Path: utzoo!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!magnus.acs.ohio-state.edu!tut.cis.ohio-state.edu!ucbvax!ucsd!ogicse!milton!dormer@cs.purdue.edu
From: dormer@cs.purdue.edu (John Dormer)
Newsgroups: sci.virtual-worlds
Subject: Re: VR on a CM or other Parallel Machine
Message-ID: <18587@milton.u.washington.edu>
Date: 17 Mar 91 18:26:03 GMT
References: <18543@milton.u.washington.edu>
Sender: news@milton.u.washington.edu
Organization: occasional on alternate Tuesdays
Lines: 21
Approved: cyberoid@milton.u.washington.edu



In article <18543@milton.u.washington.edu> darken@eesun.gwu.edu (Rudolph Darken)
 writes:
>
> Would a massively parallel machine like the CM offer any added
>speed that would  (or could) significantly improve a VR environment? 
>
>Rudy Darken
>darken@seas.gwu.edu

  And for another question, has anyone played with getting more information 
out of such machines? They are quite excellent numeric devices, but I have
seen examples of high-bandwidth requiring programs which die because they
can't get to the external hardware as quickly (reading data files from a
harddrive). With all those CPUs, certainly some of them could be made into
IO processors (or told to be IO processors). I don't know enough about the
CM hardware...perhaps one of you does?

        John Dormer
        dormer@medusa.cs.purdue.edu