Author: Uri Blass
Date: 08:38:09 03/21/04
Go up one level in this thread
On March 21, 2004 at 11:04:11, Dan Andersson wrote: > I'm talking about ETC not the Fruit implementation of it. > About the refactoring comment. Your first post made me think you either update >the attacktables for all moves before making the actual ETC probes. Or update >the attacktables for each single move before the move is probed. > >MvH Dan Andersson I will try to make things clear: 1)I do not use ETC at this moment and I never tried it. 2)My makemove update the attack tables and the same for my undomove so it clearly does not make sense to make all the moves and undo them if the target is mainly hash cut off and not mate cut off. 3)I thought that the same is also for other programs because even without attack tables you do not need to update information about the board that is not the hash key in order to use ETC. 4)I even did not look at a code that is using ETC before writing a function that update only the hash key of the move without making the move because I assumed that it is clear that this is what I need to do before implementing ETC. 5)I was surprised to find that the ETC implementation in fruit makes all the moves and it seems that the same for the code in the article that you posted except the fact that it has no undomove and seems to have only 2 functions(firstmove and nextmove when firstmove make the first move and nextmove undo the last move and make the next move in case that there is next move). When I think about it there is some potential for speed improvement here by having less updates of the attack tables by nextmove instead of undomobve but I do not plan to care about it now. 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.