Author: Steve Maughan
Date: 00:46:05 02/26/99
Go up one level in this thread
>I only get a 4-5% hash hit rate in Zarkov when searching the opening >position for about a minute. I don't store or probe quisecence nodes, >but I do count them in the #nodes. The percentage I calculate >is #hits/#nodes. I compute 4 hash addresses and replace the "worst" >of the four entries, unless the entry is near the root of the tree and >current node is not. I think this hit rate is ok. It goes way up in >the endgame. > >John WOW! Maybe it is OK after all. The way you calculate hit rate is also the same as me. As I mentioned to James R, I would intuatively expect the hit rate with iterative deepening to be about 100/ply factor ~ 16%. ie the nodes already searched by the previous iteration should be in the hash table. Thanks Steve Maughan
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.