Author: Robert Hyatt
Date: 17:53:46 07/11/03
Go up one level in this thread
On July 11, 2003 at 18:51:07, Grzegorz Sidorowicz wrote: > >>>Without null moves my program also can't find solution >> >>A couple of points. First, you ought to get to depth=25 _instantly_. You >>can try this position on Crafty to get a feel for how you are doing. I'd >>immediately inspect the hash probe and hash store code _carefully_ as it >>appears to be broken. >> >>For comparison, Crafty gets to ply=35 in 7 seconds on my PIII/750 laptop. >>It finds Kb1 at around depth=18/19 depending on hash size. If you don't >>find it by depth=26 you have a horrible bug. And if you don't get to depth >>26 instantly (my same laptop takes .5 seconds to get there) you also probably >>have a bad hashing bug. >> >>are you sure you are handling the bounds correctly when you store them and >>then look them up. Are you checking the draft right, using >= rather than >>> in the comparison? Etc. > >I'm sure I have got bug but maybe this bug is not in TT. No idea. It takes 26 plies. I don't think it can go that deep without hash tables to speed the search along. >How fast Crafty can find solution without TT. >In few days I will try debug it again but I can't find reason. >Today I was thinking about wrong random numbers. But now I'm sure >it is OK. >I will try debug it again and again....
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.