Author: Mikael Bäckman
Date: 03:58:02 10/30/04
Go up one level in this thread
On October 29, 2004 at 09:48:31, Peter Fendrich wrote: >How do you guys deal with this? > >In addition to the ordinary hash table, I have an evaluate hash table with only >absolute values (no upper and lower flags) and a lazy evaluate that just quits >evaluating when the score exceeds alpha-Magin or beta+Margin. >I don't store lazy evaluate values in the evaluate hashtab - they are not >complete. As I see it there are a few options: > >1. Don't use lazy evaluate >2. Include upper, lower and inside flags in the evaluate hashtab as well >3. Don't use a specialised evlaute hashtab. Instead use the ordinary one. > >How do you deal with this? Personally I don't like number 3. > >/Peter I do much the same thing you described. I evaluate only in quiescence search, and only store the full evaluation into the evalutation hash. I use lazy eval and probe the hashtable if lazy eval fails. In middlegames I typically get an 5-10% speed improvement, in endgames the improvement is slighty larger. /Mikael
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.