Author: Dave Gomboc
Date: 21:12:32 09/05/01
Go up one level in this thread
On September 05, 2001 at 12:18:04, Roy Eassa wrote: >On September 05, 2001 at 06:27:53, Sune Fischer wrote: > >>On September 05, 2001 at 01:37:25, Slater Wold wrote: >> >>>>>So we're a little less than 9 years away. :) >>>>> >>>>> >>>>>Slate >>>> >>>>Not quite according to my calculator: >>>> >>>> 64 GHz in 90 months ... and then comes NOT 1.28 THz in 108 months, >>>>but rather: 128 GHz in 108 months >>>> 256 GHz in 126 months >>>> 512 GHz in 144 months >>>> 1 THz in 162 months (about 13.5 years from now) >>> >>>You're right. 1000Ghz is 1.0Thz. >>> >>>Forgive me. I post while sometimes still asleep. >> >>Hmm, I get 14.22 years if I assume the doubling time is 1.5 years and current >>speed is 1.4 GHz. >> >>time(speed) = ln(speed/1.4GHz)*(T2/ln2), where T2=1.5 years. >>Now put in speed = 1000 GHz and you get 14.22 years. >> >>You could also sove two equations with two unknowns: >>x = time/1.5 years >>1000 GHz = 1.4 GHz * 2^x >> >>here I get x=9.48 and so time = 9.48*1.5 years = 14.22 >> > > >I think Moore's law used "18 - 24 months", not strictly 18 months. Perhaps >using the average, 21 months, is more reasonable going forward? (After all, PC >sales are down for the first time EVER, and sales drive everything else. This >slump we're in smells different from any other thus far and may signal a slope >change in the curve. Nothing continues unchanged forever!) When first stated, it was 24 months. The processor speed increase since that time has actually been superexponential. Dave
This page took 0 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.