Author: Robert Hyatt
Date: 12:54:37 04/01/04
Go up one level in this thread
On April 01, 2004 at 14:16:22, Artem Pyatakov wrote: >On April 01, 2004 at 13:11:14, Pallav Nawani wrote: > >>Would care to you elaborate a little? What do you mean by the most 'pouplar' >>move or how is your method different than just having 4 killer moves? >> >>Pallav > >Pallav, > >Good to know somebody in this forum still cares about computer chess programming >and little advances in the search :-) > >So to elaborate: For each of the killer moves on the ply, I store not only the >killer move itself but how often that killer move has been encountered at this >ply. If a killer move I have not seen yet comes up, I add it to the array of >killers. If this array is full, I look for the killer move that has been the >least frequent in the past and overwrite it with my current killer move. This is >actually very similar to a merge between the history heuristic and the killer >move heuristic. Make sense? > That is how everyone does killers I suspect. It is exactly how Slate described it in Chess Skill in Man and Machine... If you only use 2, the counters can probably be eliminated (I did this a long while back). >By the way, a little update is that with 10 moves it actually does do better >(the result indicating otherwise was an aberration) - 10% less nodes than the >initial 2 killer move heuristics. > >Artem What is the overhead for keeping the extra killers, updating the counts, finding the best killer, etc? IE it isn't free. Hopefully it doesn't cost 10% which would make this "break-even"...
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.