Skip to content
项目
群组
代码片段
帮助
当前项目
正在载入...
登录 / 注册
切换导航面板
P
pytensor
项目
项目
详情
活动
周期分析
仓库
仓库
文件
提交
分支
标签
贡献者
图表
比较
统计图
议题
0
议题
0
列表
看板
标记
里程碑
合并请求
0
合并请求
0
CI / CD
CI / CD
流水线
作业
日程
统计图
Wiki
Wiki
代码片段
代码片段
成员
成员
折叠边栏
关闭边栏
活动
图像
聊天
创建新问题
作业
提交
问题看板
Open sidebar
testgroup
pytensor
Commits
27cb3e25
提交
27cb3e25
authored
3月 14, 2017
作者:
Frederic Bastien
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
Fix the new gpu profile printing
上级
171346e8
隐藏空白字符变更
内嵌
并排
正在显示
1 个修改的文件
包含
11 行增加
和
12 行删除
+11
-12
basic_ops.py
theano/gpuarray/basic_ops.py
+11
-12
没有找到文件。
theano/gpuarray/basic_ops.py
浏览文件 @
27cb3e25
...
@@ -1549,46 +1549,45 @@ def profile_printer(message, compile_time, fct_call_time,
...
@@ -1549,46 +1549,45 @@ def profile_printer(message, compile_time, fct_call_time,
print
()
print
()
print
(
'Some info useful for gpu:'
)
print
(
'Some info useful for gpu:'
)
f
ct_call
=
set
()
f
graphs
=
set
()
for
node
in
apply_time
.
keys
():
for
node
in
apply_time
.
keys
():
f
ct_call
.
add
(
node
.
fgraph
)
f
graphs
.
add
(
node
.
fgraph
)
cpu
=
0
cpu
=
0
gpu
=
0
gpu
=
0
trans
=
0
trans
=
0
for
(
_
,
node
),
t
in
iteritems
(
apply_time
):
for
node
,
t
in
iteritems
(
apply_time
):
if
isinstance
(
node
.
op
.
__class__
.
__name__
,
if
isinstance
(
node
.
op
,
(
HostFromGpu
,
GpuFromHost
)):
(
HostFromGpu
,
GpuFromHost
)):
trans
+=
t
trans
+=
t
elif
node
.
op
.
__class__
.
__name__
.
lower
()
.
startswith
(
"gpu"
):
elif
node
.
op
.
__class__
.
__name__
.
lower
()
.
startswith
(
"gpu"
):
gpu
+=
t
gpu
+=
t
else
:
else
:
cpu
+=
t
cpu
+=
t
print
()
print
()
print
(
" Spent
%.3
fs(
%.
3
f
%%
) in cpu Op,
%.3
fs(
%.3
f
%%
) in gpu Op and
%.3
fs(
%.3
f
%%
) transfert Op"
%
(
print
(
" Spent
%.3
fs(
%.
2
f
%%
) in cpu Op,
%.3
fs(
%.2
f
%%
) in gpu Op and
%.3
fs(
%.2
f
%%
) transfert Op"
%
(
cpu
,
cpu
/
local_time
*
100
,
gpu
,
gpu
/
local_time
*
100
,
cpu
,
cpu
/
local_time
*
100
,
gpu
,
gpu
/
local_time
*
100
,
trans
,
trans
/
local_time
*
100
))
trans
,
trans
/
local_time
*
100
))
print
()
print
()
print
(
" Theano function input that are float64"
)
print
(
" Theano function input that are float64"
)
print
(
" <fct name> <input name> <input type> <str input>"
)
print
(
" <fct name> <input name> <input type> <str input>"
)
for
f
ct
in
fct_call
:
for
f
g
in
fgraphs
:
for
i
in
f
ct
.
input_storage
:
for
i
in
f
g
.
inputs
:
if
hasattr
(
i
.
type
,
'dtype'
)
and
i
.
type
.
dtype
==
'float64'
:
if
hasattr
(
i
.
type
,
'dtype'
)
and
i
.
type
.
dtype
==
'float64'
:
print
(
' '
,
f
ct
.
name
,
i
.
name
,
i
.
type
,
i
)
print
(
' '
,
f
g
.
name
,
i
.
name
,
i
.
type
,
i
)
print
()
print
()
print
(
" List of apply that don't have float64 as input but have float64 in outputs"
)
print
(
" List of apply that don't have float64 as input but have float64 in outputs"
)
print
(
" (Useful to know if we forgot some cast when using floatX=float32 or gpu code)"
)
print
(
" (Useful to know if we forgot some cast when using floatX=float32 or gpu code)"
)
print
(
' <Apply> <Apply position> <fct name> <inputs type> <outputs type>'
)
print
(
' <Apply> <Apply position> <fct name> <inputs type> <outputs type>'
)
for
f
ct
in
fct_call
:
for
f
g
in
fgraphs
:
for
idx
,
node
in
enumerate
(
f
ct
.
maker
.
fgraph
.
toposort
()):
for
idx
,
node
in
enumerate
(
f
g
.
toposort
()):
if
(
any
(
hasattr
(
i
,
'dtype'
)
and
i
.
dtype
==
'float64'
if
(
any
(
hasattr
(
i
,
'dtype'
)
and
i
.
dtype
==
'float64'
for
i
in
node
.
outputs
)
and
for
i
in
node
.
outputs
)
and
not
any
(
hasattr
(
i
,
'dtype'
)
and
i
.
dtype
==
'float64'
not
any
(
hasattr
(
i
,
'dtype'
)
and
i
.
dtype
==
'float64'
for
i
in
node
.
inputs
)):
for
i
in
node
.
inputs
)):
print
(
' '
,
str
(
node
),
idx
,
f
ct
.
name
,
end
=
' '
)
print
(
' '
,
str
(
node
),
idx
,
f
g
.
name
,
end
=
' '
)
print
(
str
([
getattr
(
i
,
'dtype'
,
None
)
print
(
str
([
getattr
(
i
,
'dtype'
,
None
)
for
i
in
node
.
inputs
]),
end
=
' '
)
for
i
in
node
.
inputs
]),
end
=
' '
)
print
(
str
([
getattr
(
i
,
'dtype'
,
None
)
print
(
str
([
getattr
(
i
,
'dtype'
,
None
)
...
...
编写
预览
Markdown
格式
0%
重试
或
添加新文件
添加附件
取消
您添加了
0
人
到此讨论。请谨慎行事。
请先完成此评论的编辑!
取消
请
注册
或者
登录
后发表评论