Author: David Hanley
Date: 15:51:41 08/20/02
Go up one level in this thread
Well, i understand what you are saying, and i agree with your aversion to randomness in a program. I am surprised by the checkers program missing a win due to a false positive hash. It seems like the state of a checkers game can be represented in far fewer bits and should have less false matches. I tell you what, i have an idea that may tell us something. What if we match two versions of crafty ( or some program ) against each other. One will a totally normal version of crafty. The other copy will have the high 32 bits of all its hash keys set to zero. They could play 10,000 speed games. I wonder if their scores would be significantly different. Maybe 10,000 is too much, it looks like that would take two months! As i say as a computer science person i understand you aversion to randomness. My asking is just because of a specific situation i am presented with, and is not urgent--i'm using 64 bits currently. dave
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.