Author: Volker Böhm
Date: 10:41:07 05/29/04
Go up one level in this thread
On May 29, 2004 at 13:31:16, James Swafford wrote: >On May 29, 2004 at 12:09:56, Volker Böhm wrote: > >>On May 29, 2004 at 11:10:47, James Swafford wrote: >> >>>In a recent post, Tord suggested setting a flag in >>>the search when the hash table suggests a fail high, and >>>testing whether the search would indeed fail high. >>> >>>The idea seems so simple I'm embarassed I haven't thought >>>of it before. :) >>> >>>I've been 'pretty sure' for a long time that I've got some >>>nasty hash bugs. I'm in the mood to exterminate them. >>> >>>Last night I implemented Tord's idea and, to my dismay >>>(but not to my surprise) my hash table is saying 'fail >>>high' when the search wouldn't have failed high. And- >>>it doesn't take very long. :) >>> >>>This seems like a nasty thing to debug. I'm comtemplating >>>how I might go about it. I'm hoping some of you can >>>provide some suggestions... >>> >>>-- >>>James >> >>The hash table could say fail high and the search not fail high because of >>search instability. Search instability arises from hash entries, nullmoves, >>foreward-prunings and extensions. >>Thus you have to turn those off to be sure its a bug. >> > >Without any sort of pruning,nullmoves, or extensions, it >still happens (though it does take much longer). Would you >not consider this a bug? > >-- >James > > >>Greetings Volker Are you sure that you never use a hash value? Any use of hash-values anywhere could add search instability. Do you only use "fail high flags" if they are calculated with exactly the same depth? (a hash entry with depth d+n could cause a fail high, your search with "d" don´t find). If both are answered with "yes", then I think you have a bug. Then you have to be able to analyse your search tree. Take the time to add the code! In my opinon the only way to analyze complicated search bugs. Compare the current search tree with the search tree of the hash entry. Greetings Volker
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.