SWCR: Actual results, IvanHoe, Firebird, Gull, Hiarcs

[Dr.Wael Deeb]

Hi Wael,

have a look in the comparsion CEGT to SWCR.
No engine have more as 30 ELO difference.

I think the my SWCR ratings are very exact to the conditions I used.

I think you test with very fast time controls. In this case is the difference to your rating list is clear ... think so!

Best
Frank

Hi Frank,
I test only at 20 minutes + 20 seconds on an overclocked Q9650 4.0 GHz and I think it's considered to be a long time control
Anyway,I am waiting eagerly for your results....
Dr.D

[Frank Quisinsky]

Hi Wael,

I stopped after 506 games.
Now up to Monday in the afternoon the other three new engines are playing.

The latest 416 games of Firebird 1.1 WD x64 I will start Monday in the afternoon.

Best
Frank

[Dr.Wael Deeb]

Hi Wael,

I stopped after 506 games.
Now up to Monday in the afternoon the other three new engines are playing.

The latest 416 games of Firebird 1.1 WD x64 I will start Monday in the afternoon.

Best
Frank

Hi,
Very good....it's keeping solid 72% performance thus far....I'd say an excellent performance....
Dr.D

[Frank Quisinsky]

Hi Wael,

here are the final results for your setting:
Only the first 22 ...

Code: Select all

Rank Name                          Elo    +    - games score oppo. draws 
   1 Houdini 1.03a x64            2945   22   21   920   79%  2715   29% 
   2 Rybka 4 x64                  2936   17   17  1506   80%  2698   30% 
   3 IvanHoe B52aC x64            2912   27   26   525   74%  2737   36% 
   4 Rybka 3 x64                  2902   23   22   840   78%  2684   28% 
   5 Stockfish 1.8.0 JA x64       2901   18   18  1186   74%  2713   33% 
   6 Fire 1.31 x64                2898   21   20   920   75%  2718   36% 
   7 Rybka 4                      2895   20   20   960   76%  2704   32% 
   8 FireBird 1.1 x64 WD          2894   20   19   920   73%  2732   41% 
   9 Stockfish 1.7.1 JA x64       2893   19   18  1120   76%  2703   34% 
  10 Stockfish 1.7.1 JA           2875   18   18  1200   75%  2691   31% 
  11 Stockfish 1.8.0 JA           2864   20   20   920   74%  2698   37% 
  12 Rybka 3                      2861   16   16  1520   74%  2690   31% 
  13 Critter 0.80 x64             2831   18   17  1186   66%  2716   35% 
  14 Naum 4.2 x64                 2829   15   14  1746   67%  2706   36% 
  15 Stockfish 1.6.3 JA           2828   18   18  1080   71%  2681   36% 
  16 Naum 4.2                     2825   16   16  1440   68%  2701   36% 
  17 Critter 0.80                 2815   21   20   800   66%  2704   39% 
  18 Naum 4.1                     2812   20   19   920   68%  2686   35% 
  19 Critter 0.70 x64             2804   20   20   880   65%  2698   38% 
  20 Stockfish 1.6.0 JA           2803   19   19   960   68%  2688   39% 
  21 Shredder 12                  2800   10   10  3706   64%  2703   36% 
  22 Komodo 1.2 JA x64            2799   15   15  1506   63%  2702   40% 

At the moment the latest 395 games for IvanHoe B52ac x64 are still running.

FireBird 1.1 x64 WD
920 games (+482,=379,- 59), 73.0 %

Code: Select all

Shredder 12                   :  40 (+ 19,= 16,-  5), 67.5 %
Hiarcs 13.1                   :  40 (+ 21,= 16,-  3), 72.5 %
Zappa Mexico II x64           :  40 (+ 22,= 17,-  1), 76.2 %
Protector 1.3.4 JA x64        :  40 (+ 26,= 12,-  2), 80.0 %
Naum 4.2 x64                  :  40 (+ 13,= 25,-  2), 63.7 %
Loop 2007 x64                 :  40 (+ 35,=  5,-  0), 93.8 %
SmarThink 1.20 x64            :  40 (+ 31,=  9,-  0), 88.8 %
Thinker 5.4d Inert x64        :  40 (+ 25,= 14,-  1), 80.0 %
Sjeng WC-2008 x64             :  40 (+ 26,= 13,-  1), 81.2 %
Komodo 1.2 JA x64             :  40 (+ 15,= 24,-  1), 67.5 %
Junior 11.2 x64               :  40 (+ 19,= 15,-  6), 66.2 %
Fruit 09_07_05 x64            :  40 (+ 21,= 15,-  4), 71.2 %
Rybka 4 x64                   :  40 (+  4,= 27,-  9), 43.8 %
Onno 1.2.70 x64               :  40 (+ 26,= 11,-  3), 78.8 %
Spark 0.5 x64                 :  40 (+ 22,= 15,-  3), 73.8 %
Stockfish 1.8.0 JA x64        :  40 (+ 10,= 25,-  5), 56.2 %
Hannibal 1.0a x64             :  40 (+ 20,= 19,-  1), 73.8 %
Critter 0.80 x64              :  40 (+ 14,= 25,-  1), 66.2 %
Crafty 23.3 JA x64            :  40 (+ 30,=  9,-  1), 86.2 %
Umko 1.0 x64                  :  40 (+ 26,= 12,-  2), 80.0 %
Equinox 0.83 x64              :  40 (+ 31,=  8,-  1), 87.5 %
GullChess 1.0a x64            :  40 (+ 22,= 17,-  1), 76.2 %
IvanHoe B52aC x64             :  40 (+  4,= 30,-  6), 47.5 %

Games can be download around October 18th, 2010.
Two other IvanHoes will follow, after this one I update the PGN databases on my webpages. Or did you need the games before?

Since end of the last year up to today its quit clear that the IPP family with Robbolito, Iggorit, Fire, FireBird, IvanHoes, etc. have the same rating +-30. This isn't new for the most.

Quit clear is too, that this engines are in fast blitz games a little bit stronger.

After all what I saw the playing style is more boring as the style from Rybka 3 / 4. A lot of remis games, no aggressive playing style in the middle of the games. Can you see if you make statistics with my database and if you compare it with the others.

And Houdini, for myself the same as Rybka 4 ... perhaps with other parameters, I don't know.

But let us wait of the IvanHoe results. At the moment a compile from "Peterpan" is still playing. After this one the compiles from "Vlad" and "Ahmed". Perhaps one of this engines is in the near from Rybka 4 with the conditions I used.

Code: Select all

Conditions:
Windows XP Prof. x64 Edition
SClassic 4, Fritz 11 GUI
S12: Random-Book 4.1
F12: Perfect 2010

40 moves in 10 minutes
ponder = on
each engine = 1 core
resign / learning = off
hashtables = 256Mb
4TB Nalimov, GaviotaB., egbb,
ShredderB., TotalB. & TripleB.

Intel® Core™2 Quad
4x Q9550, 2,83GHz

Best
Frank

[Frank Quisinsky]

I wrote ...

And Houdini, for myself the same as Rybka 4 ... perhaps with other parameters, I don't know.

Better formulation can be:

And Houdini:
For myself the same as Rybka 4 "WITH more BUGS"

Crash statistic:
I think very important for the IPP family, not for each other engines which running in SWCR ...

36x crashes by Houdini 1.03a x64 with ponder = on under the Shredder Classic 4.0 GUI.

53x crashes by FireBird 1.1 WD x64 with ponder = on under the Shredder Classic 4.0 GUI.

Fire 1.3.1 x64 and IvanHoe B52aC x64 works fine without one crash.

Best
Frank

PS:
You should play games with Junior, Hannibal, Spark, Hiarcs, Stockfish, Shredder, Thinker ... believe me ... you will have more fun in watching. If you like to analyze fast middlegames Rybka and Co. is interesting only. Hard but true!

[Frank Quisinsky]

If you like to analyze fast middlegames ..

Correction, sorry!
I mean late middle games!

Best
Frank

[Dr.Wael Deeb]

If you like to analyze fast middlegames ..

Correction, sorry!
I mean late middle games!

Best
Frank

Thanks Frank for taking the time to test my settings
Next time I will provide you with a monster personality of Deep Junior 11.2....much much stronger than the default one
Dr.D

P.S.Shhhh,don't tell anybody,it's a secret

[Matthias Gemuh]

Actual results

...
Frank

Hi Frank,

"Aktuelle Ergebnisse" translates to "Current Results"

Your English is rapidly improving, though !

Best,
Matthias.

[Frank Quisinsky]

Hi Matthias,

please do not joking about my bad English.
It's my weak spot, honest!

Do you know how many work it is to create an interview in English (means for myself). For each of the interviews I am sitting more as 15 hours. OK, my girlfriend helped me. Questions no problem, but I don't like English.

Weak spot --> Wunder Punkt / Achillesferse

But I think you would help a bit only ... in this case have many thanks Matthias.

Best
Frank

[Frank Quisinsky]

Hi Wael,

I hope you aren't disappointed about the results. ELOstat rating is better, 1 ELO more as Fire 1.31 x64

From all what I saw from IPP family ... the IvanHoe B52aC only is playing with a bit more aggressiveness.

Possible that in fast blitz all the ipps are clearly stronger but with longer time controls these engines will lose more and more power. Perhaps important code isn't included. Often I am thinking who is the endgame code Rybka have

Not a good idea to use this engines for long time analyzes / correspondence chess.

Ivanhoe and Fire playing a little bit other chess as Rybka. But the reason can be that the endgame code isn't included

I don't know ...
Seems that all is the same with very, very little differents.

After 620 Firebird 1.0 Beta games I got an ELO from 2.920. Now, an half year later, all the updates which comes daily are not better. I will see now the other IvanHoe compiles, perhaps one of this compiles will reach the ELO Firebird 1.0 Beta or Robbolito 0.9 have ... around 2.920!

Three games won your setting up to move 55. Only Ktulu and Komodo have a more worse statistic. And the same for the other IPPs.

I think the IPP family try to switch very fast knight, bishops, queens. Clear to win the late middlegame, slaughtering without to get an advantage in the early middle game ... and so ... many games will get the result: REMIS.

Its not nice to watch the games Wael.

But, after all ... most interesting from all are IvanHoe B52aC from the engines I saw.

And yes ...
Junior isn't so strong as the IPP family but the games are great. Very individual own ideas to open a position or to play an attack.

I worte it before:
Rybka or IPP family are very strong but strong only in the late middle game. Perhaps 100 ELO stronger as all other chess engines and interesting for analyzes ... but the games are very boring for myself. In the early middle game and in the opening I think 20 chess programs will play better and more interesting chess ... with better results.

Best
Frank