Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: MTD(f) and hash table size

Author: Uri Blass

Date: 14:17:50 08/17/03

Go up one level in this thread


On August 17, 2003 at 16:24:24, Dieter Buerssner wrote:

>Dan, did you also try with 2 bounds/scores per entry? I think, 2 bounds per
>entry could make a depth preferred scheme looking more attractive.
>
>I try to give another reason, why replace always should be preferred. Assume the
>situation you have one pos with high draft (more than search depth), but you
>need to search the position. The only reason can be, that the TT entry has the
>wrong bound (otherwise you would cutoff, and not search). Now you search to end,
>and try to store in the TT. With depth preferred, you will not store it (after
>all, the pos had a high draft in the beginning). You will keep that "wrong"
>bound, that is basically useless now (but has an attractive looking draft).
>
>Regards,
>Dieter

After reading it I think that it is possible
that is better for me to use hash table for pruning in some simple productive
way.

I cannot do it when the score is bigger than beta without risks but I think that
I can safely do it when the score is bigger than beta+something.

replace always when depth is bigger or equal may be also replaced by replacing
also when the new bound is significantly bigger than the old bound.

Uri



This page took 0.05 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.