Author: Uri Blass
Date: 12:29:19 12/30/01
Go up one level in this thread
On December 30, 2001 at 14:30:03, Sune Fischer wrote: >On December 30, 2001 at 12:01:17, Severi Salminen wrote: > >>>If I do a full QS then >80% of the time is spent in there, that is a complete >>>overkill IMO, since chess is not 80% capture moves. >> >>Try counting horizon nodes as normal nodes and only nodes actually _generated_ >>by qsearch as qnodes. Your figures might look a lot better. >> >>Severi > >Well I must have some sort of problem, I haven't counted the nodes, but the tree >explodes for sure. >Maybe it is just move ordering, I haven't really adressed that issue yet, just >using MVV-LVA, probably not good enough. I believe that MVV-LVA should be enough to avoid the problem that you described. If you get 1 more ply by limiting the qsearch then it seems that you have a serious problem. I did not check middle game positions but at least in the opening it does not happen even with TSCP. It may be productive to limit the qsearch(I am not sure about it) but in normal positions it should not be 1 ply difference unless you have branching factor that is clearly less than 2 and in this case your program can outsearch all the opponents. Uri
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.