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
机器未来
Paddle
提交
ec773f90
P
Paddle
项目概览
机器未来
/
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
提交
ec773f90
编写于
9月 01, 2018
作者:
Y
yi.wu
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
fix ut merge error
上级
1b79974a
变更
1
显示空白变更内容
内联
并排
Showing
1 changed file
with
24 addition
and
21 deletion
+24
-21
python/paddle/fluid/tests/unittests/test_dist_base.py
python/paddle/fluid/tests/unittests/test_dist_base.py
+24
-21
未找到文件。
python/paddle/fluid/tests/unittests/test_dist_base.py
浏览文件 @
ec773f90
...
@@ -21,7 +21,7 @@ import sys
...
@@ -21,7 +21,7 @@ import sys
import
six
import
six
import
signal
import
signal
import
subprocess
import
subprocess
import
six
import
argparse
class
TestDistRunnerBase
(
object
):
class
TestDistRunnerBase
(
object
):
...
@@ -30,7 +30,7 @@ class TestDistRunnerBase(object):
...
@@ -30,7 +30,7 @@ class TestDistRunnerBase(object):
"get_model should be implemented by child classes."
)
"get_model should be implemented by child classes."
)
def
get_transpiler
(
self
,
trainer_id
,
main_program
,
pserver_endpoints
,
def
get_transpiler
(
self
,
trainer_id
,
main_program
,
pserver_endpoints
,
trainers
):
trainers
,
sync_mode
):
# NOTE: import fluid until runtime, or else forking processes will cause error.
# NOTE: import fluid until runtime, or else forking processes will cause error.
import
paddle
import
paddle
import
paddle.fluid
as
fluid
import
paddle.fluid
as
fluid
...
@@ -39,33 +39,35 @@ class TestDistRunnerBase(object):
...
@@ -39,33 +39,35 @@ class TestDistRunnerBase(object):
trainer_id
=
trainer_id
,
trainer_id
=
trainer_id
,
program
=
main_program
,
program
=
main_program
,
pservers
=
pserver_endpoints
,
pservers
=
pserver_endpoints
,
trainers
=
trainers
)
trainers
=
trainers
,
sync_mode
=
sync_mode
)
return
t
return
t
def
run_pserver
(
self
,
pserver_endpoints
,
trainers
,
current_endpoint
,
def
run_pserver
(
self
,
args
):
trainer_id
):
import
paddle
import
paddle
import
paddle.fluid
as
fluid
import
paddle.fluid
as
fluid
self
.
get_model
(
batch_size
=
2
)
self
.
get_model
(
batch_size
=
2
)
t
=
self
.
get_transpiler
(
trainer_id
,
t
=
self
.
get_transpiler
(
args
.
trainer_id
,
fluid
.
default_main_program
(),
pserver_endpoints
,
fluid
.
default_main_program
(),
args
.
endpoints
,
trainers
)
args
.
trainers
,
args
.
sync_mode
)
pserver_prog
=
t
.
get_pserver_program
(
current_endpoint
)
pserver_prog
=
t
.
get_pserver_program
(
args
.
current_endpoint
)
startup_prog
=
t
.
get_startup_program
(
current_endpoint
,
pserver_prog
)
startup_prog
=
t
.
get_startup_program
(
args
.
current_endpoint
,
pserver_prog
)
place
=
fluid
.
CPUPlace
()
place
=
fluid
.
CPUPlace
()
exe
=
fluid
.
Executor
(
place
)
exe
=
fluid
.
Executor
(
place
)
exe
.
run
(
startup_prog
)
exe
.
run
(
startup_prog
)
exe
.
run
(
pserver_prog
)
exe
.
run
(
pserver_prog
)
def
run_trainer
(
self
,
place
,
endpoints
,
trainer_id
,
trainers
,
is_dist
=
True
):
def
run_trainer
(
self
,
place
,
args
):
import
paddle
import
paddle
import
paddle.fluid
as
fluid
import
paddle.fluid
as
fluid
test_program
,
avg_cost
,
train_reader
,
test_reader
,
batch_acc
,
predict
=
\
test_program
,
avg_cost
,
train_reader
,
test_reader
,
batch_acc
,
predict
=
\
self
.
get_model
(
batch_size
=
2
)
self
.
get_model
(
batch_size
=
2
)
if
is_dist
:
if
args
.
is_dist
:
t
=
self
.
get_transpiler
(
trainer_id
,
t
=
self
.
get_transpiler
(
args
.
trainer_id
,
fluid
.
default_main_program
(),
endpoints
,
fluid
.
default_main_program
(),
trainers
)
args
.
endpoints
,
args
.
trainers
,
args
.
sync_mode
)
trainer_prog
=
t
.
get_trainer_program
()
trainer_prog
=
t
.
get_trainer_program
()
else
:
else
:
trainer_prog
=
fluid
.
default_main_program
()
trainer_prog
=
fluid
.
default_main_program
()
...
@@ -132,18 +134,21 @@ def runtime_main(test_class):
...
@@ -132,18 +134,21 @@ def runtime_main(test_class):
args
=
parser
.
parse_args
()
args
=
parser
.
parse_args
()
model
=
test_class
()
model
=
test_class
()
if
role
==
"pserver"
:
if
args
.
role
==
"pserver"
:
model
.
run_pserver
(
endpoints
,
trainers
,
current_endpoint
,
trainer_id
)
model
.
run_pserver
(
args
)
else
:
else
:
p
=
fluid
.
CUDAPlace
(
0
)
if
core
.
is_compiled_with_cuda
(
p
=
fluid
.
CUDAPlace
(
0
)
if
core
.
is_compiled_with_cuda
(
)
else
fluid
.
CPUPlace
()
)
else
fluid
.
CPUPlace
()
model
.
run_trainer
(
p
,
endpoints
,
trainer_id
,
trainers
,
is_dist
)
model
.
run_trainer
(
p
,
args
)
import
paddle.compat
as
cpt
import
paddle.compat
as
cpt
class
TestDistBase
(
unittest
.
TestCase
):
class
TestDistBase
(
unittest
.
TestCase
):
def
_setup_config
(
self
):
raise
NotImplementedError
(
"tests should have _setup_config implemented"
)
def
setUp
(
self
):
def
setUp
(
self
):
self
.
_trainers
=
2
self
.
_trainers
=
2
self
.
_pservers
=
2
self
.
_pservers
=
2
...
@@ -221,9 +226,7 @@ class TestDistBase(unittest.TestCase):
...
@@ -221,9 +226,7 @@ class TestDistBase(unittest.TestCase):
# Run local to get a base line
# Run local to get a base line
env_local
=
{
"CUDA_VISIBLE_DEVICES"
:
"0"
}
env_local
=
{
"CUDA_VISIBLE_DEVICES"
:
"0"
}
env_local
.
update
(
required_envs
)
env_local
.
update
(
required_envs
)
local_cmd
=
"%s %s trainer %s 0 %s %d FLASE"
%
\
local_cmd
=
"%s %s --role trainer"
%
(
self
.
_python_interp
,
model_file
)
(
self
.
_python_interp
,
model_file
,
"127.0.0.1:1234"
,
"127.0.0.1:1234"
,
1
)
if
not
check_error_log
:
if
not
check_error_log
:
local_proc
=
subprocess
.
Popen
(
local_proc
=
subprocess
.
Popen
(
local_cmd
.
split
(
" "
),
local_cmd
.
split
(
" "
),
...
...
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