Newsgroups: comp.parallel.mpi
From: laemmer@automatix.iib.bauwesen.th-darmstadt.de (Lutz Laemmer)
Subject: Re: MPI for transputers
Organization: IIB-TH Darmstadt
Date: 23 Apr 1996 08:39:29 GMT
Message-ID: <4li501$fsb@rs18.hrz.th-darmstadt.de>

In article ol4@webster.bfm.com, rmeenaks@bfm.com (Ram Meenakshisundaram) writes:
>   Yes I know that there are MPI for FORTRAN and C.  I don't think
>there is a OCCAM version.  I was basically looking for a native version
>specifically designed for the transputers.  A good start might be MPICH,
>which has the channel implementation of MPI.  Currently, I only know of
>one version for the Parix Operating environment for Transputers, however,
>several people reported that they had problems getting it to work.  As for
>trasputers themselves, INMOS-SGS introduced the ST450 transputer just 
>recently and the T9000 is still in production.  IMHO, I think they had a
>great product and blew it when they didn't offer a language other than OCCAM.
>Is it dead, no.  Is it going to die, yes and sooner than most people think.


Hi,

public domain MPIch for transputers with Parix (proprietary by Parsytec) 
operating system is provided by our site. It works fine for both PowerPC 
and transputer systems and uses the channel implementation of MPIch.

Linkable libraries for Parsytec compiler systems are available 
on request (send me an email) to avoid the errorsome installation 
and compilation of the sources in the mpich-distribution.

I am wondering, there is no transputer MPI implementation on top of the
virtual channel router (VCR) software by the University of Southampton, UK.


Lutz Laemmer


