Newsgroups: comp.parallel.mpi
From: bds@rzg.mpg.de (Bruce Scott TOK)
Subject: Re: MPI vs. shmem
Organization: Rechenzentrum der Max-Planck-Gesellschaft in Garching
Date: 31 Mar 1998 12:53:05 +0200
Message-ID: <6fqhuhINN4mv@s4bds.rzg.mpg.de>


I switched from shmem to MPI for two reasons:

1) In MPI the arrays don't have to be in the same place, or even a
definite place, on every PE.  If you want to use pointers or allocatable
arrays this is important.

2) Process groups.  I need them in order to do sums over planes in a 3D
code and to do maxima to find the Courant condition in regions that are
independent of each other.

I've only ever done this stuff on the T3D/T3E, so for me portability is
a lesser issue (this place is a Cray shop since at least the mid 80s).

As you know, however, everyone's needs are different.

cu,
Bruce

-- 
"That's the old ecological tale that explains humans' inability to fully
appreciate global warming.  To wit: if you drop a frog in a pan of hot
water, it jumps out.  If you drop it in a pan of cold water, then turn
the heat up slowly, you can roast it to death."  -- Clive Thompson