Author: Will Singleton
Date: 13:15:19 01/23/03
Go up one level in this thread
On January 23, 2003 at 14:30:30, Peter McKenzie wrote: >On January 23, 2003 at 13:29:17, Will Singleton wrote: > >>On January 23, 2003 at 12:27:28, Peter McKenzie wrote: >> >>>On January 23, 2003 at 05:36:12, Uri Blass wrote: >>> >>>>On January 23, 2003 at 03:57:45, Peter McKenzie wrote: >>>> >>>>>Sounds like you are doing history heuristic. A couple of points (which you >>>>>might know already): >>>>> >>>>>1) Make sure you have a separate history table for each colour. >>>> >>>>I have only one history table. >>>>Is it important to have seperate history table for each color? >>> >>>Only if you want it to work well :-) >>> >> >>I just tried two tables, and i get better performance on some positions, worse >>on others (in the 5% range). A single table in which you OR in the piece value >>to the index should be OK. > >You might have to explain that last comment to me. > >Its been a long time since I implemented HH, but I seem to remember not being >able to get a very good improvement until I started using 2 tables. All >programs are different of course so who knows how the 1 vs 2 history table thing >interacts with other design choices. > >Still, there are a reasonable number of subtleties and choices with history >heuristic. Just off the top of my head: > >- how do you increase the history count? (by 1, by depth, by depth**2, etc) >- how/when do you decrease your history counts to stop overflow? >- how many history moves do you use for move ordering? (All?, 2?, 3?, 4?) >- is history used for ordering q-search moves? >- only increase history count for non capturing cutoffs (I think everyone does >it this way?) > >Peter > >> >>Will I think perhaps you're mixing up killers and history tables. Will
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.