Author: Gerd Isenberg
Date: 12:15:14 01/23/02
Go up one level in this thread
On January 23, 2002 at 09:34:46, Georg v. Zimmermann wrote: >Very interesting post ! Thx for sharing. The 2 positions where it gets wrong >results both seem to be a problem of "useless" interposing. Maybe you have to >add checks there. What are the bitboards you keep updated via makeMove ? > >Georg I hope all detection problems are solved now. In some cases i had to shrink some generalisations. In makeMove i only update the four rotated BitBoards and Piece-BitBoards indexed by (piece & 1), the piececolor and (piece >> 1) the pure piece code. This BitBoards are member of a subclass of CNode. This subclass implements increment update. All the other BitBoards used by eval are embedded in own classes (with inlined getter, initializer and constructor) and got inititialized direct before or during piece and king evaluation. I hash the positional eval with some flags in a own hashtable and also in the main transposition table and i do some kind of lazy eval depending on depth && if outside some bounds && some flags are not triggered by search or eval of (grand)parent node. Gerd
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.