Author: Christophe Theron
Date: 20:05:38 11/09/01
Go up one level in this thread
On November 09, 2001 at 20:31:41, John Dahlem wrote:
>How did you determine CT for Palm has a 2100 FIDE rating in the first place? Was
>it simply an estimate based on your play against the program or is there a
>statistal basis?
I have used several different piece of data including:
* matches against other dedicated chess units played during the beta test stage
of Chess Tiger for Palm
* matches against ChessGenius for Palm, which has a known program tested by the
SSDF (Mephisto Roma, with some adjustement due to the speed difference)
* The SSDF rating of Chess Tiger 11.8 on P90 (naturally with speed adjustement
to match the speed of the Palm)
* The SSDF rating of the latest version of Chess Tiger tested on fast hardware
(on K6-2 450, once again taking into account the speed difference with a Palm)
* Results on test suites known to be quite accurate on that class of computers
(the LCT-II for example)
This data is not bullet-proof, but all tests tend to show that Chess Tiger for
Palm is close to 2100 FIDE (SSDF) elo on a standard Palm, and approximately 100
elo points stronger on an overclocked Palm (that is, a Palm running 2.5 times
faster as a standard 20MHz one).
I wish the SSDF will test Chess Tiger for Palm, so we have a definite and
accurate rating for it.
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.