Author: Brian McKinley
Date: 18:26:16 02/02/98
I am creating hash keys using what I assume is a common technique suggested to me by Robert Hyatt. I generated 13 * 64 random numbers (64 bit in my case) and xor values together for each square depending on the piece and color or lack of a piece on each square. I then xor an additional value based on flags for empassant and castling status. The hash key is determined by using the first (n) bits from the resulting value depending on the size of the hash table. This is working much better (translate faster) than what I was doing before, but the resulting distribution in the hash table is not very even. Has anyone investigated using numbers that are not randomly generated to maximize the distance between inputs. My math skills are not up to the challenge. --Brian
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.