Author: Johan de Koning
Date: 22:48:46 05/15/04
Go up one level in this thread
On May 15, 2004 at 17:37:48, John Merlino wrote: >On May 15, 2004 at 16:48:10, James T. Walker wrote: > >>On May 15, 2004 at 12:22:16, John Merlino wrote: >> >>>On May 15, 2004 at 10:30:51, James T. Walker wrote: >>> >>>>On May 14, 2004 at 21:11:54, Dann Corbit wrote: [SSDF stats] >>>>>Therefore, 200 - 87 = 113 was the last increase for ChessMaster (approximately) >>>>>Hence, 100-150 Elo would be par for the course. >>>> >>>>Thanks for the stats Dann. Do you know if CM9K was tested using original >>>>default settings or what? >>>>Jim >>> >>>The SSDF tested the program with original settings EXCEPT a larger hash table >>>and increased selective search value (to 9, I believe). Other than that, it was >>>the default. >>> >>>jm >> >>I seem to rember they used the "SKR" settings. Also not sure if they ran CM9K >>in the Chessbase GUI with a book not ChessMasters. I would be interested in the >>facts. My memory is very poor these days. >>Jim > >I know that they started to use the SKR settings, and lots of people complained >that The King was getting special treatment, using "optimized" settings rather >than "out of the box" settings. The SSDF then stopped testing the SKR settings >and went back to the "slightly modified default". The engine had been tested with SKR settings, under CB and GENERAL.CTG. Then people complained about using learning and book that are not part of Chessmaster. I'll skip the nasty bits here, let's just say it was the time of year that a lot of people did not make many friends. :-) Then SSDF discarded the games and started testing CM out of the box. At that point I lost contact. I did check several games though, and the timing seemed to match the moves. That is assuming all default (which SS=9 was), except for a larger TT size. ... Johan
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.