Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Efficient hash algorithm?

Author: Robert Hyatt

Date: 10:31:24 09/22/98

Go up one level in this thread


On September 22, 1998 at 08:14:30, Pat King wrote:

>
>On September 22, 1998 at 05:52:43, Roberto Waldteufel wrote:
>
>> Maybe there should also be an "age"
>>entry in the tables based on the move-number at the root of the search when the
>>move was stored (or maybe last used)?
>>
>>Best wishes,
>>Roberto

I already do this, so that I can overwrite old stuff when needed, without
having to clear the table to make room for the next search...



>
>How about if you store the ply the position was encountered.  When a collision
>occurs, the new position replaces the old if...
>
>  1) old.ply < current.ply (old position now history)
>  2) new.ply < old.ply (new position nearer root)
>  3) new.ply = old.ply and new.depth > old.depth (new position has higher
>quality score)
>
>Right now, I'm using always replace, but I haven't been terribly pleased with my
>results, and the above seems to simply incorporate the ideas discussed in this
>thread.
>
>Pat



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.