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
47999c42
P
PaddleDetection
项目概览
PaddlePaddle
/
PaddleDetection
1 年多 前同步成功
通知
696
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
未验证
提交
47999c42
编写于
2月 19, 2019
作者:
Y
Yan Xu
提交者:
GitHub
2月 19, 2019
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
cherry pick: multiple process launch utily test=release/1.3 (#15741)
上级
687fb401
变更
5
隐藏空白更改
内联
并排
Showing
5 changed file
with
37 addition
and
17 deletion
+37
-17
python/paddle/__init__.py
python/paddle/__init__.py
+1
-0
python/paddle/distributed/__init__.py
python/paddle/distributed/__init__.py
+13
-0
python/paddle/distributed/launch.py
python/paddle/distributed/launch.py
+22
-16
python/paddle/fluid/__init__.py
python/paddle/fluid/__init__.py
+0
-1
python/setup.py.in
python/setup.py.in
+1
-0
未找到文件。
python/paddle/__init__.py
浏览文件 @
47999c42
...
...
@@ -25,4 +25,5 @@ import paddle.reader
import
paddle.dataset
import
paddle.batch
import
paddle.compat
import
paddle.distributed
batch
=
batch
.
batch
python/paddle/distributed/__init__.py
0 → 100644
浏览文件 @
47999c42
# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
tools/run_mp
.py
→
python/paddle/distributed/launch
.py
浏览文件 @
47999c42
...
...
@@ -37,7 +37,7 @@ default_envs = {
GPUS
=
8
def
start_procs
(
gpus
,
cmd
,
log_dir
):
def
start_procs
(
gpus
,
entrypoint
,
entrypoint_args
,
log_dir
):
procs
=
[]
log_fns
=
[]
os
.
system
(
"mkdir -p %s"
%
log_dir
)
...
...
@@ -73,12 +73,11 @@ def start_procs(gpus, cmd, log_dir):
"PADDLE_TRAINER_ENDPOINTS"
:
all_nodes_devices_endpoints
})
print
(
"starting process "
,
i
,
cmd
,
curr_env
)
print
(
"starting process "
,
i
,
entrypoint
,
entrypoint_args
,
curr_env
)
fn
=
open
(
"%s/workerlog.%d"
%
(
log_dir
,
i
),
"w"
)
log_fns
.
append
(
fn
)
procs
.
append
(
subprocess
.
Popen
(
cmd
.
strip
().
split
(
" "
),
stdout
=
fn
,
stderr
=
fn
,
env
=
curr_env
))
cmd
=
[
sys
.
executable
,
"-u"
,
entrypoint
]
+
entrypoint_args
procs
.
append
(
subprocess
.
Popen
(
cmd
,
stdout
=
fn
,
stderr
=
fn
,
env
=
curr_env
))
for
i
in
range
(
gpus
):
try
:
...
...
@@ -89,7 +88,8 @@ def start_procs(gpus, cmd, log_dir):
pass
def
main
():
def
parse_args
():
parser
=
argparse
.
ArgumentParser
(
description
=
'''start paddle training using multi-process mode.
NOTE: your train program ***must*** run as distributed nccl2 mode,
...
...
@@ -108,21 +108,27 @@ POD_IP (current node ip address, not needed for local training)
type
=
int
,
default
=
8
,
help
=
'start number of processes for every gpu'
)
parser
.
add_argument
(
'--cmd'
,
type
=
str
,
default
=
""
,
help
=
'command to run for each process, e.g. python train.py --lr 0.1'
)
parser
.
add_argument
(
'--log_dir'
,
type
=
str
,
default
=
"mylog"
,
help
=
'directory to put logs per process.'
)
args
=
parser
.
parse_args
()
if
args
.
cmd
==
""
:
parser
.
print_help
()
exit
(
0
)
start_procs
(
args
.
gpus
,
args
.
cmd
,
args
.
log_dir
)
parser
.
add_argument
(
'entrypoint_script'
,
type
=
str
,
help
=
"The entrypoint script to be launched in parallel,"
"followed by all the arguments for each process,"
"e.g. train.py --lr 0.1"
)
parser
.
add_argument
(
'entrypoint_args'
,
nargs
=
argparse
.
REMAINDER
)
return
parser
.
parse_args
()
def
main
():
args
=
parse_args
()
# launch multiple training process
start_procs
(
args
.
gpus
,
args
.
entrypoint_script
,
args
.
entrypoint_args
,
args
.
log_dir
)
if
__name__
==
"__main__"
:
...
...
python/paddle/fluid/__init__.py
浏览文件 @
47999c42
...
...
@@ -161,7 +161,6 @@ def __bootstrap__():
'times_excess_than_required_tmp_allocation'
,
'enable_inplace_whitelist'
]
core
.
init_gflags
([
sys
.
argv
[
0
]]
+
[
"--tryfromenv="
+
","
.
join
(
read_env_flags
)])
core
.
init_glog
(
sys
.
argv
[
0
])
...
...
python/setup.py.in
浏览文件 @
47999c42
...
...
@@ -100,6 +100,7 @@ packages=['paddle',
'paddle.utils',
'paddle.dataset',
'paddle.reader',
'paddle.distributed',
'paddle.fluid',
'paddle.fluid.imperative',
'paddle.fluid.proto',
...
...
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