pokey@well.UUCP (Jef Poskanzer) (11/09/88)
I've been thinging about how many CPU cycles the worm used in its
short but sweet rampage. People have been tossing around the number
6000 for systems infected. (Where did this number come from?) Let's
say these machines averaged 2 MIPS each, and wildly guess that the
average length of infection was a day. That comes out to (kachunk
kachunk) one quadrillion instructions. This may well be the most CPU
power expended so far on a single problem.
Now, if we could harness that kind of power for useful work... Imagine
a net-wide distributed ray-tracing service. If you want something traced,
you send the NFF to a central clearinghouse. All over the net, people
start ray tracing servers when they leave for the day. These servers
connect up to the clearinghouse and fetch the scene description, and a
list of pixels to trace. By morning, hundreds of scenes have been traced
and returned to their senders. (Close your mouth, you're drooling.)
By the way, I spent Sunday doing the PostScript artwork for a worm
t-shirt. Anyone interested in ordering one?
---
Jef
Jef Poskanzer jef@rtsg.ee.lbl.gov ...well!pokey
"Why should we subsidize intellectual curiosity?" -- Ronald Reagandesnoyer@Apple.COM (Peter Desnoyers) (11/10/88)
In article <7600@well.UUCP> pokey@well.UUCP (Jef Poskanzer) writes: > [estimate of CPU cycles used by virus] > >Now, if we could harness that kind of power for useful work... Already been done. DEC just factored a huge number this way in the last month or two. Made the papers and everything. Peter Desnoyers
rds95@leah.Albany.Edu (Robert Seals) (11/10/88)
In article <7600@well.UUCP>, pokey@well.UUCP (Jef Poskanzer) writes: > Now, if we could harness that kind of power for useful work... Imagine > a net-wide distributed ray-tracing service. If you want something traced, > you send the NFF to a central clearinghouse. All over the net, people > start ray tracing servers when they leave for the day. These servers > connect up to the clearinghouse and fetch the scene description, and a > list of pixels to trace. By morning, hundreds of scenes have been traced > and returned to their senders. (Close your mouth, you're drooling.) I guess most of us [well, at least the groovier of us, anyway...] have thought about this. Do we all need RPC? Why aren't we doing it now? They do it at CMU, and apparently at DEC and elsewhere, so why not 'distributed ray-tracing (or luminace modelling, or acid deposition modelling, or...) for the rest of us'? I'd contribute my meager resources each night... honest, rob