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
0bddb951
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
未验证
提交
0bddb951
编写于
12月 03, 2019
作者:
T
tangwei12
提交者:
GitHub
12月 03, 2019
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
fix async mode, test=develop (#21367)
上级
94bef035
变更
1
隐藏空白更改
内联
并排
Showing
1 changed file
with
19 addition
and
12 deletion
+19
-12
paddle/fluid/operators/distributed_ops/listen_and_serv_op.cc
paddle/fluid/operators/distributed_ops/listen_and_serv_op.cc
+19
-12
未找到文件。
paddle/fluid/operators/distributed_ops/listen_and_serv_op.cc
浏览文件 @
0bddb951
...
...
@@ -464,23 +464,21 @@ void ListenAndServOp::RunImpl(const framework::Scope &scope,
f
(
request_get_no_barrier_handler_
.
get
());
f
(
request_notify_handler_
.
get
());
// start the server listening after all member initialized.
server_thread_
.
reset
(
new
std
::
thread
(
RunServer
,
rpc_service_
));
VLOG
(
3
)
<<
"wait server thread to become ready..."
;
rpc_service_
->
WaitServerReady
();
// register SIGINT(from ctrl+C) and SIGTERM(from kill) signal handlers
signal
(
SIGINT
,
SignalHandler
::
StopAndExit
);
signal
(
SIGTERM
,
SignalHandler
::
StopAndExit
);
// Cache the type of the received vars as `sparse_vars_` and `dense_vars_`
// so that we can reset them at the end of each iteration.
// NOTE: only used in sync update
CacheVarsType
(
inputs
,
recv_scope
);
// Write to a file of server selected port for python use.
SavePort
();
if
(
sync_mode
)
{
// start the server listening after all member initialized.
server_thread_
.
reset
(
new
std
::
thread
(
RunServer
,
rpc_service_
));
VLOG
(
3
)
<<
"wait server thread to become ready..."
;
rpc_service_
->
WaitServerReady
();
CacheVarsType
(
inputs
,
recv_scope
);
// Write to a file of server selected port for python use.
SavePort
();
RunSyncLoop
(
&
executor
,
program
,
&
recv_scope
,
&
dev_ctx
,
prefetch_block_id_list
,
checkpoint_block_id
);
}
else
{
...
...
@@ -498,6 +496,15 @@ void ListenAndServOp::RunImpl(const framework::Scope &scope,
split
(
grad_to_block_id_str
[
0
],
':'
,
&
pieces
);
distributed
::
HeartBeatMonitor
::
Init
(
fan_in
,
pserver_id
==
0
,
pieces
[
0
]);
}
// start the server listening after all member initialized.
server_thread_
.
reset
(
new
std
::
thread
(
RunServer
,
rpc_service_
));
VLOG
(
3
)
<<
"wait server thread to become ready..."
;
rpc_service_
->
WaitServerReady
();
// Write to a file of server selected port for python use.
SavePort
();
RunAsyncLoop
(
&
executor
,
program
,
&
recv_scope
);
}
}
...
...
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