Author: Komputer Korner
Date: 11:26:44 10/05/98
Go up one level in this thread
On October 01, 1998 at 15:46:44, David Eppstein wrote: >On September 30, 1998 at 15:46:33, Komputer Korner wrote: >> This is probably the way that CSTAL does it, but it is hellishly complicated >> and results of computer vs computer tests have not been encouraging for the >> DOS version of CSTAL. Maybe the windows version will be better. > >I don't think Chris has told us what CSTAL does, but I was under the impression >that its eval is more or less normal (except for having more knowledge than most >so it can come up with a reasonable number even in non-quiescent positions) and >that the mysterious "Tal function" is not in the eval, but is instead something >to do with choosing how to expand the tree -- back when Chris was talking about >it he said things about how it could win even with many fewer nodes than other >programs, because it had a different set of nodes. He has indicated that his program has the most asymmetrical code of any program with the possible exception of Genius. However he does it, CSTAL has an incredible amount of knowledge that will serve it well when computer speeds increase by tenfold in the coming decade. However it looks like unless you have a parallel program you won't have a chance in comp-comp events except the world micro which still bans parallel programs working on a multi cpu machines. -- Komputer Korner
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.