Author: Pat King
Date: 16:03:29 09/22/98
Go up one level in this thread
Bob, I want to be sure of what you mean by "collision" in this context. The probability I gave was of erroneously equating a stored position with a different current position, and thus returning a wrong score. I believe that my calculation of that probability stands. As to the equivalent depth, that of course depends on your effective branching factor. Some have claimed as low as 2.5 with null move, so I used 3. I think of collision as when you are trying to store a new position, and find the slot at that hash index filled. Certainly, the number of collisions will go down as the hash table size goes up, but that should only slow down the search, not generate errors. Pat
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.