Author: Daniel Clausen
Date: 12:57:38 05/11/02
Go up one level in this thread
On May 11, 2002 at 15:04:14, pavel wrote: >On May 11, 2002 at 09:46:49, Vincent Diepeveen wrote: > >>On May 11, 2002 at 05:48:17, Matthias Gemuh wrote: >> >>for attacktables, bitboards isn't the fastest thing. >> >>Bitboards is a tradeoff. You put less information into >>a single 64 bits word. So you can use instructions like AND and OR >>more easily. >> >>However to get attacktables you might want to look at gnuchess 4.0 (not >>5.0 which was raped and rewritten to bitboards for some unknown reason). >> >>Gnuchess is putting more information into a single word for each >>square. The advantage is you can faster work with complex knowledge. >> >>The disadvantage is obviously a slowdown in speed because you use >>more knowledge. > >So in short it means, bitboard helps put more information (==more knowledge)in a >single square (64 bits?), if that is what I understand. > [snip] Not quite. In a bitboard approach, each bit represents one square of the chessboard, and you can store exactly 1 bit of information about each square. (ie square is occupued by a white piece or not, square is attacked by an enemy knight or not, you get the idea) Storing more information requires more bitboards. (1 bitboard for each bit of information :) Let's compare this to an alternative board representation, say 'int square[64]'. In this approach we have an integer (is >= 32 bit, if your compiler is ANSI-C) and therefore can store much more information about a particular square. This doesn't mean that bitboard-programs necessarily are dumber though. (ok, mine is :p) Just use more bitboards if you want to use more information :) For example Crafty is a bitboard-engine and is quite intelligent, IMHO. (YMMV) HTH 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.