From: evenson@hitl.washington.edu (Mark Evenson)
Subject: Re: SCI: Networked VR Models
Date: 23 May 93 00:58:00
Organization: R+D Hipsters @ virtual HITLab, Ithaca


In article <1tk548INNrsk@news.u.washington.edu> Jesse Schell <jscn+@andrew.cmu.edu> writes:

   I am currently involved in a research project attempting to develop
   mathematical models of networked VR systems, so that there might be a
   good way to compare different network architectures (client-server,
   etc.) without actually spending thousands of dollars to find out that
   you can't get the performance you desire. 

   In order to do this, we needed to figure out some performance metrics. I
   think we have a good list, but I was wondering if anyone could think of
   more.

[...]

   The four listed above we feel to be the key items by which someone would
   judge a networked VR system. If anyone has other ideas about the factors
   that they would use to judge a networked VR system, please email me. If
   there is sufficient interest, I can compile a list of responses. 

The paper that Mike Almquist just delivered to 3CyberConf called "Lessons
from the Gibsonian Matrix" has some numbers you might be interested in.  He
apparently did some projections based on current typical TCP/IP limitations
and showed how many workstation nodes could be operative.  He apparently
based his numbers on some simulations done on SGI hardware.

He's a hard man to reach these days, but he published this paper through
his affiliation with HITL, so you could probably (eventually) reach him at
<squish@hitl.washington.edu>.

N.B.  this is all hearsay.  I have seen neither the abstract nor the paper.
And I'm looking for a copy as well . . .

		so much information, so little duplication