Author: Heiner Marxen
Date: 05:02:41 01/19/01
Go up one level in this thread
On January 19, 2001 at 07:54:21, leonid wrote: >On January 19, 2001 at 06:34:32, Heiner Marxen wrote: > >>>Only one thing I see as practical why saving the moves that lead >>>to given position, is the taken of pawn "en passant". There you need "previous" >>>move to know if it pawn should be taken. >> >>Ah, I see. >>But I would not store a cemplete last move, but rather, if and into which >>column an e.p. may happen just now. That takes 4 bit (9 cases). >>I have also 4 bits for the castling rights, and one bit for the side to move. >>Total: 64*4 bit square contents, 4+4+1 Bit extra info. > >Interesting to see that you do this some other possible way. Actually rocks with >all the details about this (where rock already happened, what king or rook ever >changed its initial place and some other data) is kept in one variable. 2 bytes. >Previous move program keep in 4 bytes long variable as every move. This I do >probably because I could not see other way but mainly for simplicity. Even >description of the board I keep in exact 64 bytes chain. It could be reduced to >much less volume but then the simplicity will vanish. > >Leonid. Ah, yes, sure. I was not very clear. I use this compact coding (4 bit for each square etc) only for the TT. Otherwise Chest uses quite a different format. Yes, simplicity is a good thing. Heiner
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.