[GRASS-dev] Re: Parallelization of GRASS modules for GSOC 2009

[Markus Neteler]

On Thu, Mar 19, 2009 at 8:27 AM, Jyothish Soman
<jyothish.soman at gmail.com> wrote:
> I am a Masters student from International Institute of Information
> Technology-Hyderabad India, looking to participate in google summer of code
> under the OSGeo umbrella. More specifically, I want to work in the
> parallelizing GRASS modules using openMP (MPI is also fine).

That is a great idea - I am sure that many users would benefit from that.

> Slightly confused where to start poking in the code.
> Any suggestions on possible modules that have a good scope of coarse grained
> parallelization in it is welcome.

I guess that some kind of profiling is needed to see where time is
wasted. The "problem" is that much of GRASS' intelligence stays
in the modules. So heavy interpolation modules are certainly
candidates.
Also the vector library (topology building?) might be a candidate.

We could make a quick survey in grass-user to ask people where
they find GRASS slow.

> Please note: I am assuming that all the developers I have talked to earlier
> are already in this mail. If I missed anybody, please do inform me.

(all are in the list)

> Have been slightly busy yesterday, so didnt put up a full effort in going
> through the code, forgive me for that.

No problem at all! Here is also the (navigable) manual:
http://download.osgeo.org/grass/grass6_progman/

Please ask if you need specific answers.

Markus