Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: ...old version on new system vs. new version on old system

Author: Gregor Overney

Date: 11:01:00 09/07/01

Go up one level in this thread


>I agree that Deeper blue was the best at 1997 but programmers learned a lot
>after 1997.
>
I am not too sure about those algorithmic improvements that have been achieved
from 1997 until 2001. For example, take Junior 5 and the newest version of it.
Give version 5 a PIII/850 and the older one a PII/400. You can repeat this test
with Crafty from '97 and with the newest version 18.10. No great algorithmic
improvements have been made that make a slower system with the newest engine win
against an older version running on a significantly faster machine. Actually,
regarding Crafty, I sometimes get the impression that 18.10 is not the strongest
version of Crafty.

Therefore, deep fritz vs. Kramnik compares with deep blue vs. Kasparov like,
well.... apples with oranges.

6M nodes gives you not much more ply's compared to a fast PC with 0.8M nodes.
The difficulties deep blue had to win against Kasparov partially proof that IBM
was a little bit lucky to win this game in '97. With giving Kasparov more rest
and giving him the chance to play against deep blue before the match would have
made a difference.

When do we see a commercially available program/system that beats the acting
world champ in Chess under official match conditions? I guess in 2010. We are
not there yet with systems like deep fritz, even with 64 CPUs running at one
1GHz each. And with the most recent difficulties in the computer industry,
progress in CPU design might even get slowed down.

Gregor



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.