Author: Robert Hyatt
Date: 13:57:17 12/13/99
Go up one level in this thread
On December 13, 1999 at 15:38:10, paul bedrey wrote: >On December 13, 1999 at 13:25:04, Robert Hyatt wrote: > >>On December 13, 1999 at 11:28:00, James Robertson wrote: >> >>>On December 13, 1999 at 08:08:38, CLiebert wrote: >>> >>>>Crafty 17.04 running the gauntlet >>>> >> >>>Crafty 17.4 is, at least from my preliminary results _way_ stronger than 17.3. >>>Insomniac was scoring close to 50% agianst 17.3, and the latest result was 0% >>>against 17.4. :( >>> >>>James >> >> >>Remember that I posted one of the bugs. The score for a pawn on the 7th rank >>was negative, rather than positive. It would therefore never advance to the >>7th, unless it could see it could also go to the 8th and promote. And if you >>got one on the 7th, it would be _very_ happy, until it finally saw that you >>could promote it, which was often too late. Even blockading the pawn on the >>7th was very bad because of this. >> >>Totally nuts... and it had a horrific effect on performance... > >Funny but Crafty 17.3 on my machine has yet to lose a 16 game match to any of >the other Winboard engines. Bob, is Crafty tested at any specific time control? >I'm running at 40 moves in 20'. I'll download 17.4 and run a match against 17.3 >Also do you think the Nunn test is a good indicator of engine strength? If not >what is? This bug didn't hurt against _some_ programs. If the opponent doesn't know a lot about passed pawns, then this wouldn't get beat on... but if they did... As far as testing, I don't pay much attention to comp-vs-comp games as that is not my major interest at present. I pay _far_ more attention to human games as they offer significant problems that have yet to be solved well by anybody.
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.