From: wjk@pluto.doc.ic.ac.uk (William Knottenbelt)
Newsgroups: comp.parallel.mpi
Subject: Re: Performance of overlapped communication and computation
Date: 9 Jan 1999 00:32:29 GMT
Organization: Dept. of Computing, Imperial College, University of London, UK.
Message-Id: <77682t$qq7$2@lux.doc.ic.ac.uk>
References: <76tkr7$efc$1@lux.doc.ic.ac.uk>
    <772v8p$8q7$1@nnrp1.dejanews.com>
    <3695C224.4557861A@mufasa.informatik.uni-mannheim.de>
    <77505v$749$1@lux.doc.ic.ac.uk> <775lro$t00$1@walter-fddi.cray.com>


Hi Eric

: Consider: you've got a 16-cpu machine running a 16-process MPI job, and
: one of the MPI processes calls MPI_Isend(). 

It struck me that I should mention that I'm talking about
distributed-memory parallel computers with dedicated network hardware
(like the AP3000 which is made up of UltraSPARC nodes connected by a
high-speed torus network called the APnet), not an SMP machine. I'm
not sure which you are talking about since I believe SGI are famous
for their excellent SMP machines but you could conceivably be talking
about distributed-memory parallel machines too.

Cheers
Will