Author: Robert Hyatt
Date: 14:02:25 09/09/99
Go up one level in this thread
On September 09, 1999 at 15:59:00, Steve Maughan wrote: >>>Am I missing something? >> >>Yes.. You are assuming that nearly _every_ position is a hash hit. This is >>wrong. IE only about 10% or so of typical middlegame positions result in >>hash hits. Which means that 90% of the time I am relying on _other_ move >>ordering to pick the best move. Most of the time this is a winning capture, >>but killers and history do help. >> > >I see. Still by the virtue of the fact that you use Internal Iterative >Deepening is it not the case that you virtually force a hash hit - ie when you >do not find a hash move Crafty searches Depth-2 after which, assuming that it >finds a move score of > Alpha, it will store the position in the hash table and >this will be used as the effective hash move. Hence for most positions there >will be a hash hit. Am I still wide of the mark? > >You help is appreciated! > >Steve Maughan If you look at search, you'll notice that IID is used only when alpha and beta are set to their original values. Since I use PVS, you will notice that this is only true right on the PV path, and nowhere else. IE when I start a new iteration, and step down ply by ply, the _first_ time I search a move at each ply after starting a new iteration, if I don't have a hash move I try IID to get one. This hardly ever happens... and when it does, it is only on 10-12 nodes in a normal iteration. IE _not_ at every position where there is no hash move, but only on PV nodes where there is no hash move...
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.