From: Sascha Ziemann <szi@aibon.ping.de>
Newsgroups: comp.parallel.pvm
Subject: PCM Comunication between processes on the same machine
Date: 06 Oct 1998 08:49:10 +0200
Organization: Aibon
Sender: szi@olivia.aibon.ping.de
Message-Id: <7ud886nq1l.fsf@olivia.aibon.ping.de>


Hi,

I would like to build a system for simulating video algorithms on top
of PVM. But after reading the PVM book there is still one question. A
system for video algorithms uses often blocks of the same size (for
example 720x576x8bit). And it is likely that many simulations will run
on only one host. It is important that the inter task communication is
done via shared memory to avoid copying the video frames, because I
would like to be able to create quite small simulation modules (for
example only a simple delay or an addition). Is the PVM communication
optimized in that way? Are there any throughput benchmarks for PVM3?

-- 
/* In the beginning was the Word: */
typedef long SCM;