[GRASS5] [bug #2155] (grass) integrating GRASS r.mapcalc with MPI

Helena hmitaso at unity.ncsu.edu
Tue Oct 14 12:09:35 EDT 2003

You can try to write to Christoph who has already written some GRASS code
for MPI
christoph.troyer at jrc.it
I put his parallel version of s.surf.rst at
I haven't submitted it yet as it needs more testing and I am not sure whether
it should go into 5.3 or 5.7 and it may also require some discussion
on developers list how to handle MPI code in general.


here are some of Christoph's notes:
I modified 'configure.in' such that configure would test whether MPI is
installed. Actually it tries to find 'mpicc', which comes with most
MPI implementations and which is a wrapper for cc. If you use mpicc
you don't have to worry about the location of your MPI libraries and
If mpicc exists the compiler will generate an additional executable
called 's.surf.rst.mp'.

To start the program on let's say 4 processors just type

    mpirun -np 4 s.surf.rst.mp [options]

on one of the machines in your cluster.

Request Tracker wrote:
> this bug's URL: http://intevation.de/rt/webrt?serial_num=2155
> -------------------------------------------------------------------------
> Subject: integrating GRASS r.mapcalc with MPI
> hi, 
> i was wondering... can i try to apply the MPI (message passing interface) in 
> grass.. i download the sample r.mapcalc from a gRASS website and i just 
> did some simple work as it takes a map and reproduce a thoroughly same 
> map as an output. i was thinking whether it is possible that i run that on a 
> beowulf cluster. but as i think further, i dunno how am i suppose to run the 
> program in the end. coz the grass need to be sort of compiled with gmake5 
> while the mpi needs to be run as mpirun -v filename. 
> -------------------------------------------- Managed by Request Tracker
> _______________________________________________
> grass5 mailing list
> grass5 at grass.itc.it
> http://grass.itc.it/mailman/listinfo/grass5

More information about the grass-dev mailing list