From: Mark Hampsey <markh@engmail.spamnewcastle.edu.au>
Newsgroups: comp.parallel.pvm
Subject: Re: PVM on a multiprocessor machine
Date: Tue, 18 May 1999 12:38:02 +1000
Organization: University of Newcastle, Australia
Message-Id: <3740D28A.C11D3143@engmail.spamnewcastle.edu.au>
References: <Pine.SGI.3.96.990517213801.9646B-100000@kali01.rz.fhtw-berlin.de>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Xref: ukc comp.parallel.pvm:8394


The standard answer is: "Start at least one process per processor".  This is
the easiest
way to spread your load over your available processors - each processor
works on its
own task.  I'm just new to this, so I'm sure you'll get some good answers
from gurus, but
this method works really well for me on my cluster of SMP machines.

Cheers,
Mark

Dirk Janke wrote:

> Hi,
>
> i have to write a diploma thesis about parallel data processing in
> visualisation on multiprocessor workstations. Do you know if PVM is
> suitable for using to this problem? PVM should give me the possibility to
> parallelizis the processing of geometric data on one single computer but
> with more than one processor!!!
>
> thanks a lot
>
> Kermit
> email: D.anke@rz.fhtw-berlin.de

