Author: Frank Schneider
Date: 21:07:23 10/19/99
Go up one level in this thread
On October 19, 1999 at 18:19:41, Daniel Clausen wrote: >Hi > >On October 19, 1999 at 11:23:30, Robert Hyatt wrote: > >>On October 19, 1999 at 07:45:33, Antonio Dieguez wrote: >> >>>I have read somewhere "atacking bitboards" >>>What are exactly them? it has for example the data about all squares atacked by >>>white and black? but how are they updated? it looks very complicated. >>> >>> >>>me. >> >> >>It is somewhat complicated, but doesn't have to be computationally expensive. >>IE for non-sliding pieces, you can simply look up in a table to find which >>squares a piece on "SQ" attacks. For sliding pieces it is more complicated, >>but rotated bitmaps turns this into a series of table lookups as well.. > >I'm using bitboards in my chess engine and I'm concentrating on the eval >for once. [:)] I noticed that as long as I can't answer questions like >"is square XY attacked by the opponent" and things like dat, then my eval >tends to be too static, ie it plays the same moves in all games. (more or >less :) > >Ideally I would have a bitboard attacks[64], which shows has all the pieces >attacking a certain square. If I don't have that, I basically have to >make a testCheckForSquare(XY). Gromit uses your kind of attacktables and updates them every move. > >So I'd like to implement these attacks[64] bitboard, but they prolly should >be updates incrementally after each move. :) Do you think this is possible >in a reasonable amount of time? I thought about that for some time, but every >time I do this my head hurts after 10 mins because it seems rather complex. It is maybe a little more complex than other approaches. I Gromit I use two basic methods a) removePiece(sq) and b) placePiece(sq) which are called to update the attacktables a) before the move is made on the board and b) after the move is made on the board. Most of the complexity comes from EnPassant, Castles and Promotions which are handled by separate routines. >So, the question to all bitboard hackers is: do you think this is a good >idea? :) I like them, but it is expensive to compute them. Maybe a good basis for a 'knowledgebased' program but not for a fast searcher. Frank > >Kind regards, > -sargon
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.