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
ad2dfef4
P
Paddle
项目概览
PaddlePaddle
/
Paddle
大约 1 年 前同步成功
通知
2298
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
体验新版 GitCode,发现更多精彩内容 >>
未验证
提交
ad2dfef4
编写于
2月 12, 2018
作者:
Y
Yang Yang(Tony)
提交者:
GitHub
2月 12, 2018
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
Update parallel_do.md
上级
549c74a9
变更
1
隐藏空白更改
内联
并排
Showing
1 changed file
with
75 addition
and
1 deletion
+75
-1
doc/design/parallel_do.md
doc/design/parallel_do.md
+75
-1
未找到文件。
doc/design/parallel_do.md
浏览文件 @
ad2dfef4
...
...
@@ -41,6 +41,7 @@ This implementation allows to write mixed device program like this
# get embedding feature on CPU
feature
=
some_cpu_only_op
(
data
)
gpu_places
=
get_place
(
use_gpu
=
True
)
# parallel processing on multiple GPUs
pd
=
ParallelDo
(
gpu_places
)
with
pd
.
do
():
...
...
@@ -51,6 +52,38 @@ prediction = pd()
loss
=
cross_entropy
(
prediction
,
label
)
```
And the programDesc are like the following
```
# start_program will be run by executor(CPUPlace), all w1, w2 will be allocated on CPU
start_program
{
vars: w1, w2
ops: init(w1), init(w2)
}
main_program
{
block0 {
vars: data, places, w1, w2
ops: data, get_place, parallel_do(block1),
parallel_do_grad(block2),
sgd(w2, w2_grad),
sgd(w1, w1_grad)
}
block1 {
vars: data, h1, h2, loss
ops: fc, fc, softmax
}
block2 {
vars: data_grad, h1_grad, h2_grad, loss_gard, w1_grad, w2_grad
ops: softmax_grad,
fc_grad
fc_grad
}
}
```
## Proformance Imporvement
There are serial places we can make this parallel_do faster.
...
...
@@ -78,6 +111,47 @@ We can avoid this step by making each device have a copy of the parameter. This
1.
`allreduce`
operators need to be called in async mode to achieve maximum throughput
1.
apply gradients related op(i.e. cliping, normalization, decay, sgd) on different devices in parallel
By doing so, we also avoided "backward: accumulate param@grad from different devices to the first device"
By doing so, we also avoided "backward: accumulate param@grad from different devices to the first device".
And the ProgramDesc looks like the following
```
# w1, w2 will be allocated on all GPUs
start_program
{
block0 {
parallel_do(block1)
}
block1 {
vars: w1, w2
ops: init(w1), init(w2)
}
}
main_program
{
block0 {
vars: data, places, w1, w2
ops: data, get_place, parallel_do(block1),
parallel_do_grad(block2), # append_backward
parallel_do(block3) # append_optimization
}
block1 {
vars: data, h1, h2, loss
ops: fc, fc, softmax
}
block2 {
vars: data_grad, h1_grad, h2_grad, loss_gard, w1_grad, w2_grad
ops: softmax_grad,
fc_grad, allreduce(places, scopes, w1_grad),
fc_grad, allreduce(places, scopes, w2_grad)
}
block3 {
vars: lr
ops: sgd(w2, w2_grad),
sgd(w1, w1_grad)
}
}
```
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