Author: Christophe Theron
Date: 17:17:03 05/01/00
Go up one level in this thread
On May 01, 2000 at 17:41:21, Bertil Eklund wrote:
>
>Played on 2 AMD450 K6-2, auto232 no problem at all. 5min blitz.
>
>Nimzo7.32-Genius6.5 35-27 (62)
>Hiarcs7.32-Genius6.5 43,5-30,5 (74)
>
>Not so bad!?
>
>I believe that Hiarcs and Nimzo is about 200 elo stronger than Genius6.5 in
>2h/40.
>
>Bertil
I completely agree that this is not too bad, and I also agree that the
difference is bigger at long time controls.
But as you know, Bertil, there was a time when such an experiment would have
been considered as unfair for its opponents, as Genius was standing well above
the pack in blitz.
It would be interesting to replay the matches at 2min/game. I'm sure there is a
point where Genius outperforms its opponents.
In my opinion, and I apologize if this is a bit technical, Genius performs less
well at long time controls because of its Effective Branching Factor.
That means that Genius is less effective than other programs in pruning out
moves in deep trees. The complexity of its tree grows more than the complexity
of other top chess programs trees. The difference gets bigger and bigger with
increasing depths. I think this could be shown by measuring the times needed to
complete searches of increasing depths, and by plotting the times of Genius and
the times of other chess programs.
Anyway I like a lot the Genius programs. They have very stable evaluation
functions and their search is still efficient at fast time controls. I still use
Genius 5 as a sparring partner for Tiger from time to time, and I have bought
the Palm version of Genius.
Great program, a legend I would say. Any computer chess lover should have at
least one version of Genius in his collection.
Christophe
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.