Author: Gerd Isenberg
Date: 11:59:29 09/12/03
Go up one level in this thread
On September 12, 2003 at 14:39:46, Dieter Buerssner wrote: >Hi Gerd, > >my answer is not exactly referring to your post, but an idea I always had. >Instead of Zorbrist type hashing, one could use some "classical" hashing >function: See all the position data (or pawn pos data) as a string, and >calculate a hash - perhaps a 32 bit hash. Many schemes don't need any modulo. >One could use a (32-bit) modulo in the last step. Perhaps one could try the >typical CRC scheme (source available for example in zip). Crypto experts >certainly could suggest even better schemes. Hi Dieter, yes, i believe it too, that there are faster approaches with a 128-bit stream. Xoring consecutive number of bits is a bad idea with pawn bitboards - i guess. Maybe a magic de Bruijn multiplication. Do you have any suggestions or links? >Incremental updates would perhaps >be difficult or impossible, but might be possible in some schemes. For pawn >hashing, storing the whole pawn structure as 2 bitboards, plus perhaps 12 bytes >for the KK pos for verification of clashes might be no problem (I guess, most >that use pawn hashing already store lot's of things about the pawnstructure >anyway, so the size overhead should not be too bad, and would have the >advantage, that one really has certainty). > Yes, that's the way i do it too. Regards, Gerd >Regards, >Dieter
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.