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
5e6276ed
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
5e6276ed
编写于
3月 22, 2018
作者:
T
typhoonzero
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
fix transpiler bug
上级
1eec9261
变更
2
显示空白变更内容
内联
并排
Showing
2 changed file
with
9 addition
and
6 deletion
+9
-6
paddle/fluid/operators/send_op.cc
paddle/fluid/operators/send_op.cc
+4
-4
python/paddle/fluid/distribute_transpiler.py
python/paddle/fluid/distribute_transpiler.py
+5
-2
未找到文件。
paddle/fluid/operators/send_op.cc
浏览文件 @
5e6276ed
...
@@ -68,7 +68,7 @@ class SendOp : public framework::OperatorBase {
...
@@ -68,7 +68,7 @@ class SendOp : public framework::OperatorBase {
for
(
size_t
i
=
0
;
i
<
ins
.
size
();
i
++
)
{
for
(
size_t
i
=
0
;
i
<
ins
.
size
();
i
++
)
{
if
(
NeedSend
(
scope
,
ins
[
i
]))
{
if
(
NeedSend
(
scope
,
ins
[
i
]))
{
VLOG
(
3
)
<<
"sending "
<<
ins
[
i
]
<<
" to "
<<
epmap
[
i
];
VLOG
(
2
)
<<
"sending "
<<
ins
[
i
]
<<
" to "
<<
epmap
[
i
];
rpc_client
->
AsyncSendVariable
(
epmap
[
i
],
ctx
,
scope
,
ins
[
i
]);
rpc_client
->
AsyncSendVariable
(
epmap
[
i
],
ctx
,
scope
,
ins
[
i
]);
}
else
{
}
else
{
VLOG
(
3
)
<<
"don't send no-initialied variable: "
<<
ins
[
i
];
VLOG
(
3
)
<<
"don't send no-initialied variable: "
<<
ins
[
i
];
...
@@ -77,20 +77,20 @@ class SendOp : public framework::OperatorBase {
...
@@ -77,20 +77,20 @@ class SendOp : public framework::OperatorBase {
PADDLE_ENFORCE
(
rpc_client
->
Wait
());
PADDLE_ENFORCE
(
rpc_client
->
Wait
());
for
(
auto
&
ep
:
endpoints
)
{
for
(
auto
&
ep
:
endpoints
)
{
VLOG
(
3
)
<<
"batch barrier, ep: "
<<
ep
;
VLOG
(
2
)
<<
"batch barrier, ep: "
<<
ep
;
rpc_client
->
AsyncSendBatchBarrier
(
ep
);
rpc_client
->
AsyncSendBatchBarrier
(
ep
);
}
}
PADDLE_ENFORCE
(
rpc_client
->
Wait
());
PADDLE_ENFORCE
(
rpc_client
->
Wait
());
if
(
outs
.
size
()
>
0
)
{
if
(
outs
.
size
()
>
0
)
{
for
(
size_t
i
=
0
;
i
<
outs
.
size
();
i
++
)
{
for
(
size_t
i
=
0
;
i
<
outs
.
size
();
i
++
)
{
VLOG
(
3
)
<<
"getting "
<<
outs
[
i
]
<<
" from "
<<
epmap
[
i
];
VLOG
(
2
)
<<
"getting "
<<
outs
[
i
]
<<
" from "
<<
epmap
[
i
];
rpc_client
->
AsyncGetVariable
(
epmap
[
i
],
ctx
,
scope
,
outs
[
i
]);
rpc_client
->
AsyncGetVariable
(
epmap
[
i
],
ctx
,
scope
,
outs
[
i
]);
}
}
PADDLE_ENFORCE
(
rpc_client
->
Wait
());
PADDLE_ENFORCE
(
rpc_client
->
Wait
());
// tell pservers that current trainer have called fetch
// tell pservers that current trainer have called fetch
for
(
auto
&
ep
:
endpoints
)
{
for
(
auto
&
ep
:
endpoints
)
{
VLOG
(
3
)
<<
"send fetch barrier, ep: "
<<
ep
;
VLOG
(
2
)
<<
"send fetch barrier, ep: "
<<
ep
;
rpc_client
->
AsyncSendFetchBarrier
(
ep
);
rpc_client
->
AsyncSendFetchBarrier
(
ep
);
}
}
PADDLE_ENFORCE
(
rpc_client
->
Wait
());
PADDLE_ENFORCE
(
rpc_client
->
Wait
());
...
...
python/paddle/fluid/distribute_transpiler.py
浏览文件 @
5e6276ed
...
@@ -563,6 +563,8 @@ class DistributeTranspiler:
...
@@ -563,6 +563,8 @@ class DistributeTranspiler:
orig_var_name
=
""
orig_var_name
=
""
if
suff_idx
>=
0
:
if
suff_idx
>=
0
:
orig_var_name
=
varname
[:
suff_idx
]
orig_var_name
=
varname
[:
suff_idx
]
else
:
orig_var_name
=
varname
return
orig_var_name
return
orig_var_name
def
_append_pserver_ops
(
self
,
optimize_block
,
opt_op
,
endpoint
,
def
_append_pserver_ops
(
self
,
optimize_block
,
opt_op
,
endpoint
,
...
@@ -577,7 +579,8 @@ class DistributeTranspiler:
...
@@ -577,7 +579,8 @@ class DistributeTranspiler:
grad_block
=
None
grad_block
=
None
for
g
in
self
.
param_grad_ep_mapping
[
endpoint
][
"grads"
]:
for
g
in
self
.
param_grad_ep_mapping
[
endpoint
][
"grads"
]:
if
same_or_split_var
(
if
same_or_split_var
(
self
.
_orig_varname
(
g
.
name
),
opt_op
.
input
(
key
)[
0
]):
self
.
_orig_varname
(
g
.
name
),
self
.
_orig_varname
(
opt_op
.
input
(
key
)[
0
])):
grad_block
=
g
grad_block
=
g
break
break
if
not
grad_block
:
if
not
grad_block
:
...
@@ -748,7 +751,7 @@ class DistributeTranspiler:
...
@@ -748,7 +751,7 @@ class DistributeTranspiler:
param_names
=
[
param_names
=
[
p
.
name
for
p
in
self
.
param_grad_ep_mapping
[
endpoint
][
"params"
]
p
.
name
for
p
in
self
.
param_grad_ep_mapping
[
endpoint
][
"params"
]
]
]
if
op
.
input
(
"Param"
)
in
param_names
:
if
op
.
input
(
"Param"
)
[
0
]
in
param_names
:
return
True
return
True
else
:
else
:
for
n
in
param_names
:
for
n
in
param_names
:
...
...
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