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
41af738a
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
41af738a
编写于
6月 12, 2017
作者:
H
Helin Wang
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
fix according to comments
上级
54e8263c
变更
2
隐藏空白更改
内联
并排
Showing
2 changed file
with
37 addition
and
32 deletion
+37
-32
go/connection/conn.go
go/connection/conn.go
+1
-0
go/master/service.go
go/master/service.go
+36
-32
未找到文件。
go/connection/conn.go
浏览文件 @
41af738a
...
@@ -62,6 +62,7 @@ func (c *Conn) Connect(addr string) error {
...
@@ -62,6 +62,7 @@ func (c *Conn) Connect(addr string) error {
c
.
waitConn
=
nil
c
.
waitConn
=
nil
}
}
}
else
{
}
else
{
client
.
Close
()
return
errors
.
New
(
"client already set from a concurrent goroutine"
)
return
errors
.
New
(
"client already set from a concurrent goroutine"
)
}
}
...
...
go/master/service.go
浏览文件 @
41af738a
...
@@ -11,10 +11,6 @@ import (
...
@@ -11,10 +11,6 @@ import (
"github.com/PaddlePaddle/recordio"
"github.com/PaddlePaddle/recordio"
)
)
const
(
targetTaskCount
=
300
)
// Service is the master server service.
// Service is the master server service.
type
Service
struct
{
type
Service
struct
{
chunksPerTask
int
chunksPerTask
int
...
@@ -23,7 +19,7 @@ type Service struct {
...
@@ -23,7 +19,7 @@ type Service struct {
ready
chan
struct
{}
ready
chan
struct
{}
mu
sync
.
Mutex
mu
sync
.
Mutex
init
Began
bool
init
Done
bool
taskQueues
taskQueues
taskQueues
taskQueues
}
}
...
@@ -104,54 +100,35 @@ func (s *Service) snapshot() error {
...
@@ -104,54 +100,35 @@ func (s *Service) snapshot() error {
return
nil
return
nil
}
}
// SetDataset sets dataset to dispatch for the master server.
func
getChunks
(
globPaths
[]
string
)
([]
Chunk
,
error
)
{
//
// SetDataset can be call multiple times. But only the first call will
// be honored.
func
(
s
*
Service
)
SetDataset
(
globPaths
[]
string
,
dummy
*
int
)
error
{
if
len
(
globPaths
)
==
0
{
return
errors
.
New
(
"no dataset specified"
)
}
s
.
mu
.
Lock
()
defer
s
.
mu
.
Unlock
()
if
s
.
initBegan
{
// SetDataset already called. All trainer will call
// SetDataset, but we only handle the first one. Treat
// other calls as successful but do nothing.
return
nil
}
s
.
initBegan
=
true
var
chunks
[]
Chunk
var
chunks
[]
Chunk
var
paths
[]
string
var
paths
[]
string
for
_
,
s
:=
range
globPaths
{
for
_
,
s
:=
range
globPaths
{
match
,
err
:=
filepath
.
Glob
(
s
)
match
,
err
:=
filepath
.
Glob
(
s
)
if
err
!=
nil
{
if
err
!=
nil
{
panic
(
err
)
return
nil
,
err
}
}
paths
=
append
(
paths
,
match
...
)
paths
=
append
(
paths
,
match
...
)
}
}
if
len
(
paths
)
==
0
{
if
len
(
paths
)
==
0
{
return
errors
.
New
(
"no valid datset specified"
)
return
nil
,
errors
.
New
(
"no valid datset specified"
)
}
}
for
_
,
path
:=
range
paths
{
for
_
,
path
:=
range
paths
{
f
,
err
:=
os
.
Open
(
path
)
f
,
err
:=
os
.
Open
(
path
)
if
err
!=
nil
{
if
err
!=
nil
{
panic
(
err
)
return
nil
,
err
}
}
index
,
err
:=
recordio
.
LoadIndex
(
f
)
index
,
err
:=
recordio
.
LoadIndex
(
f
)
if
err
!=
nil
{
if
err
!=
nil
{
return
err
return
nil
,
err
}
}
err
=
f
.
Close
()
err
=
f
.
Close
()
if
err
!=
nil
{
if
err
!=
nil
{
return
err
return
nil
,
err
}
}
count
:=
index
.
NumChunks
()
count
:=
index
.
NumChunks
()
...
@@ -164,14 +141,41 @@ func (s *Service) SetDataset(globPaths []string, dummy *int) error {
...
@@ -164,14 +141,41 @@ func (s *Service) SetDataset(globPaths []string, dummy *int) error {
}
}
}
}
return
chunks
,
nil
}
// SetDataset sets dataset to dispatch for the master server.
//
// SetDataset can be call multiple times. But only the first call will
// be honored.
func
(
s
*
Service
)
SetDataset
(
globPaths
[]
string
,
dummy
*
int
)
error
{
if
len
(
globPaths
)
==
0
{
return
errors
.
New
(
"no dataset specified"
)
}
s
.
mu
.
Lock
()
defer
s
.
mu
.
Unlock
()
if
s
.
initDone
{
// Already initialized. All trainer will call
// SetDataset, but we only handle the first one. Treat
// other calls as successful but do nothing.
return
nil
}
chunks
,
err
:=
getChunks
(
globPaths
)
if
err
!=
nil
{
return
err
}
s
.
taskQueues
.
Todo
=
partition
(
chunks
,
s
.
chunksPerTask
)
s
.
taskQueues
.
Todo
=
partition
(
chunks
,
s
.
chunksPerTask
)
err
:
=
s
.
snapshot
()
err
=
s
.
snapshot
()
if
err
!=
nil
{
if
err
!=
nil
{
return
err
return
err
}
}
close
(
s
.
ready
)
close
(
s
.
ready
)
s
.
initDone
=
true
return
nil
return
nil
}
}
...
@@ -193,7 +197,7 @@ func (s *Service) GetTask(dummy int, task *Task) error {
...
@@ -193,7 +197,7 @@ func (s *Service) GetTask(dummy int, task *Task) error {
// TODO(helin): client need to retry in this
// TODO(helin): client need to retry in this
// error case. Gotcha: RPC client can't
// error case. Gotcha: RPC client can't
// compare returned error with predefined
// compare returned error with predefined
// erros like io.EOF. Because interface don't
// erro
r
s like io.EOF. Because interface don't
// have same dynamic value when in different
// have same dynamic value when in different
// process.
// process.
return
errors
.
New
(
"no more available task"
)
return
errors
.
New
(
"no more available task"
)
...
...
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