Skip to content

P
pytensor
提交 50740a0c authored 作者: Frederic Bastien's avatar Frederic Bastien
浏览文件
操作

Updated the profiler output.

上级 e347153f
隐藏空白字符变更
内嵌 并排
正在显示 包含 46 行增加 和 47 行删除
doc/hpcs2011_tutorial/presentation.tex
浏览文件 @ 50740a0c
......@@ -687,17 +687,15 @@ To replace the default mode with this mode, use the Theano flags \texttt{mode=Pr
To enable the memory profiling use the flags \texttt{ProfileMode.profile\_memory=True}
\begin{Verbatim}
Time since import 1.486s
Theano compile time: 1.017s (67.9% since import)
Optimization time: 0.805s
Linker time: 0.199s
Theano fct call 0.002s (0.1% since import)
Theano Op time 0.001s 0.0%(since import) 36.8%(of fct call)
Theano function overhead in ProfileMode 0.001s 0.1%(since import)
63.2%(of fct call)
11 Theano fct call, 0.000s per call
Rest of the time since import 0.479s 32.0%
Time since import 2.697s
Theano compile time: 1.046s (38.8% since import)
Optimization time: 0.804s
Linker time: 0.230s
Theano fct call 0.028s (1.0% since import)
Theano Op time 0.026s 1.0%(since import) 93.7%(of fct call)
Theano function overhead in ProfileMode 0.002s 0.1%(since import) 6.3%(of fct call)
11 Theano fct call, 0.003s per call
Rest of the time since import 1.623s 60.2%
\end{Verbatim}
\end{frame}
......@@ -708,8 +706,8 @@ Theano outputs:
\begin{Verbatim}
Theano fct summary:
<% total fct time> <total time> <time per call> <nb call> <fct name>
97.1% 0.002s 1.64e-04s 10 train
2.9% 0.000s 4.91e-05s 1 predict
97.2% 0.027s 2.70e-03s 10 train
2.8% 0.001s 7.84e-04s 1 predict
\end{Verbatim}
\end{frame}
......@@ -722,15 +720,14 @@ Single Op-wise summary:
<% of local_time spent on this kind of Op> <cumulative %>
<self seconds> <cumulative seconds> <time per call> <nb_call>
<nb_op> <nb_apply> <Op name>
30.8% 30.8% 0.000s 0.000s 1.86e-05s 10 1 1 <'Gemv'>
23.8% 54.6% 0.000s 0.000s 1.58e-06s * 91 10 10 <'Elemwise'>
18.3% 72.9% 0.000s 0.000s 1.10e-05s 10 1 1 <'Alloc'>
15.9% 88.7% 0.000s 0.001s 8.71e-06s 11 1 2 <'Dot'>
7.7% 96.4% 0.000s 0.001s 1.49e-06s * 31 2 4 <'DimShuffle'>
2.0% 98.4% 0.000s 0.001s 1.22e-06s * 10 1 1 <'Sum'>
1.6% 100.0% 0.000s 0.001s 9.78e-07s * 10 1 1 <'Shape_i'>
82.0% 82.0% 0.021s 0.021s 2.13e-03s 10 1 1 <Gemv>
14.1% 96.1% 0.004s 0.025s 3.33e-04s 11 1 2 <Dot>
2.9% 98.9% 0.001s 0.026s 8.24e-06s * 91 10 10 <Elemwise>
0.6% 99.6% 0.000s 0.026s 1.69e-05s 10 1 1 <Alloc>
0.3% 99.9% 0.000s 0.026s 2.43e-06s * 31 2 4 <DimShuffle>
0.1% 100.0% 0.000s 0.026s 1.91e-06s * 10 1 1 <Sum>
0.0% 100.0% 0.000s 0.026s 1.19e-06s * 10 1 1 <Shape_i>
(*) Op is running a c implementation
\end{Verbatim}
\end{frame}
......@@ -743,16 +740,17 @@ Op-wise summary:
<% of local_time spent on this kind of Op> <cumulative %>
<self seconds> <cumulative seconds> <time per call>
<nb_call> <nb apply> <Op name>
31.4% 31.4% 0.000s 0.000s 1.93e-05s 10 1 Gemv{inplace}
16.9% 48.3% 0.000s 0.000s 1.04e-05s 10 1 Alloc
15.5% 63.8% 0.000s 0.000s 8.65e-06s 11 2 dot
5.0% 68.8% 0.000s 0.000s 3.05e-06s * 10 1 Elemwise{
Composite{scalar_softplus,{mul,scalar_softplus,{neg,mul,sub}}}}
4.3% 73.1% 0.000s 0.000s 1.27e-06s * 21 3 InplaceDimShuffle{x}
3.3% 76.4% 0.000s 0.000s 2.00e-06s * 10 1 Elemwise{sub,no_inplace}
2.9% 79.3% 0.000s 0.000s 1.79e-06s * 10 1 Elemwise{gt,no_inplace}
2.5% 84.5% 0.000s 0.001s 1.53e-06s * 10 1 InplaceDimShuffle{1,0}
... (remaining 9 Apply account for 18.3%(0.00s) of the runtime)
82.0% 82.0% 0.021s 0.021s 2.13e-03s 10 1 Gemv{inplace}
14.1% 96.1% 0.004s 0.025s 3.33e-04s 11 2 dot
1.4% 97.5% 0.000s 0.025s 3.63e-05s * 10 1 Elemwise{Composite{
scalar_softplus,{mul,scalar_softplus,{neg,mul,sub}}}}
0.6% 98.1% 0.000s 0.026s 1.69e-05s 10 1 Alloc
0.4% 98.5% 0.000s 0.026s 1.02e-05s * 10 1 Elemwise{Composite{
exp,{mul,{true_div,neg,{add,mul}}}}}[(0, 0)]
0.3% 98.8% 0.000s 0.026s 8.80e-06s * 10 1 Elemwise{
ScalarSigmoid}[(0, 0)]
0.2% 99.0% 0.000s 0.026s 2.40e-06s * 21 3 InplaceDimShuffle{x}
... (remaining 10 Apply account for 1.0%(0.00s) of the runtime)
(*) Op is running a c implementation
\end{Verbatim}
\end{frame}
......@@ -766,14 +764,14 @@ Apply-wise summary:
<% of local_time spent at this position> <cumulative %%>
<apply time> <cumulative seconds> <time per call>
<nb_call> <Apply position> <Apply Op name>
29.8% 29.8% 0.000s 0.000s 1.96e-05s 10 15 Gemv{inplace}
(<TensorType(float64, vector)>, {-0.1}, InplaceDimShuffle{1,0}.0, Elemwise{Composite{exp,{mul,{true_div,neg,{add,mul}}}}}[(0, 0)].0, {0.998})
15.8% 45.6% 0.000s 0.000s 1.04e-05s 10 10 Alloc(Elemwise{inv,no_inplace}.0, Shape_i{0}.0)
14.0% 59.6% 0.000s 0.000s 9.20e-06s 10 1 dot(<TensorType(float64, matrix)>, <TensorType(float64, vector)>)
5.6% 65.2% 0.000s 0.000s 3.67e-06s 10 9 Elemwise{Composite{scalar_softplus,{mul,scalar_softplus,{neg,mul,sub}}}}(<TensorType(float64, vector)>, Elemwise{Composite{neg,sub}}[(0, 0)].0, Elemwise{sub,no_inplace}.0, Elemwise{neg,no_inplace}.0)
3.2% 68.4% 0.000s 0.000s 2.12e-06s 10 4 Elemwise{sub,no_inplace}(TensorConstant{[ 1.]}, <TensorType(float64, vector)>)
2.9% 71.3% 0.000s 0.000s 1.93e-06s 10 12 Elemwise{gt,no_inplace}(Elemwise{ScalarSigmoid{output_types_preference=transfer_type{0}}}[(0, 0)].0, TensorConstant{[ 0.5]})
... (remaining 14 Apply instances account for 28.6%(0.00s) of the runtime)
82.0% 82.0% 0.021s 0.021s 2.13e-03s 10 15 Gemv{inplace}(
w, TensorConstant{-0.1}, InplaceDimShuffle{1,0}.0, Elemwise{Composite{exp,{mul,{true_div,neg,{add,mul}}}}}[(0, 0)].0, TensorConstant{0.998})
11.5% 93.4% 0.003s 0.024s 2.99e-04s 10 1 dot(x, w)
2.6% 96.1% 0.001s 0.025s 6.81e-04s 1 1 dot(x, w)
1.4% 97.5% 0.000s 0.025s 3.63e-05s 10 9 Elemwise{Composite{scalar_softplus,{mul,scalar_softplus,{neg,mul,sub}}}}(y, Elemwise{Composite{neg,sub}}[(0, 0)].0, Elemwise{sub,no_inplace}.0, Elemwise{neg,no_inplace}.0)
0.6% 98.1% 0.000s 0.026s 1.69e-05s 10 10 Alloc(Elemwise{inv,no_inplace}.0, Shape_i{0}.0)
0.4% 98.5% 0.000s 0.026s 1.02e-05s 10 13 Elemwise{Composite{exp,{mul,{true_div,neg,{add,mul}}}}}[(0, 0)](Elemwise{ScalarSigmoid{output_types_preference=transfer_type{0}, _op_use_c_code=True}}[(0, 0)].0, Alloc.0, y, Elemwise{Composite{neg,sub}}[(0, 0)].0, Elemwise{sub,no_inplace}.0, InplaceDimShuffle{x}.0)
... (remaining 14 Apply instances account for 1.5%(0.00s) of the runtime)
\end{Verbatim}
\end{frame}
......@@ -786,17 +784,18 @@ Profile of Theano functions memory:
(This check only the output of each apply node. It don't check the
temporary memory used by the op in the apply node.)
Theano fct: train
Max without gc, inplace and view (KB) 4
Max FAST_RUN_NO_GC (KB) 0
Max FAST_RUN (KB) 0
Memory saved by view (KB) 3
Memory saved by inplace (KB) 0
Max without gc, inplace and view (KB) 2481
Max FAST_RUN_NO_GC (KB) 16
Max FAST_RUN (KB) 16
Memory saved by view (KB) 2450
Memory saved by inplace (KB) 15
Memory saved by GC (KB) 0
<Sum apply outputs (bytes)> <Apply outputs memory size(bytes)>
<created/inplace/view> <Apply node>
3200B [3200] v InplaceDimShuffle{1,0}(<TensorType(float64, matrix)>)
800B [800] i Gemv{inplace}(<TensorType(float64, vector)>, TensorConstant{-0.1}, InplaceDimShuffle{1,0}.0, Elemwise{Composite{exp,{mul,{true_div,neg,{add,mul}}}}}[(0, 0)].0, TensorConstant{0.998})
32B [32] c Alloc(Elemwise{inv,no_inplace}.0, Shape_i{0}.0)
<created/inplace/view> is taked from the op declaration, not ...
2508800B [2508800] v InplaceDimShuffle{1,0}(x)
6272B [6272] i Gemv{inplace}(w, ...)
3200B [3200] c Elemwise{Composite{...}}(y, ...)
\end{Verbatim}
\end{frame}
......
Markdown 格式
0% 或
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！
请 注册 或者 后发表评论