Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Reproducibility of Nunn matches is not so good

Author: Christophe Theron

Date: 00:49:22 05/02/00

Go up one level in this thread


On May 02, 2000 at 03:37:48, Ulrich Tuerke wrote:

>On May 02, 2000 at 03:29:51, Georg v. Zimmermann wrote:
>
>>
>>A small difference (happens all the time : Computer is a bit faster / slower,
>>initial swaping, a bit less time, HT filled with a different numbers, parallel
>>randomness .... ) can result in a different move which will result in a
>>different game.
>>
>>Ever heared of Chaos Theory ? :-)
>>
>>Sincerely,
>>Butterfly
>
>I understood, the match had been repeated on the same machines, each prog
>getting 100% of the cpu.



It's closer to 97%. Windows steals approx. 3%.



> I consider the chance for producing another game from
>the same start position as microscopically small.



The probability is BIG. I happens all the time.




>But in principle, this possibilty is non-zero; so far I agree. But I would never
>exepect to observe such drastic differences as posted here !



I was surprised myself when I first tried.

The phenomenon increases as time control decreases. Because the time granularity
becomes bigger.

A one second search is divided into approx 18 clock ticks. So a 5% time
difference is very likely in this case, and this is enough to disturb a chess
program's time allocation manager.



    Christophe




>Of course, the thing is very different if you play from a book containing
>weighted moves.



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.