Author: Robert Hyatt
Date: 17:05:51 11/23/98
Go up one level in this thread
On November 23, 1998 at 18:04:03, Amir Ban wrote: >On November 23, 1998 at 17:01:04, Robert Hyatt wrote: > > >> >>I let this run for a while here on a single P6... I just finished 19 and my >>score for Qb6 is down to +.97, on the other machine, axb5 is still holding on >>at +.6, but that is one ply behind the machine with Qb6 and has stopped because >>of a power failure kicking that machine in the head. >> > >Huh ? If you play 36. axb5 axb5 you get the position for move 37, and I remember >that you posted here a week ago that you ran this on Crafty to get draw values >for this move, so where does this +.6 come from ? > >Amir I searched from the following position (after axb5 + black's move), ie the point where Qb6 or Be4 can be played... and I searched to either depth=21 or depth=22, which translates to 23 or 24 from axb5 position... It took a *long* time. It hasn't been running that long here. I don't recall the specifics now, but I ran that test on a P6/200 that was totally idle for a month, and I believe that it took multiple weeks.. It was far enough back in time that I don't remember specifics now... And note that "draw" is evaluation-specific... IE the side that is behind will go for the draw while the side that is "ahead" will avoid it if possible. This is a version of crafty 18 months separated from the version that was tested on the DB/Kasparov position. Who knows what has changed internally since then. My extensions are different, my evaluation is markedly different, my q-search is different, etc...
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.