Scope
*
recv_scope
,
framework
::
BlockDesc
*
prefetch_block
)
const
;
framework
::
ProgramDesc
*
program
)
const
;
void
Stop
()
override
;
...
...
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