Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: SSDF question

Author: Paul Petersson

Date: 10:27:57 02/23/98

Go up one level in this thread


On February 23, 1998 at 07:13:22, Andreas Mader wrote:

>Is Fritz the only program that played its games mainly with 64MB hash
>tables? And if yes, don't you think that this is an unfair advantage
>given to one single program? What will you do if the next version of
>Fritz' autoplayer will only work if Fritz is given a Pentium II 400?

I can only speak for myself and my own testing. Fritz is not the only
program that have had 64 MB hash when tested. I have played Nimzo98 and
Rebel9 with 64 MB and Nimzo98 and Shredder2 with 48 MB. Only a few of my
games have been played with 32 MB, mostly with Shredder2, Rebel9 and
Genius5.

The main problem with testing Fritz5 is the autoplayer. It doesn´t
function very well. Not only does Fritz crash and hang occasionally, it
sometimes crash the opponent as well. This happend to me when Fritz
played against Nimzo98. Only a Ctrl-Alt-Del, and killing Nimzo would get
out of the hang, same with Fritz. Fritz autoplayer doesn´t work in
exactly the same manner as Donningers does. Instead of playing a preset
number of games with either white or black, Fritz always starts with
white, and when the game is over it switches colours so the opponent
plays white the next game. This goes on until the tester interrupts the
testing.

If ChessBase demanded that a new version of Fritz (or any other engine)
would need more than 64 MB, or a P II 400, they would bee out of luck.
It simply wouldn´t be tested. At least not for the next 18 months or so.
Many of SSDF´s testers will soon have 64 MB in their machines, so the
problem with different testers having different amounts of memory will
slowly fade away.

Paul

>
>Andreas



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.