Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Mate in 1 - but Fritz 6 needs 1 hour!!!

Author: Adrien Regimbald

Date: 08:31:10 08/13/00

Go up one level in this thread


Hello,

>Agreed.  I don't use guesswork pruning algorithms so that I know my analysis is
>correct. My method applies whatever searching algorithm/pruning you employ.  I
>use pv-enhanced negamax with a-b and hash tables etc... and so I can guarantee
>to see every mate down to a certain ply.  You don't have to do brute force - a-b
>pruning is tactically sound.


I was about to say "Yeah, I guess if you don't use anything like null move, etc
then you could be right" .. but then I saw that you use hash tables :P  I don't
think that I have to point out to you that the very fact that you're using hash
tables means you can't guarantee that your search is 100% inaccurate .. due to
possible collisions.


>All I'm claiming is that my method reduces the search time by a negligible
>amount *regardless of algorithms employ to actually do the search* and it
>occasionally results in enormous speed increases.
>
>Take this position for example;
>
>[D]4Q3/2p3p1/7p/p1k1B3/2p1P2P/2P3P1/1r2q3/6K1 w - -
>
>which ColChess searches *fully* to depth 13 ply in just over 3 seconds on my
>p3-450.  It is a mate in 9.  Crafty sees the mate quite early because of its
>advanced extensions, but after 30 seconds had still only just reached 11 ply in
>its search.  There are probably better examples where Crafty can't see the mate,
>and I unfortunately can't remember the board setup before this position, which I
>know to be one of them.


Well, that is a very impressive gain .. but I'm curious - does it hurt ColChess
overall to be doing this?


Regards,
Adrien.



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.