On 22 August 2014 13:47, Ben Acland <[email protected]> wrote: > Is there a way to run permutation testing using multiple processes? >
Not at the moment, as far as I know. Currently we have parallelization in the searchlight and some of the surface-based voxel selection code, but the code is non-unified. A while ago I proposed a more unified parallellization approach: https://github.com/PyMVPA/PyMVPA/pull/109 but it would require extensive testing before it could be merged into master. If I remember correctly, some things broke when trying to parallelize repeated measures because of conditional attributes that were not properly passed through when pickling / hd5-ing the data. I don't know how this would play out for the MC code... Once we get to looking into the parallelization stuff, I'll try to make sure we have a look at the MC code as well.
_______________________________________________ Pkg-ExpPsy-PyMVPA mailing list [email protected] http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-exppsy-pymvpa
