Author: Uri Blass
Date: 08:45:23 03/09/03
Go up one level in this thread
On March 09, 2003 at 11:13:47, Axel Grüttner wrote: >Hi, > >i want to test my movegen() performance. i read about "Vince´s"-Position. >how should i continue. I think that vincent's position is a bad idea. generating the same moves again and again will tell you nothing because this is not what you do in a chess game. >is it ok to build a do-while loop around my movegen() an let it run 1 second? >Do i only create white moves? of course it is not ok. In a game you generate moves for both sides. If you test only generating moves for one side then it is a good reason not to trust your test. >does this test including makemove()? I suggest you to ignore vincent tests. A test that include generating moves is going to tell you nothing because it is possible to make one function faster at the price of doing the second function slower. what i mean is should i only calculate >possible next moves or do i have to calculate one, make() it, unmake() it and >then calculate the next one? > >axel The only relevant test of your move generator is relative to previous version. If you change your move generator and you search in games more nodes per seconds without changing nothing in your evaluation or search algorithm then you have an improvement. 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.