Newsgroups: comp.parallel.mpi
From: xiaobai@cs.duke.edu (Xiaobai Sun)
Subject: Re: MCI or AT&T // MPI or PVM ?
Organization: Duke University Department of Computer Science, Durham, NC
Date: 21 Jun 1995 12:53:09 GMT
Message-ID: <3s94nl$h3g@duke.cs.duke.edu>

In article <3s822i$3tu@NNTP.MsState.Edu>,
Tony Skjellum <tony@aurora.cs.msstate.edu> wrote:
>your mail is not specific enough for anyone to give you
>a credible answer.  It seems like a blanket criticism, from
>someone who could benefit from some minutes using a Mosaic browser.
>

Glad to get attention from a major MPI contributor :) and hear
a criticism on some users who spend hours and days with MPI. :(
Hope the MPI developers can see the other side as well,
people who really criticize MPI don't even bother to use it.

>There is significant effort to document MPI ... look at
>http://www.erc.msstate.edu/mpi, for instance.
>

And another two in the circle :)

http://www.mcs.anl.gov/mpi 
http://www.osc.edu/lam.html

>There are several MPI implementations (for instance, vendors
>are implementing them).  This is to provide you with higher
>quality results.  You can read about that at the above
>URL, or elsewhere on the web.  Argonne/Mississippi State is
>one, LAM from Ohio State is another.  Commercial systems will
>be available too.  Much like the comparison of GNU software
>to vendor-specific compilers...  
>

>As to CM-5 problems, please refer an actual bug report to
>the developers.  MPICH bugs can be referred to mpi-bugs@mcs.anl.gov.
>We get reports and fix bugs and make improvements.
>

Yes, we have been sending reports of good and bad news. 

>-Tony Skjellum
-- 
Department of Computer Science, Duke University, Durham, NC 27708-0129
Internet:	USER@cs.duke.edu

