Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Some experiments

Author: Robert Hyatt

Date: 19:33:20 08/08/01

Go up one level in this thread


On August 08, 2001 at 16:05:25, Alvaro Jose Povoa Cardoso wrote:

>
>In my experiments I noticed that in many cases one sucessor is in the hash table
>and causes a cutoff on the next ply but it may not cause a cutoff on this ply
>(because at this ply I check if the score >= BETA and only return if this test
>is true).
>Can this possibly be correct?
>
>Thank you,
>Alvaro Cardoso


Correct, but inconsistent.  If, when you searched these positions, you could
save _both_ hash entries, then things would be ok.  But it is possible that
you overwrite something shallower by something deeper (same position different
depth) for one of them but not the other. That will cause some mischief.

Hashing is full of that kind of problem..

The fuller the hash table, the more of that you can see.



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.