Author: Ulrich Tuerke
Date: 04:17:11 01/19/99
Go up one level in this thread
On January 19, 1999 at 07:00:40, Harald Faber wrote: >On January 19, 1999 at 06:44:46, Ulrich Tuerke wrote: > >>>So now it is done, 40 games were played with the following result: >>> >>>Genius3-MCP7: 9.0-11.0 >>>MCP7-Genius3: 7.0-13.0 >>> >>>Sum: Genius3(!!) won 22-18. >>> >>>Conditions were: >>> >>>Genius3: AMD-K6-200, 32MB hash, tournament book (!) >>>MChessPro7.1: Intel166MMX, 40MB hash >>>(Genius3 had 50% speed advantage) >>> >>>Who would have expected this result? >> >>This is not really a surprise. Genius3 was rated about 2410 on P90 by SSDF. Add >>another 90 to scale to P200/MMX and you see that G3 and MCP7 are very close, and >>that's what your match seems to indicate. > >OK, if one adds 90 points. But it was not what one could have expected because >the jump from 486/66->P90 was less than 60 points! >I'd have expected 50 pts which makes 2460 and 166MMX reduce max. to 2480 (MCP) >and therefore a close win for MCP. > >If you add 90 pts (!) for Genius3 P90->200MMX, it would be higher rated than >Genius5 in SSDF. :-))) >(I know, only 4 points but not as worse as one could expect by the huge version >number difference that should indicate major changes). I admit that an increase of 90 is probably an overestimate. If I remember correctly, Thoralf Karlsson obtained an average increase of 75. Nevertheless, the resulting numbers for P200/MMX will be rather close. Don't forget that you'd have to play hundreds of games in order to reliably verify a difference of the order of 40/50 ELO points. Uli
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.