Author: Mogens Larsen
Date: 11:36:04 10/21/03
Go up one level in this thread
On October 21, 2003 at 13:03:24, Mike S. wrote: >Testing the CM9000 default personality is useless by now. I see. >Did you never open a CM9000 personalities thread, in CCC? Occasionally, both on purpose and by accident. And I do own a copy of CM8000. >I didn't mean ad hoc changes, but well known tweaks which have been proven to be >stronger in thousands! of games. This is known for a long time now, that the CM >defaults are not the strongest settings for The King, on up to date hardware >against other chess programs. This is not a "may or maybe not" case: These >settings are definetely much stronger. Ask the experts, i.e. Kurt Utzinger: > >http://www.utzingerkurt.com/cm9_skr_report.htm > >On other ratings lists, King 3.23 runs at least with selectivity changed to 12, >and of course larger hash (default would be only 4 MB). My recommended settings >(with ks 150/150 in addition) are an "easy derivative" of successfull settings, >which usually have increased king safety settings in common. > >There is no audience for SSDF tests of the CM defaults IMO, because every >serious King "nerd" just doesn't use them. - That's so obvious, also known from >a lot of CCC threads, tournament results etc. that I suspect you don't really >know what this is all about. Of course your'e not obliged :-)) to know every >detail of every progam, I do neither, but the quality of arguments is related to >the information they are based on. The problem with arguing for SSDF testing a personality setting is irrelevance. AFAIK the SSDF test the engine in an out-of-the-box configuration. The only deviances being hashsize, GUI and other minor details. In short, it doesn't matter what you, I or Kurt Utzinger thinks about the default settings or which personality setting we might think is strongest. If I'm not mistaken default settings were used when testing CM6000 and CM8000. Similiarly, one could argue that other engines with adjustable settings should be tested when someone finds settings that may or may not be stronger. Maybe Nimzo8_codpiece is forthcoming or a few extra Shredder incarnations. They are sorely missed ;-). The question should be default or nothing. It doesn't matter a great deal to me either way. Good luck and keep fightin'! ;-) >When you say you don't see the attraction of The King being tested, than I think >you said this from the theoretical viewpoint of the publisher (?). For the >computerchess fans, I don't see why it should be less attractive than to test >Hiarcs or why testing Rebel should be more attractive than King, etc. These are >all top engines, and King 3.23 is one of them, but missing in the list still. In principle as many engines as possible should be tested. However, one could question the relevance. I doubt that being in the top ten or not would make a difference for those selling and using Chessmaster. Mainly because strength might not be the primary parameter. If strength is a parameter the difference between the top engines is neglible for practical purposes IMO and collecters don't care. Some would probably disagree and they're more than welcome. Regards, Mogens
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.