Author: Uri Blass
Date: 01:58:20 12/21/02
Go up one level in this thread
On December 21, 2002 at 03:36:59, John Lowe wrote: >On December 21, 2002 at 02:57:48, Uri Blass wrote: > >>On December 20, 2002 at 23:49:19, John Lowe wrote: >> >>>When you program castling, you have to checkout the empty squares for attack - >>>theres no reason why you couldn't do this in your situation looking "through" >>>the king. It seems much more efficient to let the king go there and find it's in >>>check. >> >> >>No it is not >> >>When the king castle I can simply use my attack table to find if the square is >>under attack >> >>When the king is in check I cannot trust them and I need to use some more >>complicated algorithm. >> >>Uri > >Hi Uri > >I use two routines for move-list generation - one simply generates peudo-legal >moves and the other simultaneously collects info on attacks, defences, pins >veiled attacks etc - but both write pseudo-legal moves and live with the >consequences later. > >How expensive (in time) is the generation/updating of an attack table ...... or >is it fundamental to your program anyway? > >John Here are my times for the initial position on p850 perft 6=119060324 time=17.25 seconds Movei need to make more than 5000,000 moves for that target and it always update attack tables after making moves. 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.