Author: Adrien Regimbald
Date: 10:53:23 06/23/00
Go up one level in this thread
Hi, >In The Crazy Bishop, I mark all positions from the game history (=past positions >in the game) which are in the hash table as unreliable (ie, do not use the eval, >but the best move is still useful) before starting a new search. Removing them >completely from the hash table would work too. This should solve your problem, >if I understood it well. Hmm.. you loop over all entries in the hash table and reset the depths to 0? (I presume that this is after you have actually made your move, and not after every iterative depth) Couldn't this be a problem in very fast games, in terms of the time wasted doign this? >I also always test for 3-fold repetition before probing the hash table during >the search. I do this too ;) However, the problem is that the draw is achieved by the opponent's reply, thus you would see the draw by 3-fold repetition at ply 3, but you get a hash table hit at ply 2 :P >I think that the combination of these two things makes the search very safe in >respect of hash-table-related 3-fold repetition problems. Do you notice any major performance hit when resetting the depths? >(I got a "Document contains no data" error when trying to reply to the original >message). I actually got that too when I was trying to post at all.. Thanks, Adrien.
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.