From: A.Dergatchev@tn.utwente.nl (Andrei A. Dergatchev)
Newsgroups: comp.parallel.mpi
Subject: Questions using MPICH on SunOS 5.6 ?
Date: Tue, 27 Jul 1999 15:17:29 GMT
Organization: University of Twente, Enschede, The Netherlands
Message-Id: <379dc078.2968312@news.nic.utwente.nl>
Xref: ukc comp.parallel.mpi:5381


Hi,

I've encountered a seemingly strange (for me) problem
working with speeding up a simulation package. I've
compiled the MPICH package on Sun E3500 under
SunOS. The machine has 5 CPUs, and I expected
to get a considerable increase in speed as calculations
are highly parallized. However, currently I see a strong
*increase* (~50%) in computation time going from 2
to 4 processors ("mpirun -np 2" -> "mpirun -np 4").

Why it would be ? Currently I'm thinking about trying
PVM package - does any one have any experience comparing
their (with MPICH) speed on multiply-processors machines ?
Or maybe there's other package (except native Sun HPC)
which might be better suitable for a case of calculations
on 1 multimple-processor computer (not cluster of
workstations) ?

Thanks for any hints,
Rgds,

Andrei