Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Counting moves with 'perft n'

Author: Robert Hyatt

Date: 08:28:18 09/11/02

Go up one level in this thread


On September 11, 2002 at 11:09:20, Ruben Razquin wrote:

>People have recently posted results of move generation speed for several engines
>using de 'perft n' command. I did not get the same number of moves when I
>performed some tests with the engine I am developing.
>
>'perft 1' gives 20 moves for new game's piece position.
>'perft 2' gives 420 moves, while other engines give 400 moves. I am counting 20
>moves generated for root level plus 400 moves generated at the next level.

this should just count "terminal positions".  Not interior as well...

No particular reason not to count them other than this is the way I first
wrote this code, and everybody else has been using it to check their move
generator and make/unmake code by comparing the numbers.

Note that illegal moves/positions do not count.


> White
>has 20 possible moves and black can reply with 20 diferent moves. That
>represents 400 possible chess positions, but the engine has to generate and call
>the move/unmove functions 420 times. Why do other engines count 400 moves only?
>Is that correct?
>
>Ruben Razquin



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.