Computer Chess Club Archives


Search

Terms

Messages

Subject: round 40: rybka betas versus fruit 2.2.1; thoughts on time mangement

Author: Joseph Ciarrochi

Date: 17:42:45 01/18/06


Pentium 446 mhz . 40/4 repeated. Noomen test set.
Fruit 2.2.1 in gauntlet against rybka


Rybka beta 10d optimistic   23.5- 40
Rybka beta 10d default      23.0 -40
Rybka beta 10d ultrasolid   22.0- 40
rybka v1 beta 1  default    20.5 -40



Too early to make conclusions, but thought you might be interested.

I've decided that CEGT got it right to have this kind of increment (x moves in a
certain time). I suspect it reduces error due to differences in time management.
(e.g., rybka 10d search might be improved in the middle game, but if it uses all
its time in the endgame, we won't see the improvement much).

I have heard the argument that time management is a valid part of performance,
and i can not argue with this. However, for those of us who use engines for
analysis, time management issues are not so important.

best
Joseph






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.