Author: Ian Osgood
Date: 11:47:45 10/03/01
Go up one level in this thread
On October 03, 2001 at 07:46:00, Dan Andersson wrote: >Good work. Did you implement hash tables also? For conventional chips, I mean. >Post the code on comp.lang.forth to get suggestions of improvements, they are >realy nice :) > >MvH Dan Andersson Thanks, it has been a fun excercise. It only has TSCP's features: simple but worthy eval, history table, MVV-LVA move sorting, quiescence, root a=-inf, b=inf searches, check extension. No hash, no killer moves, no null-move. Jeff Fox, who ported an earlier chess program onto a Forth chip, already thinks it is way too big with 64K of data structures. :) I did change from mailbox to 0x88 edge detection, since in Forth 088 AND is easier than a double table indirection. I also have been optimizing some inner routines (inCheck, incremental update of material & pawn structure for eval, attack detection), and added a narrower root a-b window with fail-high/low handling. It's hard to call a chess program "done", isn't it? :) I'd like to contact Tom Kerrigan to let him know of my work, but his web page seems to be down. Does anyone have his current email address? Ian
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.