[SciPy-user] How to profile scipy code ?

[David Cournapeau]

Hi there,

    I was wondering if other people had any insight on how to 
efficiently profile scipy code. What I want is simply being able to see 
the bottleneck of one algorithm, see if it can be easily improved (using 
other algorithms in scipy, or using C extensions, etc...), and see the 
impact of the improvement.
    Under matlab, the profiler, quite rudimentary, works perfectly for 
this task; for people who don't know matlab's profiler, to profile one 
script, you simply do

profile on;
function_to_profile()
profile off;
profile report

profile report call creates a html report. The report gives cumulative 
time, and you can easily browse to the code of any called function (as 
long it is a matlab file, and not a "native" compiled file).

Is there anything like this for python ?

David