Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Is the Steve LopezTablebase formula different for Pentiums than Athlons?

Author: Bob Durrett

Date: 10:20:05 10/30/02

Go up one level in this thread


On October 30, 2002 at 12:11:55, Brian Katz wrote:

>Steve Lopez of Chessbase has given the Formula of
>
>2 X Processor Speed X Avg. Seconds Per Move  Divided by 1000 = MB Hash Tables
>
>My question is, is this formula the same for Athlons as well as the Pentiums?
>
>Or should this formula if accurate, be based on the Athlons Performance rating
>such as in the Athlon 2.0 gig 2400+ ?
>Should it be based on the 2.0 Gig or on the 2400+ rating?
>
>My guess would be to base it on the Performance rating since clock speeds are
>different than the Pentiums.
>The other question is about Tablebase in general.
>Is this formula really accurate or is trial and error the best way to find the
>proper setting?
>
>Perhaps the formula is only relevant to much slower processor speeds? Or not
>necessary for any processor speeds.
>
>Any info on this would be greatly appreciated.
>
>Brian Katz

I forgot to mention:

The best estimator for "optimum hash table size" may depend on choice of engine.
 Certain engines might work best on the Intel processors and others work best on
the AMD processors.  It could even be worse than that:  It might depend on the
specific processsor.  Other significant variables may be processor speed, number
of processors used by the engine, etc.

There may not be any simple answer to your question.

 The formula you gave was for Fritz 5.32 running on then-existing single
processor machines.

Bob D.



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.