Author: Uri Blass
Date: 04:44:49 12/30/01
Go up one level in this thread
On December 30, 2001 at 06:27:59, Severi Salminen wrote: >>>>>>0 2868842 >>>>>>1 3611982 >>>>>>2 997968 >>>>>>3 1564826 >>>>>>4 1196247 >>>>>>5 1214999 >>>>>>6 1184129 >>>>>>7 1153555 >>>>>>8 1180282 >>>>>>9 1153618 >>>>>>=10 1153584 > >>I increase node counter after making moves(my program generates only legal >>moves). > >Also in qsearch()? Yes every makemove is a node. And how did you actually compute the numbers? They just look >so strange to me: how can deeper search reduce the number of nodes from >3'612'000 to 998'000? And how can a search without qsearch result 2'869'000 >nodes and with qsearch a little above one million. There has to be an error >somewhere. > >Severi If I limit my qsearch to 0 ply without doing something intelligent in the leaves I get the following pv's at the end of iterations 5,6,7 depth 5:1.e3 e5 2.Bc4 Nc6 3.Bxf7+ 0.70 for white depth 6:1.d3 b5 2.Nc3 Bb7 3.Bf4 Bxg2 0.46 for black [D]rn1qkbnr/p1pppppp/8/1p6/5B2/2NP4/PPP1PPbP/R2QKBNR w KQkq - depth 7:1.Nc3 e6 2.e4 Bb4 3.Nb5 Nc6 Nxc7+ 0.90 for white [D]r1bqk1nr/ppNp1ppp/2n1p3/8/1b2P3/8/PPPP1PPP/R1BQKBNR b KQkq - The program changes it's mind often and it is a good reason for searching more nodes If I limit the qsearch to 7 plies I do not get usually strange pv's and evaluations even when I use the stupid eval for static positions. 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.