Newsgroups: comp.parallel.pvm
Path: ukc!uknet!newcastle.ac.uk!eata!nrk
From: R.Kerr@newcastle.ac.uk (R. Kerr)
Subject: Re: experiences with PVM?
Nntp-Posting-Host: eata
Message-ID: <CnFIA8.9AK@newcastle.ac.uk>
Organization: University of Newcastle upon Tyne, UK, NE1 7RU
References: <CnEoHG.F3w@knot.ccs.queensu.ca> <1994Mar29.064838.45829@rs6000.cmp.ilstu.edu>
Date: Tue, 29 Mar 1994 13:46:56 GMT
Lines: 27

eschandl@katya.acs.ilstu.edu (Eric Scott Chandler) writes:

>	PVM seems to be at the forefront in heterogeneous network
>implementation.  I suggest that you also look at other systems as well. 
>Maybe you will find them "better" in your own opinion.  I'm not a PVM
>advocate, or anything.  Try also: p4, Parallaxis, C-Linda (Suns only), and a
>few more that I cant remember off hand.

Linda is definitely worth a look. Instead of being a bunch of
message passing and synchronisation subroutines it represents a
simple but versatile parallel programming model and so shifts the 
programmer's emphasis from the mechanics of message construction
and transmission to the logic of the interaction of the parallel
components. In so doing, it is also much less error-prone.

Linda versions of both C and FORTRAN are available on considerably
more systems than the Suns, mentioned above, and we have 
successfully used it with C++ as well (and Pascal through automated
translation to C).

Cheers....Ron

------------------------------------------------------------------------
  Ron Kerr, Computing Service, Newcastle University, NE1 7RU, England.
            Tel. +44 91 222 8187       Fax. +44 91 222 8765 
------------------------------------------------------------------------


