Author: Richard A. Fowell
Date: 19:56:13 06/06/02
Go up one level in this thread
On June 06, 2002 at 18:22:46, Kevin Strickland wrote: <snip> >Does a commercial program contain more knowledge in the endgame due to lack of >way to improve the opening/middlegame so the authors concentrate on the endgame >due to the amount of knowledge there is to put in there? <snip> One factor is that the older commercial programs originated when memory was expensive enough that it was not too practical to support databases. When I started testing chess programs, none of the commercial programs I had access to (including HIARCS and Chessmaster) supported tablebases. In those days, Computer Chess Reports published a suite of endgame test positions that I used. Based on those (ancient) results, I would expect you to find that HIARCS will perform excellently in endgames without tablabases. If one is developing a chess program today, the incremental benefit of adding endgame knowledge on top of tablebases is less, so there is somewhat less incentive to do the (difficult) work of adding such terms to the eval. Arguably, there would be some benefit (if you can count on tablebases) to strip endgame knowledge out of the older engines. -Richard
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.