Newsgroups: comp.parallel.pvm
From: brown@altair.krl.caltech.edu (C. Titus Brown)
Subject: Re: I don't believe PVM !!!
Organization: Avida Artificial Life group
Date: 18 Jul 1994 02:48:43 GMT
Message-ID: <30cqib$e8m@gap.cco.caltech.edu>

In article <30apsn$pgr@news.cis.nctu.edu.tw>,  <is80064@cis.nctu.edu.tw> wrote:
>   Hi every net-frient,
>      Currently, I am studying the PVM. But I don't find that
>   there is anything more convenient than the UNIX SOCKET.
>   Maybe I don't adapt it now. Could any net-frient analysis
>   what the advantage and disadvantage in the two. thankx.

I had the same question when I first started using PVM.  The thing that
PVM provides (or provided as of 2.something, I haven't been using it
recently) is convenience.  Sure, it has more features than you might
want to use (or anyone might want to use :); but it's convenient for
several reasons.

It's been ported to almost all the parallel machines I know about, thereby
allowing code to run on both networks of workstations and on tightly
connected parallel machines without a change to the code.

It can also work on machines that don't have sockets (or TCP/IP); this way,
you get another layer of abstraction 'tween the actual machine and thus don't
need to worry about annoying and unproductive things like whether your code
is compatible: if PVM has been ported, your code *will* be portable (comm,
at least :)  Obviously, if you're enamored of sockets, you may just want to
port *them* over, but then you could get into all sorts of icky problems.

Also, PVM can take advantage of the innate (hardware) communications
abilities.  I'm working on porting our Artificial Life system to PVM,
and it looks like most of the computer systems we'd be running on
have PVM already, and it uses native communications methods.  That means
that the program's more flexible on such machines, and I don't have to do
a thing to change it.

I'm sure other people have their reasons for PVM usage; these are mine!
(Oh, one more thing; it's FREE!!)

--Titus
-- 
C. Titus Brown <- http://www.krl.caltech.edu/~brown/plan.html -> brown@reed.edu
    --> GCS/GSS:@d--,-p+,c++++,l,u(++),e+,m+,s+/,n+,h+,f+,g+,w+,t-,r-,y? <--
            Sysadmin at Caltech KRL / Guest sysadmin at Reed College
               Member of the Avida Artificial Life research group

