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
BaiXuePrincess
Paddle
提交
ca1cab3e
P
Paddle
项目概览
BaiXuePrincess
/
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
0
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
0
Issue
0
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
未验证
提交
ca1cab3e
编写于
9月 08, 2022
作者:
P
pangyoki
提交者:
GitHub
9月 08, 2022
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
new executor support compiled_program constructed by graph (#45836)
上级
ef53e1b4
变更
1
隐藏空白更改
内联
并排
Showing
1 changed file
with
11 addition
and
11 deletion
+11
-11
python/paddle/fluid/executor.py
python/paddle/fluid/executor.py
+11
-11
未找到文件。
python/paddle/fluid/executor.py
浏览文件 @
ca1cab3e
...
...
@@ -708,6 +708,11 @@ class _ExecutorCache(object):
# NOTE(Ruibiao): Not all changeable item is considered for key at present,
# ONLY: program, feed, and fetch_list
if
isinstance
(
self
.
program
,
compiler
.
CompiledProgram
):
if
not
self
.
program
.
_program
:
# The program holds no _program, maybe it is constructed by graph.
# Convert graph to program in order to generate key.
self
.
program
.
_program
=
framework
.
IrGraph
(
self
.
program
.
_graph
).
to_program
()
self
.
key
=
hash
(
_get_strong_program_cache_key_for_new_exe
(
self
.
program
.
_program
,
feed
,
fetch_list
))
...
...
@@ -1125,6 +1130,7 @@ class Executor(object):
warnings
.
warn
(
"The program holds no _program, maybe it is constructed by graph."
)
return
feed
else
:
global_block
=
program
.
global_block
()
...
...
@@ -1566,13 +1572,7 @@ class Executor(object):
if
compiled
:
compiled_program
=
program
if
isinstance
(
program
,
compiler
.
CompiledProgram
)
else
program
.
_graph
# Unsupported case 1 : the CompiledProgram is constructed by Graph
if
compiled_program
.
_program
is
None
:
warnings
.
warn
(
"Standalone executor is not used for Graph"
,
UserWarning
)
return
False
# Unsupported case 2: data parallel
# Unsupported case 1: data parallel
if
compiled_program
.
_is_data_parallel
and
len
(
compiled_program
.
_get_places
(
place
,
compiled_program
.
_places
))
!=
1
:
...
...
@@ -1581,7 +1581,7 @@ class Executor(object):
UserWarning
)
return
False
# Unsupported case
3
: parallel graph
# Unsupported case
2
: parallel graph
if
core
.
globals
()[
'FLAGS_enable_parallel_graph'
]
in
[
1
,
'1'
,
True
,
'True'
,
'true'
]:
...
...
@@ -1590,21 +1590,21 @@ class Executor(object):
UserWarning
)
return
False
# Unsupported case
4
: inference
# Unsupported case
3
: inference
if
compiled_program
.
_is_inference
:
warnings
.
warn
(
"Standalone executor is not used for inference"
,
UserWarning
)
return
False
# Unsupported case
5
: CUDA Graph
# Unsupported case
4
: CUDA Graph
if
compiled_program
.
_build_strategy
is
not
None
and
compiled_program
.
_build_strategy
.
allow_cuda_graph_capture
:
warnings
.
warn
(
"Standalone executor is not used for CUDA Graph"
,
UserWarning
)
return
False
# Unsupported case
6
: async mode
# Unsupported case
5
: async mode
if
compiled_program
.
_build_strategy
is
not
None
and
compiled_program
.
_build_strategy
.
async_mode
:
warnings
.
warn
(
"Standalone executor is not used for async mode"
,
...
...
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