Shredder 11 rating

Discussion of computer chess matches and engine tournaments.

Moderator: Ras

IWB
Posts: 1539
Joined: Thu Mar 09, 2006 2:02 pm

Shredder 11 rating

Post by IWB »

Hi all

This is a part of my rating list. Of course I deleted all Betas, therefore some engines have a higher numbers of games. Overall I played 53000+ games for this test.

Conditions:

1Thread all versions!
Ponder on
6 min /game + 3 sec/move (averages at about 25h per 100 games)
256 MB Hash
Only 3 + 4 Pc Tbs or Sbases
No books
No learning
50 Nunn2 positions, changing colors - 100 games engine vs engine

I played a full round robin with the important engines, nevertheless a lot of Shredder-betas played vs some engines. If this engine is by chance a under- or overperformer vs Shredder its rating might have shifted a bit up or down. Especially in the middle part all engines are very close together so their ranking has to be seen with the nessesary caution!

S10 is fixed at 2830 as this is the same rating as in the last SSDF rating list, the only other rating list with ponder on I know off.

Code: Select all


    Program                          Elo    +   -   Games   Score   Av.Op.  Draws

  1 Rybka 2.3.2a mp 1T x64         : 3045   12  12  2400    71.7 %   2883   35.7 %
  2 Rybka 1.2f x64                 : 2992    8   8  4800    67.9 %   2862   31.5 %
  3 Deep Shredder 11 x64           : 2922   14  14  1600    58.8 %   2860   33.7 %
  4 Zappa Mexico X64 Final         : 2913   13  13  1600    53.7 %   2887   39.2 %
  5 Rybka 1.0 Beta x64             : 2911   13  13  1900    53.5 %   2887   36.1 %
  6 Naum 2.2 x64                   : 2863   12  12  1900    45.5 %   2894   43.9 %
  7 Rybka 1.0 Beta 32-bit          : 2862    8   8  4400    50.1 %   2862   33.2 %
  8 Fruit 05/11/03 x32             : 2856   11  11  2400    46.8 %   2878   39.2 %
  9 HIARCS 11.1                    : 2855    8   8  5100    47.5 %   2872   35.8 %
 10 Fruit 2.3 x32                  : 2851   22  22   600    51.9 %   2838   39.2 %
 11 DS 10 Balmung x64              : 2850   10  10  2600    49.9 %   2851   42.5 %
 12 Loop 13.5 x64                  : 2847    8   8  4100    45.5 %   2878   38.0 %
 13 LoopMP 12.32                   : 2846    9   9  3400    50.7 %   2841   34.2 %
 14 Loop M1-P x64                  : 2842   24  24   500    49.2 %   2848   39.6 %
 15 Toga II 1.2.1a x32             : 2841    7   7  6400    47.1 %   2861   35.2 %
 16 ListMP 11.64b                  : 2837   12  12  2100    44.2 %   2877   35.6 %
 17 Deep Shredder 10 x64           : 2830    8   8  5000    45.1 %   2864   37.8 %
 18 HIARCS 11 MP 1T                : 2827   19  19   800    44.9 %   2862   36.4 %
 19 Naum 2.1 x64                   : 2825    9   9  3700    44.6 %   2863   36.4 %
 20 Toga II 1.3x4                  : 2818   19  19   800    44.6 %   2856   38.9 %
 21 Hiarcs X54 64bit               : 2816   22  22   600    43.9 %   2859   36.5 %
 22 Spike 1.2 Turin                : 2795    7   7  5400    40.4 %   2863   36.2 %
 23 DS 9.02 1T x64                 : 2786   17  17  1200    41.1 %   2848   30.4 %
 24 Deep Sjeng 2.7 x32             : 2781   17  17  1000    36.6 %   2876   38.1 %
 25 Glaurung 2-epsilon/5 x64       : 2764   17  17  1100    37.0 %   2856   31.2 %
 26 Deep Sjeng 2.5 x32             : 2705   20  20   900    29.9 %   2853   30.6 %
My very personal impression, which is not secured by a significant number of games but just on a few long games on playchess with my old DUAL AMD vs some big animals there is, that with longer time controls S11 will be a hard opponent for them!

This is just to give a first impression about the new Shredder 11. Please have a look at the common rating lists after its release to get another impression!

Bye
Ingo
Alessandro Scotti

Re: Shredder 11 rating

Post by Alessandro Scotti »

Good news but... I hope this time there is something for us Mac users too! :roll:
User avatar
geots
Posts: 4790
Joined: Sat Mar 11, 2006 12:42 am

Re: Shredder 11 rating

Post by geots »

Ingo, im assuming this Shredder 11 you are running is a beta you are testing for Stefan, and nothing has been released to the public. This is a correct assumption, right? Thanks,
Tony Thomas

Re: Shredder 11 rating

Post by Tony Thomas »

Very promising, there is one correction. WBEC uses ponder on.
IWB
Posts: 1539
Joined: Thu Mar 09, 2006 2:02 pm

Re: Shredder 11 rating

Post by IWB »

Hello
geots wrote:Ingo, im assuming this Shredder 11 you are running is a beta you are testing for Stefan, and nothing has been released to the public. This is a correct assumption, right? Thanks,
The Shredder 11 in the list is no longer a beta, but you are right, it is not published yet. Just a few days and it is out.

Bye
Ingo
IWB
Posts: 1539
Joined: Thu Mar 09, 2006 2:02 pm

Re: Shredder 11 rating

Post by IWB »

Tony Thomas wrote:Very promising, there is one correction. WBEC uses ponder on.
Hello

Now I know another one besides the SSDF.

I just had a look, unfortunately the number of games is ... very limited, therefore the number of engines is huge.

Thx
Ingo
Tony Thomas

Re: Shredder 11 rating

Post by Tony Thomas »

IWB wrote:
Tony Thomas wrote:Very promising, there is one correction. WBEC uses ponder on.
Hello

Now I know another one besides the SSDF.

I just had a look, unfortunately the number of games is ... very limited, therefore the number of engines is huge.

Thx
Ingo
Yes, it is true that the number of games are low. It is more of a question of quality vs quantity.
Spock

Re: Shredder 11 rating

Post by Spock »

Nice also to see that WBEC use the Bayesian Elo Rating system
Tony Thomas

Re: Shredder 11 rating

Post by Tony Thomas »

Spock wrote:Nice also to see that WBEC use the Bayesian Elo Rating system
I think he was the first one who started using it. If I remember correctly he was using Bayesian Elo even back during the 12th edition.
IWB
Posts: 1539
Joined: Thu Mar 09, 2006 2:02 pm

Re: Shredder 11 rating

Post by IWB »

Tony Thomas wrote:
Spock wrote:Nice also to see that WBEC use the Bayesian Elo Rating system
I think he was the first one who started using it. If I remember correctly he was using Bayesian Elo even back during the 12th edition.
Hello

I tried Byesian ELo as well but the difference where within a 1 Elo range due to rounding. I played allways the same number of black and white games vs each opponent. In that case Elostat or Bayesian are identical. The only difference is the usage. Elostat is much more simple to use.

Bye
Ingo