Author: Bo Persson
Date: 09:46:39 01/09/02
Go up one level in this thread
On January 08, 2002 at 19:23:37, David Rasmussen wrote: >On January 08, 2002 at 18:03:18, Gian-Carlo Pascutto wrote: > >>On January 07, 2002 at 20:46:48, David Rasmussen wrote: >> >>>First of all, it may take more time. It's probably not important though. >>>Secondly, I couldn't think of a simple enough way of doing it. >> >>I didn't do promotions in quiescent search/SEE. I just added it. These >>are my test results: >> >>[...] >> >>Warning: node count is different! >> >>Node count is for same depth only. >> >> Old New Ratio >>----------------------------------------------------- >>Nodes: 1046622533 1057808269 101.07% >>Time: 123966 133292 107.52% >>Depth: 11.71 11.66 -0.05 >>Solved: 63 62 >> >>At best, it's not going to make a lot of difference whether you >>do it or not. >> >>This is tricky for me too. If I keep them, I'm going 7% slower >>overall. On the other hand, I _know_ I will be seeing some stuff a ply >>earlier. >> >>The question is now how often do I think it will happen where it matters. >> >>I have no idea. >> >>-- >>GCP > >Yes, it was stuff like this I was considering. > >Also, even if you don't do promotions i quiescensce, and don't want to assess >promotions directly, you might still want to deal with promotions _during_ the >capture sequence that you are SEE'ing, otherwise your SEE results will sometimes >be incorrect. The question is (as you imply): Does it matter? > >/David Another idea is to *not* go into quiescence search when a promotion is imminent, but instead extend the search another ply. Bo Persson bop2@telia.com
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.