Author: Uri Blass
Date: 13:44:05 05/01/02
Go up one level in this thread
On May 01, 2002 at 15:48:23, Christophe Theron wrote: >On April 30, 2002 at 13:43:42, Uri Blass wrote: > >>On April 30, 2002 at 13:36:18, Uri Blass wrote: >> >>>On April 30, 2002 at 13:32:33, Uri Blass wrote: >>> >>>>On April 30, 2002 at 13:09:42, Jorge Pichard wrote: >>>> >>>>>On April 30, 2002 at 12:54:31, Peter Berger wrote: >>>>> >>>>>>On April 30, 2002 at 11:31:50, Christophe Theron wrote: >>>>>> >>>>>>>Match it against Comet and you might be even more surprised. >>>>>>> >>>>>>>And after that we will talk again about your supposed 350 elo points difference >>>>>>>between the two and the supposed superiority of knowledge. >>>>>>> >>>>>>> >>>>>>> >>>>>>> Christophe >>>>>> >>>>>>Judging by the SSDF results of Fritz 1 >>>>>> >>>>>>148 Fritz 1.0 486/33 MHz 1941 48 -47 215 55% 1908 >>>>> >>>>>I get 2007 on a 386 25 Mhz 1996 SSDF List >>>>> >>>>>http://www.geocities.com/CapeCanaveral/Launchpad/2640/ssdf/1996/ssdf9612.htm >>>>> >>>>>2041 on this 1998 SSDF List: >>>> >>>>These numbers of the old lists should be resuced by 100 elo because the ssdf >>>>reduced 100 elo from all programs. >>>> >>>>I believed peter about the rating above 2000 of Fritz1 and this is the reason >>>>that I tried to use bugs to explain possible 350 elo difference. >>> >>>sorry. >>> >>>It should be probably someone else(maybe you but I am not sure now) >>> >>>I see that peter said in another post that 350 elo is a good guess based on the >>>1941 and not based on the over 2000 elo but I remember that I used bugs because >>>of reading the over 2000 estimate. >>> >>>Uri >> >>I can add that it may be interesting to test Fritz1 against palm tiger. >> >>Fritz1 has 1941 on 486/33 Mhz. >> >>If we assume that palm tiger has 2200 and if we assume that Fritz1 earns 50 elo >>per doubling of the speed then it means that Fritz1 needs hardware that is more >>than 32 times faster than 486/33 Mhz to get 50%. >> >>Uri > > > >Doubling the speed generally results in a 70 elo points increase. > > > > Christophe I believe that often one of the advantages of good program relative to significantly weaker programs is that they earn more from doubling of the speed. The results of Comet against Fritz1 suggest that 350 elo difference is not a bad guess. Fritz1 is losing 5.5-2.5 against Comet inspite of that fact that the hardware of comet is significantly faster(seems to be 4 times faster based on the Mhz but I know that Mhz means nothing). 250-300 elo difference could be a better guess based on that result but the statistical error is too high to say that 350 is wrong. Uri
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.