Author: Robert Hyatt
Date: 06:21:25 01/05/00
Go up one level in this thread
On January 05, 2000 at 01:24:52, Landon Rabern wrote: >On January 05, 2000 at 01:03:47, Robert Hyatt wrote: > >>On January 05, 2000 at 00:49:50, Landon Rabern wrote: >> >>>After move generation, when putting a killer/pv/hash table move on top of the >>>list I search through the list of moves until I find the move I am looking for, >>>shift the whole list down and put the move I wanted on top. Is there a better >>>way to do this? The way I am doing it seems inefficient. >> >>There are multiple answers. >> >>1. Don't generate moves before you try the hash move. If it produces a >>cutoff you avoid generating anything. Otherwise, try it, and if it doesn't >>cutoff, generate only captures (if you can do this easily). Try them next. >>Before generating other moves, try the killer moves since they can be tried >>without generating moves and they too may produce a cutoff that will avoid >>move generation. Otherwise, generate all moves and then for the first pass >>to select a history move, delete the killer moves and the hash move. >> >>2. Don't shift things around. Just make the move you want and zero it in >>the list. no move should be represented as 0, so you can tell that an >>entry is a move that has already been tried and cleared... no point in >>shuffling stuff around as it uses a lot of memory bandwith, something the >>PC has very little of. > > >What if a killer move is not a legal move at some nodes? > >Landon I store FROM, TO, MOVING_PIECE, CAPTURED_PIECE, PROMOTION_PIECE in my 21 bit move format. With that information, I can test for psuedo-legality by making sure that MOVING_PIECE is on FROM, CAPTURED_PIECE is on TO, etc. IE I have enough information to be sure that moves are legal enough to not corrupt the board. Later the in_check legality check will finish the testing..
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.