IP: Bell v. Hillis: What about "Toy Story"?

[Dave Farber <farber () central cis upenn edu>]

Date: Wed, 20 Dec 1995 21:03:07 -0500
To: Dave Farber <farber () central cis upenn edu>
From: pcw () access digex net (Peter Wayner)


The Upside article made it seem as if Hillis lost completely.
While I don't know the terms of the bet nor do I know much about
the technology that created the big hit movie "Toy Story", I
wonder whether it might be the answer that Hillis needs.


For instance, there were probably more than 100 machines working
at once doing floating point calculations. They weren't in the
same box. They didn't have a fancy name like CM-5. But I would
conjecture that more many intents and purposes, massively
parallel rendering was going on at Pixar.


There are many cavaets. For instance, I would guess that many
teams of programmer/artists were working through different parts
of the movie at once. They could probably start and stop their
rendering job without affecting the other people. So some might
not consider it  "massively parallel." But what if there was one
final job that was done when everyone was finished with the
final colors and parameters? One final compile might win the
prize.


I'm sure that there are technical reasons why this might not
answer the bet. It certainly wasn't the type of massively
parallel jobs from the DARPA grand challenges. But what's the
difference?


Perhaps "Toy Story" proves Gordon Bell's point even better?
Pixar could have bought a CM-x, but it chose to cob together
many Sparc boxes. Massively parallel computing might continue,
but not at the price charged by the big box boys.


If any one has any insight or direct knowledge about processor
counts in the "Toy Story" build, I would like to hear them.


-Peter Wayner