Author: Peter Fendrich
Date: 12:59:03 06/22/98
Go up one level in this thread
Don! A couple of years ago I tried a lot of different schemes for Killer tables and no one outperformed the "counter version". At that time I hadn't too much memory and the Killer tables had much more impact on performance than today with the big hash tables. Every new killer move is stored with counter = 0; In each node after the first move is generated (I'm using PVS) I set the counters at plydepth+1 to 1. That gives the old killer moves a small advantages to the new ones but keeps the killers local and reduces the effects you are describing. I used to set the counters at plydepth+2 to 1 but I don't remember why I changed this to plydepth+1. Has anyone tried this? //Peter
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.