Author: Steve Maughan
Date: 01:05:41 01/02/01
Go up one level in this thread
Bob, >Those occur when I get and EXACT hit. I search no deeper, store the table >value as the best score here, and then stuff the PV with the moves from the >root to the previous ply. I then set the <HT> flag since the PV should >have more moves, but I don't store PVs in the hash entry (this is doable >but expensive in terms of memory). Yes of course I do the same. >It has to be a bug. The most common problem is backing up a garbage PV when a >node fails _low_ as there can be no best move, much less a PV. If you fail >high, you also have no PV, only a best move at _this_ ply. But the ply below >this one is a fail low with no best move of any kind... > >Hashing is the special case... if you get a hit, you have to fake up a PV >since there is going to be no further searching to back one up. It is up >to your hash lookup/hit code to recognize this and build a PV from each move >played up to the previous ply... I'll try and hunt the bug down. I'm still not convinced but I have to do some more experimentation. Thanks again, Steve
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.