Author: Sune Fischer
Date: 07:17:01 12/20/02
Go up one level in this thread
On December 19, 2002 at 23:18:55, Miguel A. Ballicora wrote: >When I started, my first idea was to make a pgn cruncher. So, I figured that >a legal move generator could be more useful (as Peter Fendrich points out >in another message). As an engine, first unreleased versions of Gaviota had a >legal generator. It was cool, detected all sort of pins statically using >bitboards, but later I realized that the program was faster when I turned into a >pseudo-legal generator. >A legal move generator gives the advantage of knowing _exactly_ how many moves >you can make, if you use that information (Uri does, I don't) it could be >useful, otherwise, I think a pseudo legal should be simpler, tighter, faster and >less prone to bugs. IMHO. I hold the same opinion. >I use a legal move generator only when I am in check. I think it is useful there >and the slow down is negligible. Actually it shouldn't slow you down, if it does I bet you are one of those that adds illegal nodes to their node count :) I think that is a bad idea, because illegal moves is a search overhead and higher nodes per sec is a good thing, counting illegal nodes is simply wrong as it gives you the impression a large search overhead is a good thing. Doing check evasions is relative slow, but it is should still be faster than making and unmaking illegal moves. I do see a small speedup in kNps when doing check evasions. -S. >Regards, >Miguel > > >> >>pseudo legal move generator is the easy way but I do not think that it is better >>and the fact that chest is the best mate solver suggest that legal move >>generator may be better. >> >>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.