Author: Matthew Hull
Date: 10:49:54 12/03/04
Go up one level in this thread
On December 03, 2004 at 09:50:55, Uri Blass wrote: >On December 02, 2004 at 23:44:59, Matthew Hull wrote: > >>On December 02, 2004 at 17:21:21, Wilhelm Hudetz wrote: >> >>>Hi all, >>> >>>what are your results with different Crafty versions? >>>19.08SE give me at the moment the best results, so I run a small >>>tourney to compare a little bit. >>>All SE-versions run with SE=on and standard personality. >> >> >>Two 100 game matches: >> >>Time control 5 0 >>xboard: Match Crafty-19.16 vs. Crafty-19.17: final score 2-5-93 >> >>Time control 30 0 >>xboard: Match Crafty-19.16 vs. Crafty-19.17: final score 24-23-53 >> >>In the second match, 19.17 was down 6 full points with 15 games to go. That's >>what I call a rally. >> >> >>Duron 900mhz 512m ram, ponder on, learning on. >>hash 96m >>hashp 24m >>book width 99 >> >>The 30 0 match would be similar to a blitz match on more up-to-date hardware. >> >>Not a lot of difference between these two. Don't even mess with versions prior >>to 19.15 as these contain a significant endgame mis-evaluation (read main.c for >>more details). > >It seems that Crafty19.17 under winboard contains more mis-evaluations and I >remember that Mike byrne found that Crafty19.17 shows different output in text >mode and under winboard. During the matches on the Linux box, I tailed the log files for both programs. The eval's differed by as much as a 1/2 pawn. However, Crafty uses assymetric evaluation terms, which accounts for most of the delta. But Bob did say 19.17 "corrected" some evals that were out of balance. But it's hard to tell the difference between them. 19.16 seems slightly more aggressive, but 19.17 plays good "comeback" chess. > >Unfortunately that post is too old to find it and it seems it was something like >10 days ago. > >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.