Author: Jari Huikari
Date: 04:32:17 12/11/97
Go up one level in this thread
On December 10, 1997 at 16:58:09, Dan Newman wrote: >it was in one of Levy's books) it has been suggested that a move >generator might be constructed that produced moves in some >reasonable order to start off with. Certainly one could produce >capture moves ahead of non-capture moves and moves by less valuable >pieces before more valuable ones--which might do well enough >(especially if you try null moves, killers, transposition table >moves, and so on, ahead of doing move generation). Good move >ordering is of course a *very* big win since it can give orders >of magnitude better performance than random move ordering. I make several searches. First I try all possible moves and responses to them. That's how I find legal moves in current position. Then I search one move deeper and so on until time limitation. While searching I give bad scores to those moves I'm not interested to search in further plies. The index in my move table tells which move is which because they are always searched in the same order. Very primitive method, but Nero is even more primitive. :-) Jari
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.