Author: Georg v. Zimmermann
Date: 06:25:11 05/13/02
Go up one level in this thread
Hi, hash tables are essential and should get you around 3x speed improvement on average. The whole idea of iterative deepening is (other than easier time management) to fill the hash tables, so clearing them between iterations is a bit silly :-). Have you compared your implementation with the one in resp or Beowulf for example ? Maybe you want to post the relevant source ? Noticable search instabilities most likely mean a serious bug (storing wrong draft, overwriting with too big numbers, wrong adjustment of alpha/beta etc etc) in your implementation. Kind regards, Georg On May 13, 2002 at 08:48:36, Vladimir Medvedev wrote: >I'm going to implement hashtables in GreKo (they will be in ver. 0.42, now I'm >testing their work). > >The question is, how to manage hashtables while iterative deepening is used? >Now I clear hashtable on each iteration, because holding them unchanged since >last iteration causes search instability. But I feel that I lose some useful >information this way... > >Please, could anybody give me some information, weblinks or advice on this >technique? >In addition to http://www.seanet.com/~brucemo/topics/hashing.htm, which is the >main knowledge source for me now?
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.