Author: Drexel,Michael
Date: 21:05:27 11/20/03
Go up one level in this thread
On November 20, 2003 at 22:58:25, Dann Corbit wrote: >On November 20, 2003 at 22:51:04, Uri Blass wrote: > >>On November 20, 2003 at 22:35:43, Dann Corbit wrote: >> >>>as we often see cited. >>> >>>See: >>>http://www.talkchess.com/forums/1/message.html?329011 >> >>No >> >>Assume that all the program of the ssdf have 2500 at 100 mhz and 2570 at 200 >>Mhz. >> >>Your formula is going to give them 35 elo per doubling and not 70 elo. > >You are right. I divided by the ratio of the two speeds to normalize and forgot >to divide by 0.5 (or multiply by 2). > >70 is the better figure. OK,my random sample was not representative. I get 69.45 ELO per doubling, however all this math is not quite correct anyway. The result of Gandalf 5.0 is based on 102 games... 47 Gandalf 5.0 128MB K6-2 450 MHz 2538 73 -68 102 60% 2465 ...and the result of Deep Fritz is based on 1071 games for example. 25 Deep Fritz 128MB K6-2 450 MHz 2649 22 -21 1071 58% 2589 So this result is almost meaningless IMO. You can't ignore the number of games. Michael
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.