Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!rutgers!ames!lll-tis!ptsfa!ihnp4!inuxc!iuvax!pur-ee!uiucdcs!uiucdcsm!grunwald
From: grunwald@uiucdcsm.cs.uiuc.edu
Newsgroups: comp.arch
Subject: Re: chewing up mips with graphics
Message-ID: <3300008@uiucdcsm>
Date: Fri, 19-Jun-87 11:05:00 EDT
Article-I.D.: uiucdcsm.3300008
Posted: Fri Jun 19 11:05:00 1987
Date-Received: Mon, 22-Jun-87 05:35:19 EDT
References: <8270@amdahl.amdahl.com>
Lines: 13
Nf-ID: #R:amdahl.amdahl.com:8270:uiucdcsm:3300008:000:619
Nf-From: uiucdcsm.cs.uiuc.edu!grunwald    Jun 19 10:05:00 1987


I have to agree with Eugene Miya on the ``ray tracing is trivally parallel only
if you do trival things''.

It's true that you can just sub-divide your view space. I moved a ray tracer
to a hypercube and got almost linear speedup. However, the ``trival'' part
dies soon. Doing larger scenes (which is what's desired anyway)
requires sub-dividing your data-space as well. Suddenly, your problem is
no longer as simple as it used to be & you start to make trade-offs between
replication of data & how much data & speed-up & messaging.

But, an hour or two is faster than waiting a day for the vax to crank out a
picture.