Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: curious Rybka behaviour

Author: Kurt Utzinger

Date: 13:25:00 01/27/06

Go up one level in this thread


On January 27, 2006 at 16:13:20, Bernardo Wesler wrote:

>Kurt: do u say that because of a pragmatic thought or do u have statistical data
>based upon the basis of science?
>I am asking this to u, first, because u are one of the few seriuos opinions in
>CCC, and mainly because it looks like (at least to me) that "hash tables sizes"
>is a secret theme or really unknown theme.....
>Thx in advance and my kindest regards.
>Bernardo.

      Hi Bernardo
      Hash size seems to be a difficult theme. The most common rule
      is to give the engines as much hash as possible but avoid at
      any cost swapping the HD. I am not so sure about this rule.
      I have once started a test (engine/engine matches) giving
      the engines different amount of hash A = 16 MB, B = 128 MB
      then A = 32 MB, B = 128 MB and then changed engines. And as
      far as I can remember (I have lost this material !!), the
      engines with 32 MB got slightly better results. As soon as
      I have a modern PC (Dual Core AMD X2 4400+ or higher) with
      2 GB of ram, I would like to repeat such a time consuming
      test to find out if things have changed in the meantime.
      Best regards
      Kurt



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.