Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: ELO rating as a function of hash table size

Author: Enrique Irazoqui

Date: 06:56:07 09/27/99

Go up one level in this thread


On September 27, 1999 at 09:27:40, Michael de la Maza wrote:

>I'm interested in gathering estimates of how the ELO rating of a program changes
>as a function of hash table size.  For example, if a program has an ELO rating
>of 2300 with a hash table of 64M, what would its ELO rating be with a hash table
>of 0M?
>
>I know that this relationship is non-linear, difficult to compute, etc.  I'm
>interested in ballpark estimates.
>
>Michael

It depends on the program. Based on the SSDF scale and on a P200MMX, I have the
following:

P200MMX
CM6K 128MB         2591
CM6K 16MB          2591
F532 128MB         2590
F532 46MB          2579
N99a 128MB         2574
N99a 46MB          2562
J5 128MB           2538
J5 46MB            2538

Enrique




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.