[GRASS-dev] Re: parallelizing GRASS modules

Hamish hamish_b at yahoo.com
Mon Dec 5 09:37:02 EST 2011


Markus Metz wrote:
> IOW, there may be other modules that are easier to optimize using
> parallelization than the ones using the segment library.

what do you think about the recently optimized v.surf.bspline module?
Thanks to Sören the G_math_solver_cholesky_sband() is now parallelized,
which helps, but I think it would be more efficient to run each of the
slightly-overlapping subregions in their own threads, then wait for all
subregions to finish before continuing on. Technically the segment
library is involved, but the subregion bit of the code seems somewhat
separated from the segmenting parts.


fwiw I don't really consider parallelization to be "optimizing" anything,
rather just a way to throw more brute force at the problem.


> Attachment FYI

very impressive.  the v.surf.icw addon script thanks you.


Hamish



More information about the grass-dev mailing list