Author: Vincent Diepeveen
Date: 17:14:33 11/19/01
Go up one level in this thread
On November 19, 2001 at 18:36:45, Uri Blass wrote: >On November 19, 2001 at 17:49:48, Vincent Diepeveen wrote: > >>On November 19, 2001 at 14:43:38, Ernst Walet wrote: >> >>>This is a nice king safety position from the Rebel site showing (likely) Fritz7 >>>indeed has more king safety knowledge. >>>Futhermore Fritz7 (noMMX) seems to be the latest development of fritz6, as >>>surgested by many indeed. >> >>well let's delay opinion upon default fritz7 for some time, but fritz7nommx >>looks like a positional dumber version. its minimum improvement is 0.03 >>of a pawn. So pawn=32 or frans searches with [alfa+3,max(alfa+4,beta)] >>Anyway something that makes it positional more silly. > >Ir is not clear that doing it make the program weaker positionally. well if all you think you need is a beancounter, then you're kind of wrong. sure at bullet/blitz level fritz7nommx will be stronger as it gets quicker to the tactical barrier. but i'd say why not compare fritz7 (btw set contempt at 0 first before you let it play, because against computers contempt = 0 works better it seems) with fritz7nommx at 40 in 2 level i'd say. > >0.03 pawn is not much and it is possible that by searching 1 ply deeper you can >find difference of 0.04 pawn for a better positional move when without searching >1 ply deeper you are not going to find the better positional move. > >> >>It doesn't look like a preprocessor to me. That's what Frans says about >>all fritz versions after fritz6a, so that'll be the reason that it had >>to be positional dumber to search that deep. > >It is not clear because searching deeper may make the program positionally >smarter for the reason that I explained in this post. > >Uri
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.