Newsgroups: comp.parallel.mpi,comp.parallel.pvm
From: tony@aurora.cs.msstate.edu (Tony Skjellum)
Subject: Re: MPI or PVM?
Organization: Mississippi State University
Date: 12 Dec 94 19:05:45 GMT
Message-ID: <tony.787259145@aurora.cs.msstate.edu>

mucci@cs.utk.edu (Philip Mucci) writes:

>	Hmmm...

>	Perhaps we should submit a request for comp.parallel.advocacy 
>for posts such as this. (and the inevitable fires they ignite.)

>	I wonder if I should post my list: "The Top Ten Things MPI can't do"
>...nah..I'll just give you a clue. #1 is "Start a new process"...

>	PVM is intended for use at a higher level than MPI. In fact, in the
>future you should expect to see ports of MPI written on top of native MPI 
>implementations. And yes, PVM is not written in stone, but evolves 
>as the need arises. (Try getting a new feature added to MPI and 
>see how far you get). In short, we're talking apples and oranges here folks...
>Just look at the level of functionality each provides and it'll become 
>quite apparent. (ok..maybe not oranges... pears maybe... :-) 

>-- 
>Philip J. Mucci (mucci@cs.utk.edu) | Internet, The last free media
>GRA in CS, U of Tenn, Knoxville    | George Washington grew hemp!

Please post your list of top ten items!  You are teasing us.

And I think your arguments above actually support the MPI point of view.
-Tony

--
Anthony Skjellum, Asst. Professor, MSU/CS/ERC, (601)325-8435; FAX: 325-8997.
Mosaic: http://www.erc.msstate.edu/~tony; e-mail: tony@cs.msstate.edu
	.	.	.	.	.	.	.	.      .
"There is no lifeguard at the gene pool." - C. H. Baldwin

