Author: Stuart Cracraft
Date: 13:33:06 07/26/04
Go up one level in this thread
On July 20, 2004 at 01:57:43, Richard Pijl wrote: >On July 19, 2004 at 21:06:47, Stuart Cracraft wrote: > >>On July 19, 2004 at 18:21:38, Richard Pijl wrote: >> >>>On July 19, 2004 at 18:07:53, Stuart Cracraft wrote: >>> >>>>So I recently added some code to search the PV first >>>>but with an add-on whose effect may be a curveball >>>>in the longrun but appears okay at first. >>>> >>>>PV is searched first but with the add-on that any other >>>>position that is searched (but not on the PV) there is a check >>>>made for whether a PV-move at the same depth is available >>>>in this position and if so it is searched first. >>>> >>> >>>Do you use killer heuristics in move ordering? >>> >>>Richard. >> >>Only in the sense of history heuristic. No explicit killers. > >I think what you're proposing here is similar to killers (although it is not the >same). It would be nice to see how this method compares to using killers, and if >it also gives an improvement when you are using killers. > >> >>Do you think explicit killers will help the history heuristic that much? > >I believe the killer heuristic in move ordering really helps. When I added >killermoves to the Baron it made a nice improvement. > >Richard. I tried adding killers but there was no speedup. I have PVS, search the PV move first otherwise the standard hash move if available, and then normal ordering (captures mvv/lva with like captures broken by centrality), then normal moves broken by centrality. History heuristic value is added to all moves and is the primary sort. A paper of Schaffer's says that history heuristic and transposition table alone are sufficient for the bulk of all move ordering improvements and I believe him. Stuart
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.