[pymvpa] Time for single permutation varies between a few seconds and several minutes

[Yaroslav Halchenko]

On Wed, 08 Apr 2015, Jan Derrfuss wrote:

> Hi all,

> I'm stumped. I'm running a permutation for a searchlight analysis
> and print the time when the individual permutations start. I turns
> out that the onset times are sometimes mere seconds and sometimes
> several minutes apart.

> I've played around with setting or not setting nproc, and with
> varying the number of nproc. With more processors, the times
> decrease, but can still amount to minutes. Just as an example, with
> nproc=20 a permutation usually takes about 14 s, but sometimes it
> might also take 4-6 minutes:
> >   Running permutation 0 11:39:27
> >   Running permutation 1 11:39:40
> >   Running permutation 2 11:40:04
> >   Running permutation 3 11:44:30
> >   Running permutation 4 11:50:52
> >   Running permutation 5 11:51:08
> >   Running permutation 6 11:51:21
> >   Running permutation 7 11:51:36
> >   Running permutation 8 11:51:50
> >   Running permutation 9 11:52:08

to make it clear -- are you running permutations at the level  of entire
searchlight measure or  searchlight of measures with permutations?

sharing code snippet could be of help to troubleshoot/replicate.  It
would all "depend" on specifics, right?

> Interestingly, even when the permutation takes several minutes, top
> shows 20 processors being busy.

so good -- it is crunching ;)
-- 
Yaroslav O. Halchenko, Ph.D.
http://neuro.debian.net http://www.pymvpa.org http://www.fail2ban.org
Research Scientist,            Psychological and Brain Sciences Dept.
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
Phone: +1 (603) 646-9834                       Fax: +1 (603) 646-1419
WWW:   http://www.linkedin.com/in/yarik