Author: Thorsten Czub
Date: 03:55:22 08/19/04
Go up one level in this thread
On August 19, 2004 at 05:19:03, Ed Schröder wrote: >To state it's a bug is premature. when i have games where a program plays strange weak moves that are not on the same level like the rest of the moves in that game, it looks like a bug that causes the bad moves. if the hash table clear stops the program from playing those moves, it is an indicator that the weird moves come from older hash table information. >Clearing the hash table means a different move ordering, a different move >ordering sometimes will produce different moves, nothing weird. yes. the change in the hash table is not the problem, the weird moves in the games before without clearing the tables are the problem. >That's okay. In eng-eng matches I assume the loss of speed is neglectful, with >ponder on the loss of speed is estimated about 10% the last time I checked >(Rebel6). yes. but eng-eng have no ponder. so it will not damage. and maybe the weird moves disappear. if you can reproduce the weird moves, you should be able to find out what component produces them. >I will do some testing myself. > >Ed yes. maybe something weakens pro deo.
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.