Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: futility pruning?

Author: Koundinya Veluri

Date: 09:01:24 11/10/03

Go up one level in this thread


On November 10, 2003 at 05:24:08, Daniel Shawul wrote:

>If I am following you correctly,you do the futility cut off before making the
>move.How do you intend to figure out whether it is a cheking move or not?
>Because I think it is very hard to determine revealed checks(indirect checks
>by pieces other than the moving piece).

It's very easy with bitboards. The last time I tested the difference in speed
between checking for illegal move (move into check) before or after making the
move, I found that it was faster to check before making the move.

>Doing the futility before making the move
>may cost you some time.So what I think is easy is to make the move and see if it
>is a checking move.And if you apply the futility after you make the move,there
>is no saving in make/unmake.So what I am saying is FP at frontier is
>gonna cost you more than its advantage so it's better to rely on the stand pat
>cutoff.

I think the main advantage of futility pruning at frontier nodes is saving an
eval.

IMO the whole point of SEE-related pruning is based on the assumption that SEE
is much faster than eval. That is especially needed for futility pruning at
frontier nodes. If you use lazy eval or any shortcut that speeds up the eval by
a lot (and it works well) then most likely there will be no advantage (maybe
even a disadvantage) from using this pruning, and maybe not even much advantage
from using SEE pruning in quiescence search. In King of Kings, the eval crawls
and lazy eval is impossible because of the high changes in evaluation, so
SEE-related pruning techniques help a lot.

Regards,
Koundinya



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.