Author: Eugene Nalimov
Date: 09:42:54 09/09/02
Go up one level in this thread
Their tablebases are smaller because they contain less information. AFAIK they store only "win in N or not win" for the side to move. If STM does not win they have to probe all the successors of that position to find the exact score. That can be easily done when the position is OTB, but it will kill the program if it'll try it during the search. Probably that's the reason why they are not probing in the search, only at the root. And of course if you probe only at the root you can achieve better compression by using algorithm with larger blocks. Our compression algorithm produces 10% smaller files if you increase block size from 8k to 64k -- but than it noticeable slows down the probes in the search... Thanks, Eugene On September 09, 2002 at 09:01:53, m.d.hurd wrote: >On September 09, 2002 at 08:44:46, Daniel Clausen wrote: > >>Please excuse the typos.. I should proof-read more carefully.. that or Tim >>implements a 'edit post' feature. :) >> >>Sargon > >If they improve on the Nalimov table bases then it would be good for computer >chess if they were made an open standard so that any one could use them, in the >same way that the Nalimov table bases are used. > >Which version is it by the way ? 2.5 or 3, sargon that is ;-]] > >Regards > >Mike.
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.