Author: Sune Fischer
Date: 15:53:28 01/23/03
Go up one level in this thread
On January 23, 2003 at 18:23:11, Bruce Moreland wrote: >On January 23, 2003 at 09:56:18, Jon Dart wrote: > >>I use a history table, but I only take the top 2 highest-scores moves out of it, >>so in effect it becomes a big version of a killer table. The other moves are >>tried in random order. (Sorting all moves by history score was too expensive an >>operation for too little benefit, last time I tried it). > >In a perfectly ordered alpha-beta tree, you will always have to either search >all of the successors, or one of them. > >So if your move ordering is good, if you have to search more than a couple of >moves, chances are that you will have to search them all. > >This being the case, it doesn't matter what order you search these moves in. >The only time it matters is if you make a severe mistake with your move >ordering. Is *anything* going to help you? You've done the *best* you can for >the first few moves, and it either didn't work, or is never going to work. Why >do *more* of this? > >bruce Lat time I checked my move ordering was ~30% on the history moves. That means every 3rd time a history move produces a cutoff the history sorting helped in searching it first. It works for sure, question is if it slows you down too much, only a test can tell you that. -S.
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.