Author: Ed Schröder
Date: 04:48:10 01/19/00
Go up one level in this thread
>Posted by Tina Long on January 19, 2000 at 02:19:14: > >Pasted from the latest SSDF: 8 beats 9 by a point, 6 beats 7 by 5 points. > >My interpretation is "there was very little improvement in COMPUTER vs >COMPUTER (sorry to shout) strength from Rebel6 to Rebel 9." > >Perhaps this is because the Author was concentrating on improving his COMPUTER >vs HUMAN results, and preparing to pit his program against Grandmasters. > >All this conjecture would be clearer if both Rebel 10b and Rebel 10c had both >appeared on SSDF (although I understand and accept why they're not on SSDF). > >Perhaps the results of these Rebel's are inaccurate due to Autoplayer problems >(Ed?, Bertil?) That is very hard to say. The damage is proven that's for sure. How much? Impossible to tell. >29 Rebel 8.0 Pentium 90 MHz 2438 18 -18 1435 55% 2403 >30 Rebel 9.0 Pentium 90 MHz 2437 23 -23 890 47% 2458 Hundreds of Rebel9 games were played with the wrong settings: 1) without Rebel's main-book 2) without Rebel9's learner Also Rebel9 had a big bug in the hash table code. The effect was that the bigger you made the hash table the worse Rebel9 (and previous Rebel's) would play. Rebel9 with 13Mb played a lot better than with 28, 40 or 60 Mb in fact 13 Mb performed best. Fixed since version 10c. Ed > 31 Hiarcs 6.0 Pentium 90 MHz 2434 18 -18 1437 51% 2430 > 32 Genius 5.0 DOS Pentium 90 MHz 2431 18 -18 1558 47% 2450 > 33 Hiarcs 5.0 Pentium 90 MHz 2427 38 -36 363 59% 2362 > 34 MChess Pro 6.0 Pentium 90 MHz 2412 17 -17 1726 45% 2445 > 35 Genius 3.0 Pentium 90 MHz 2408 26 -25 785 61% 2328 > 36 Rebel 6.0 Pentium 90 MHz 2407 31 -30 540 60% 2335 > 36 MChess Pro 5.0 Pentium 90 MHz 2407 26 -25 769 63% 2315 > 38 Genius 4.0 DOS Pentium 90 MHz 2404 23 -23 944 60% 2335 > 39 Rebel 7.0 Pentium 90 MHz 2402 26 -25 759 61% 2319 > >Tina Long
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.