Author: Robert Hyatt
Date: 08:02:04 10/29/03
Go up one level in this thread
On October 29, 2003 at 10:02:24, Dan Andersson wrote: > I know I'm being too paranoid :) For Zobrist hashing a concatenated number will >be reasonably safe unless you use a LCG with defects or a baad seed. > >MvH Dan Andersson Here is a good experiment to run: (I won't reveal the results yet). Take a good chess engine, and modify the HashProbe() code so that every N nodes, you match the table entry whether the hash signature really matches or not. Once you do this false match, you use the entry normally. IE the depth is tested, etc. If you make N == 10, would you expect it to kill the program's search results? You might be surprised if you really run the test. I was. I don't think the randomness of the numbers is important at all based on the results I have been producing testing this particular issue...
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.