Author: Uri Blass
Date: 05:50:35 06/04/02
Go up one level in this thread
On June 04, 2002 at 07:11:20, Vincent Diepeveen wrote: >On June 04, 2002 at 02:11:40, Russell Reagan wrote: > >To speak for diep in i-csvn2, on average i was outsearched >by 3 ply by the commercial programs i played there. > >Even The King (Chessmaster) nowaday searches deeper than it used to do. >It got like 12 ply now where it in past times would have gotten >9 ply at same hardware. > >One game it lost and not because it was outsearched but by its own stupidiness. > >I wasn't feeling bad for getting outsearched. Instead i saw so many times >my score go up at for example 11 ply with a certain mainline. Then after >making my move opponent needed a huge number of plies more to get a fail >low and get to the same move. > >Nullmove is a great enhancement of nowadays programs, but i get impression >most just prune too much. Also all preprocessors i play against get >completely annihilated. It is clear that the glory times of preprocessing >and getting a few ply more instead, that those days are over. > >Nevertheless i get impression that most here do not know what is happening >last year. Last years we have seen no new ideas which were giving a >search speedup like nullmove, hashtables and alfabeta have given. > >Instead everyone worked at his evaluation bigtime. I have ideas how to improve the search speed but I decided in the near future to think about the evaluation. It does not mean writing a lot of source code but thinking how to do the biggest improvement in evaluation by relatively small changes. 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.