Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Dynamic move ordering for capture/promotions?

Author: Scott Gasch

Date: 11:55:58 04/03/01

Go up one level in this thread


On April 03, 2001 at 03:55:26, Gian-Carlo Pascutto wrote:

>The results seem to vary quite a bit. In silent positions it seems to hurt
>more than help, but I got slight speedups on some tactical positions.
>

Well, I spent quite a while reworking some move generator code last night to
facilitate some experiments.  I found the same thing as you did, I think.

1. Allow losing captures as killer moves: tree sizes increased and first move
beta percentage fell in normal positions.  I did find some rare positions where
this was beneficial, though.

2. Allow losing captures as killer moves if they beat beta by a certain margin:
same results... scaled with margin.

Looks like this is not such a great idea.  Back to wading through the output of
my non-first move beta cutoff dump for me.

Thanks,
Scott



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.