Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Rybka 1.01 Beta 7 32-bit results from all four settings-nunn2

Author: Vasik Rajlich

Date: 05:32:28 01/06/06

Go up one level in this thread


On January 05, 2006 at 14:18:52, Jonas Cohonas wrote:

>Hardware P4 2.8 Ghz 1 Gb ram
>WinXP PRO
>
>All engines with 256 mb ram
>Nunn2
>Gauntlet
>Ponder=off
>Time control 4 + 2
>No EGTB's
>
>You can download all the games in .pgn and chessbase format (you will find that
>in a zip folder of the zip folder :) and i have done all the elo and stats
>calculations for each match relevant setting:
>
>http://www.fusionweb.dk/chess/rybka/rybkagames.zip
>
>VP=Very Positional
>SP=Slightly Positional
>ST=Slightly Tactical
>VT=Very Tactical
>
>Program                          Elo    +   -   Games   Score   Av.Op.  Draws
>
>1 Rybka 1.01 Beta 7 32-bit       : 2629   46  45   160    58.1 %   2572   30.0 %
>2 Fruit 2.2.1                    : 2589   59  59    80    44.4 %   2629   41.2 %
>3 Fritz 9                        : 2554   70  71    80    39.4 %   2629   18.8 %
>
>    Individual statistics:
>
>  1 Rybka 1.01 Beta 7 32-bit  : 2629  160 (+ 69,= 48,- 43), 58.1 %
>
>    Fruit 2.2.1                   :  80 (+ 28,= 33,- 19), 55.6 %
>    Fritz 9                       :  80 (+ 41,= 15,- 24), 60.6 %
>
>Games        :    160 (finished)
>
>White Wins   :     65 (40.6 %)
>Black Wins   :     47 (29.4 %)
>Draws        :     48 (30.0 %)
>Unfinished   :      0
>
>White Perf.  : 55.6 %
>Black Perf.  : 44.4 %
>
>ECO A =     48 Games (30.0 %)
>ECO B =    112 Games (70.0 %)
>ECO C =      0 Games ( 0.0 %)
>ECO D =      0 Games ( 0.0 %)
>ECO E =      0 Games ( 0.0 %)
>
>As you can see a very good score against both opponents, but this is all the
>settings results combined and some performed better than others, so have a look
>at the details for a better view of what setting did best.
>
>Watching a great portion of the games was a real treat, i will try and come up
>with some more indepth feedback, but will first do so after some longer timed
>matches where the moves themselves have more relevance as opposed to this
>experiement where it was more about testing the different settings.
>
>Vas you have done one heck of a job, i am _really_ impressed!!
>
>Regards
>Jonas

Hi Jonas,

thanks for testing. To summarize, you found:

SP 2649
VP 2645
VT 2619
ST 2604

It's not a lot of games, but this is the exact same order in which I've ranked
these four settings after now several thousand games from various sources.

In the next Rybka release, there is a new default "moderately positional"
setting which is about 75% SP and 25% VP. More importantly, a bunch of other
things were changed and some new settings were added.

Best regards,
Vas



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.