Author: Andrew Williams
Date: 10:34:18 01/04/02
Go up one level in this thread
On January 04, 2002 at 10:33:27, John Coffey wrote: >On January 04, 2002 at 10:14:17, Severi Salminen wrote: > >>>There has been some confusion in my mind for years >>>about move ordering. I know that good move ordering will reduce the size of >>>the tree. How important is it to get all the moves in the best order as >>>oppose to just starting with the best candidate? >> >>The optimal solution is a movegenerator that allways generated the best move. >>But since that is quite impossible it is best to try to generate only those >>moves that are likely needed and get a cutoff as fast as possible. 1. try the >>hash move and don't generate others. 2. generate captures and try the good >>ones. 3. try killers. 4. generate the rest and order them by history values. 5. >>try the bad captures. In each step try to do as little as possible, generate >>only the moves you need and do it fast. So if you get a cutoff with a hash move, >>you really don't need to know _anything_ about the rest of the possible moves. >> >>Severi > >How many programs actually do this and how many generate the entire move list? > >John Coffey I generate all mine at once. I've tried in the past generating them in two stages (captures+promotions then non-captures), but I never got any huge improvement and dropped it in favour of simplicity. Andrew
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.