From: "Alexander N. Andreyev" <alexander@vvv.srcc.msu.su>
Newsgroups: comp.parallel.mpi
Subject: Building MPICH on SMP: ch_shmem or ch_p4 (?)
Date: Thu, 29 Apr 1999 16:37:07 +0400
Organization: SRCC Nonlinear Computation Lab InterNetNew site
Distribution: world
Message-Id: <7g9jq9$trq$1@vvv.srcc.msu.su>
Mime-Version: 1.0
Content-Type: text/plain; charset="koi8-r"
Content-Transfer-Encoding: 7bit
Xref: ukc comp.parallel.mpi:5011


I am building MPICH on an SMP machine.There are two methods of configuring
MPICH on multiprocessors, proposed in documentation:

1) device=ch_shmem
and 2) device=ch_p4, comm=shared

What are the differences in MPI communications between these two modes?
Which one should perform better?

Should I expect message transfer rates be close to memory copy speed, as
measured, for example, by the STREAM benchmark?

Thanks in advance,
--
Alexander N. Andreyev
Laboratory of Parallel Information Technologies, SRCC, MSU
alexander@vvv.srcc.msu.su, http://alex.motor.ru, ICQ: #3523091
Parallel Computing: http://parallel.srcc.msu.su [russian language]