Newsgroups: comp.parallel.mpi
From: impellus@ames.ucsd.edu (Tom Impelluso)
Subject: Massively Vs Distributed
Organization: UC San Diego, AMES Department
Date: 8 Sep 1995 03:55:29 GMT
Message-ID: <42oerh$mpj@soenews.ucsd.edu>



Hi all

I just had dinner with some friends who gave me
information about what they thought was the future
direction of SGI.

In a nutshell:

SGI realizes that the Onyx will never make money.
The new Impact is/has (not quite sure: bear with me) an 
add on box that is quite small and can be hooked to a PC.

Thus, SGI realizes the future is going to be PC (my friends meant Intel)
based and that they will design all new equipment for that purpose.

They also said that there are new Intel PC chips designed with
multi-processing involved (he mentioned the P4 - Pentium chip - at
this point).

Now, I do work in Mech. Eng. and specifically, I envision the
necessity to bind Finite Element and Multi-Body Dynamics with
new imaging facilities such as Open Inventor.  To that end,
I worrry now about my research plan of using massively
parallel machines for the intense number cruching of the
physics (and if not Massivley parallel, than at least Onyx Based
multi-processing).

My friend now has me thinking to dis' the Onyx and T3D et. al.
in favor of multi-processing PC chips.

Could people please comment.  Could you give me your opinion
of what you think the Impact means both for the future of:
1) the Onyx
2) Massively Parallel machines
3) Distributed Processing

(regarding the last two, and this is specific for comp.parallel.mpi:
Do you think that MPP is a dinosaur?)


I DO NOT WANT A RELIGIOUS WAR STARTED.  If you can, please
just e-mail me your opinions.  I really am very interested in this.

Thanks,
Tom

