[matplotlib-devel] Profiling. was Re: Experimental, traited config module available in svn

[Eric F. <ef...@ha...>]

Fernando,

Here is some more information about profiling that others may find useful:

1) The profile module reports times that are much longer than normal; I 
don't know to what extent the relative times are still correct; my 
impression is that they do still give a decent idea of where the 
bottlenecks and resource hogs are.

2) The combination of hotshot and hotshot2calltree (or similar) gets 
badly confused by duplicate function names such as __init__, so the 
kcachegrind view can be quite misleading.

3) A solution is to use the cProfile module in Python 2.5 with an 
appropriate reformatter.  The following script runs a python script 
through cProfile and generates a *.log file that kcachegrind reads:

http://www.gnome.org/~johan/lsprofcalltree.py

So the sequence is simply

python lsprofcalltree.py testscript.py
kcachegrind testscript.py.log &

Here are a couple of relevant Google results:

https://lists.dulug.duke.edu/pipermail/yum-devel/2007-January/003045.html
http://ddaa.net/blog/python/lsprof-calltree

Eric

Fernando Perez wrote:
> On 7/31/07, Darren Dale <dd...@co...> wrote:
> 
>> I am attaching the results for two different tests. One is loading the default
>> mpl-data/matplotlib.conf, the other is loading an empty
>> ~/.matplotlib/matplotlib.conf. I haven't done much work with profiling, maybe
>> others can make some comments about the results.
> 
> I'm afraid I can't help with this right now, but you may want to look at
> 
> http://amath.colorado.edu/faculty/fperez/python/profiling/
> 
> the little code I put in there makes using kcachegrind very easy for
> python profiling, and it's really a useful tool for analyzing
> profiling info.
> 
> Cheers,
> 
> f