Author: Mike Siler
Date: 14:43:58 01/24/04
Go up one level in this thread
On January 24, 2004 at 16:45:36, Volker Böhm wrote: >On January 24, 2004 at 16:20:14, Mike Siler wrote: > >>On January 24, 2004 at 16:07:04, Volker Böhm wrote: >> >>>On January 24, 2004 at 15:57:50, Mike Siler wrote: >>> >>>>In an average middlegame position, around 80-85% of the nodes my program >>>>searches are quiesce nodes. I have a static exchange evaluator and I only search >>>>captures with SEE value > 0. It seems like other engines are always under 25% >>>>qnodes. What else should I be doing to reduce these numbers? >>>> >>>>Michael >>> >>>What do you count as quiesce node? Do you count a totally quiet horizont >>>position a quiesce node? If yes, 80-85% is ok. Else you´ve got a bug. >> >>Every time quiesce() is called, I increment qnodes. >Suppose you search 1 ply deep with only two kings on the board (no hit moves >possible). Whits turn, white got 9 moves to go. Do you call quiescence 9 times >for eval and count 9 nodes? >Then all is ok as you count the horizont positions as quiesce. That's a good point. I didn't realize until now that it doesn't really make sense to count horizon positions. Thanks. Michael > > >> >>> >>>To reduce nodes in quiesce: >>> >>>1. if static-eval > beta -> finished >> >>That much, I've already got. >> >>>2. if static-eval + piece-to-hit + bonus < alpha -> prune. >> >>I just added this, but it didn't seem to make much difference. I suppose I >>probably have some sort of bug. >> >>> >>>that should be enough to reduce nodes to 25% if not searching "bad SEE hits" >>> >>>P.S. search if SEE == 0 too >>> >>>Greetings
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.