Author: Ed Schröder
Date: 08:47:42 03/03/00
Go up one level in this thread
On March 03, 2000 at 11:35:22, Brian Richardson wrote: >Looking for suggestions with a hash table "problem". Running > >8/8/7k/8/4p1K1/8/5P2/8 b - - bm e3; id "Fine position 16"; > >Tinker likes e3 with some hash table sizes (usually at around ply 16), but Kg6 >at other sizes. Even stranger is that e3 is picked at some smaller hash sizes, >and with very large sizes, but not for "medium" size hash tables. > >Of course, I may just have a _nasty_ bug in my hash table code. > >Or, it may be something subtle. It _may_ have something to do with move >ordering--if there are two branches that happen to have the same score at the >same depth, in one case one position is hashed and it never "sees" the other >one? > >Have others run into this? > >Any suggestions welcomed. >Thanks. Never trust your hash table code, that is right. On the other hand don't bother too much about it as the behaviour you describe is quite natural. You are dealing with pruning based on scores, extensions which can cause different depths. My rule is to count the plies a move must been seen by search and then add 1 ply as a limit search should see the move to be found. Good luck with the problem. Ed
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.