Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: CM3000? Slowing System?

Author: Didzis Cirulis

Date: 06:38:22 12/03/98

Go up one level in this thread


On December 02, 1998 at 14:52:39, Christophe Theron wrote:

>On December 01, 1998 at 15:02:48, John Merlino wrote:
>
>>On November 30, 1998 at 02:03:43, Christophe Theron wrote:
>>
>>>On November 30, 1998 at 00:16:47, Reynolds Takata wrote:
>>>
>>>>I attempted to play CM3000(in micro chess layout) against CM4000, CM5500, and
>>>>CM6000.  CM3000 was actually winning about 30% of the games!  I expected that it
>>>>wouldn't win any, as i have found CM3000 positionally weak, though the tactics
>>>>aren't too bad.  These games were on one comp running both progs, how much
>>>>effect should this have?
>>>
>>>If you play programs on one computer, you should disable thinking on opponent's
>>>time on both, or else the result means nothing.
>>>
>>>
>>>    Christophe
>>
>>The main problem is verifying that both engines are getting equal CPU time.
>>Unfortunately, this is highly unlikely because Chessmaster 3000 is a DOS
>>application and 4000 and 5500 are Windows applications. Due to the way Windows
>>switches back and forth between DOS applications, it is very unlikely that both
>>programs are getting relatively equal time to think, regardless of whether you
>>have both engines thinking during their opponent's time.
>>Windows tends to give a higher priority to DOS applications, which accounts for
>>CM3000's unexpectedly high win percentage.
>>Turning off "deep thinking" will not make a difference as far as CPU time goes,
>>because Windows switches back and forth between apps very quickly. What it WILL
>>do, however, is make the engines play worse overall. And, I imagine that this is
>>unintended.
>>
>>jm
>
>
>You can play a DOS program against a Windows program giving equal chances to
>both programs. You can also play 2 DOS programs.
>
>It will work with most programs except some bad behaviour Windows programs.
>
>The MAIN thing is to disable permanent brain (or deep thinking, or thinking on
>opponent's time) in BOTH programs.
>
>Then just set up the properties of your DOS box to:
>* Run the DOS program full screen and switch between the programs with Alt-Tab
>* Give the DOS box maximum CPU time when it has the focus (sensitivity slider
>all the way to the left).
>* Give the DOS box NO CPU time when it is in the background (always suspend when
>in the background).
>
>When you use these settings, Windows will give 97% CPU time to the program that
>gets the focus, and will not switch back and forth between apps as you say. You
>should of course close any other program.
>
>However this does not work with some bad written Windows programs. For example
>Genius5 will steal 30% of CPU time to the DOS program. To solve this, use the
>DOS version of Genius5 and use the settings described for a DOS box.
>
>Before doing a match, you should ensure that no program runs slower. Use a test
>position at a given depth, and run the program on this position:
>1) when it is alone in the system
>2) when the other program is open in the background
>
>If you follow these rules, you can play fair matches manually using only one
>computer. I do this very often, and believe me I am much concerned about
>fairness, as I use these matches to decide if Chess Tiger plays better or not.
>
>
>    Christophe

If you are interested, I could explain HOW to do calculations for the fair
matches of two chess programs on one and the same PC. Email me or wait until
this information appears in a subsection of my Computer Chess Site.
http://www.konts.lv/usr/Didzis/index.html
I am doing the testing of Chess Tiger for Christophe... :-)

Didzis

Didzis



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.