Hello, MPI requires an overhead (of passing data between the processors, etc.). For short simulations you might not gain (wallclock) time at all, or even lose some.
Try a long / large simulation and see if it has any effect. Good luck and regards, Georg On Thu, Sep 22, 2011 at 3:17 PM, Mehdi Askari <ask...@shirazu.ac.ir> wrote: > > Dear all, > Does anyone know how I can run my simulation with a multiprocessor system > to get a faster run? > I have installed meep-mpi and mpich2 on my ubuntu. For a parallel meep > running I do steps below, > 1) I type "mpd &" in terminal > 2) I type "mpirun -np 8 meep-mpi test.ctl" in terminal. > It starts running the simulation with all 8 processors of my computer. > Everything seems to go well but unfortunately the simulation ends in a very > much longer time (about 10 times longer). I am confused, please help. :-( > Thanks. > Mehdi > > > _______________________________________________ > meep-discuss mailing list > meep-discuss@ab-initio.mit.edu > http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss >
_______________________________________________ meep-discuss mailing list meep-discuss@ab-initio.mit.edu http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
