Author: Robert Hyatt
Date: 21:27:34 02/13/00
Go up one level in this thread
On February 13, 2000 at 23:30:15, Tom Kerrigan wrote: >On February 13, 2000 at 23:10:26, Robert Hyatt wrote: > >>On February 13, 2000 at 21:32:05, Tom Kerrigan wrote: >> >>>On February 13, 2000 at 20:40:03, Robert Hyatt wrote: >>> >>>>Do the math again. If you run at 1ghz (1ns) and drop the temp 40 degrees, >>>>and pick up 10 picoseconds, that doesn't make much of a difference. If you >>> >>>For some reason you seem to think that there's only one gate switch per clock >>>cycle... >>> >>>-Tom >> >> >>No I don't. I factored in up to 100 gates/clock, just for an upper limit >>and got 10%. 10% doesn't come anywhere near 33% that Kryotech is getting >>with the Athlon... > >Well, we know that the Athlon runs at 850MHz without special cooling. > >So if you figure a clock = 850MHz = 1.176 ns = 100 gates, then 1 gate takes >11.76 ps. > >To get to 1GHz, then 1 gate has to take 10 ps. > >So you're saying that a huge drop in temperature could not possibly lower the >switching time from 11.76 ps to 10 ps? > >Of course, this totally ignores the fact that AMD may be giving Kryotech >specially sorted chips that have no problems running > 850MHz. > >If I have time tonight, I might try to work out some of the math to see how much >of an effect temperature can have. > >-Tom Before you pop off, why don't you visit Kryo's web page? They claim to be using 750mhz parts. So they are overclocking by 33%. No I don't think dropping the temp 40 degrees will get anywhere near that. Otherwise we would be up to our armpits in such machines. Even new crays still use air-cooling. Reckon they would take advantage of that if it existed?
This page took 0.01 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.