Author: Georg v. Zimmermann
Date: 15:55:50 11/18/03
Hi, whats happening here ? It seems like I am missing something fundamental. I just implemented perft and suprisingly enough got correct numbers. Then I was in for another suprise. I wrote a little .bat file to in the future can test my moveGen - correctness faster. It is very very simple and looks like this: ---------- echo we want to see the following numbers echo starting pos: 5072213, pos1: 4185553, pos2: 94716 sunsetter < perft.txt ---------- perft.txt looks like this : ---------- perft 5 setboard r3k2r/p1ppqpb1/bn2pnp1/3PN3/1p2P3/2N2Q1p/PPPBBPPP/R3K2R w KQkq - perft 4 setboard 8/PPP4k/8/8/8/8/4Kppp/8 w - - perft 4 quit ---------- Now when I run this .bat file it goes at half the speed than when I type the same command from the command line ! And I tested multiple times, this is reproducable ! Please take a look at the log. The first one is just typed in by hand : ----------- Sunsetter C10a (Mon Nov 17 22:26:54 2003) (c) Ben Dean-Kawamura, Georg v. Zimmermann See http://sunsetter.sourceforge.net for more info. Created space for 1048576 transposition table entries. < perft 5 Total Time: 26108 Total Nodes: 5072213 < quit ------------ The second one looks totally the same, other than the time taken, and is the one I got from running the bat file : ------------ Sunsetter C10a (Mon Nov 17 22:26:54 2003) (c) Ben Dean-Kawamura, Georg v. Zimmermann See http://sunsetter.sourceforge.net for more info. Created space for 1048576 transposition table entries. < perft 5 Total Time: 52005 Total Nodes: 5072213 [...] ------------- Obviously I checked task-manager, and in both experiments Sunsetter got > 90%. So what on earth is going on here ? Seems like I can throw away all data I ever got from running .bat files ? That can't be ! Georg
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.