YABRL: Fruit 1.5, the "bombshell" in action ...

[Robert Allgeuer]

Geschrieben von:/Posted by: Robert Allgeuer at 01 July 2004 11:48:54:

I think everyone must have expected Fruit 1.5 to score very well, but I was then still a bit surprised, when under my conditions Fruit scored at the same level as engines such as Yace, SoS and Pepito.
For exact conditions, please refer to the link below. YABRL will go now into a break, because I will leave on holiday for 4 weeks ...
Robert




    Program                     Elo    +   -   Games   Score   Av.Op.  Draws
 01 Chess Tiger 15.0 normal   : 2718   17  27   834    73.0 %   2545   27.3 %
 02 Ruffian v2.1.0            : 2676   16  24   958    69.4 %   2534   25.3 %
 03 List v5.12                : 2674   18  25   838    68.4 %   2540   27.0 %
 04 Ruffian v2.0.0            : 2673   17  25   900    70.1 %   2524   27.3 %
 05 Ruffian v2.0.2dvbk        : 2672   17  24   894    68.4 %   2538   27.4 %
 06 Ruffian v1.0.1            : 2649   16  22  1015    67.7 %   2520   27.9 %
 07 DeepSjeng v1.6ntb         : 2624   18  22   938    62.9 %   2532   24.7 %
 08 Rebel v12.00.01           : 2617   21  22   818    59.9 %   2547   25.2 %
 09 Aristarch v4.50           : 2616   19  22   898    62.0 %   2531   25.6 %
 10 Patriot v1.2.3cpbk        : 2599   20  21   897    59.6 %   2531   25.6 %
 11 SmarThink v0.17a          : 2593   19  19  1019    58.6 %   2532   26.3 %
 12 Ktulu v4.2                : 2590   19  20  1012    58.6 %   2529   23.8 %
 13 Crafty v17.14DC           : 2584   18  16  1183    58.2 %   2527   33.1 %
 14 Ktulu v5.0                : 2582   21  22   874    57.2 %   2532   19.9 %
 15 Thinker v4.5b             : 2581   20  17   976    56.9 %   2533   33.6 %
 16 Aristarch v4.37           : 2581   20  17   959    56.7 %   2534   36.4 %
 17 Aristarch v4.21           : 2580   18  17  1177    57.2 %   2529   25.2 %
 18 Crafty v19.06DCntb        : 2578   19  17  1059    56.6 %   2532   30.8 %
 19 Delfi v4.3                : 2564   19  17  1140    55.2 %   2528   24.8 %
 20 El Chinito v3.25          : 2563   20  18   980    54.7 %   2530   27.3 %
 21 Delfi v4.2                : 2561   25  25   580    58.1 %   2504   27.2 %
 22 Crafty-MPC v18.15DC       : 2560   19  17  1121    54.6 %   2528   27.8 %
 23 Little Goliath 2000 v3.9  : 2559   18  16  1220    54.5 %   2528   26.5 %
 24 SmarThink v0.16b++        : 2558   21  21   836    58.0 %   2503   24.3 %
 25 SoS 3                     : 2557   18  17  1219    54.2 %   2528   23.3 %
 26 Pepito v1.59 profile      : 2556   19  16  1220    54.1 %   2528   24.3 %
 27 Crafty v18.15DC           : 2556   22  22   741    59.0 %   2493   29.0 %
 28 Fruit v1.5                : 2556   24  20   739    52.6 %   2538   25.8 %
 29 Yace Paderborn            : 2553   19  16  1200    53.9 %   2526   26.6 %
 30 Aristarch v4.4            : 2547   36  34   319    54.1 %   2518   20.4 %
 31 SoS 4                     : 2547   21  18   999    52.3 %   2531   23.7 %
 32 Yace v0.99.56             : 2541   33  29   380    54.3 %   2511   26.6 %
 33 Little Goliath 2000 v3.5  : 2537   31  25   440    53.6 %   2511   30.9 %
 34 Green Light Chess v3.00   : 2536   20  15  1220    51.1 %   2528   26.0 %
 35 Tao v5.6                  : 2525   18  22   935    48.7 %   2534   24.4 %
 36 Anmon v5.30               : 2524   17  22   979    49.0 %   2531   26.5 %
 37 Amyan v1.59               : 2510   17  21  1030    48.6 %   2520   26.2 %
 38 Pharaon v2.62             : 2509   17  20  1139    48.3 %   2521   23.7 %
 39 Crafty v19.01DC           : 2499   24  19   815    50.4 %   2496   25.5 %
 40 LambChop v10.99           : 2495   17  18  1217    45.2 %   2529   22.5 %
 41 Gromit v3.8.2             : 2491   17  18  1192    44.6 %   2529   23.9 %
 42 Ktulu v3.9                : 2490   19  24   779    48.6 %   2500   26.1 %
 43 SlowChess v2.89b          : 2488   18  20  1013    45.1 %   2522   24.3 %
 44 KnightDreamer v3.2        : 2484   18  19  1120    43.5 %   2529   24.0 %
 45 Anmon v5.22               : 2483   19  22   899    46.4 %   2508   26.7 %
 46 Comet B44-2               : 2481   18  18  1140    43.0 %   2529   25.4 %
 47 SoS v11-99                : 2476   33  34   359    46.0 %   2505   17.3 %
 48 Tao v5.4                  : 2475   19  19  1039    44.1 %   2516   21.0 %
 49 Amy v0.8.3                : 2472   19  17  1193    42.0 %   2528   19.4 %
 50 Dragon v4.4.3             : 2466   19  18  1081    41.2 %   2528   25.7 %
 51 Comet B62-3               : 2461   19  18  1119    40.3 %   2529   25.3 %
 52 Francesca M.0.0.9         : 2444   19  16  1218    37.9 %   2530   24.4 %
 53 PostModernist v1.007      : 2441   20  17  1138    37.5 %   2530   23.9 %
 54 Comet B60                 : 2434   22  21   780    41.2 %   2496   25.6 %
 55 Leila v0.53h              : 2425   22  17  1098    36.0 %   2525   20.0 %
 56 Tcb v0045                 : 2421   21  16  1139    34.7 %   2531   23.9 %
 57 Resp v0.19                : 2402   22  16  1120    32.4 %   2530   22.9 %
 58 Arasan v7.4               : 2398   27  18   808    31.9 %   2529   23.0 %
 59 Nejmet v3.07              : 2388   25  18   876    33.2 %   2510   22.3 %
 60 SlowChess v2.78           : 2378   27  19   790    33.5 %   2497   19.6 %
 61 Exchess v4.03             : 2324   29  14  1119    23.3 %   2532   20.6 %
 62 Beowulf v2.2              : 2306   32  14  1120    22.2 %   2524   17.2 %

Games        :  29929 (finished)
White Wins   :  12241 (40.9 %)
Black Wins   :  10167 (34.0 %)
Draws        :   7521 (25.1 %)
Unfinished   :      0
White Perf.  : 53.5 %
Black Perf.  : 46.5 %
ECO A =   4553 Games (15.2 %)
ECO B =   8937 Games (29.9 %)
ECO C =   5730 Games (19.1 %)
ECO D =   5359 Games (17.9 %)
ECO E =   5350 Games (17.9 %)

(28) Fruit v1.5                : 739 (+293,=191,-255), 52.6 %
Ruffian v2.1.0                :  20 (+  2,=  7,- 11), 27.5 %
Chess Tiger 15.0 normal       :  20 (+  5,=  3,- 12), 32.5 %
Rebel v12.00.01               :  20 (+  4,=  5,- 11), 32.5 %
List v5.12                    :  20 (+  4,=  5,- 11), 32.5 %
SoS 3                         :  20 (+  6,=  3,- 11), 37.5 %
SoS 4                         :  20 (+  7,=  3,- 10), 42.5 %
DeepSjeng v1.6ntb             :  20 (+  7,=  7,-  6), 52.5 %
Aristarch v4.50               :  20 (+  5,=  4,- 11), 35.0 %
Patriot v1.2.3cpbk            :  20 (+  7,=  6,-  7), 50.0 %
Ktulu v4.2                    :  20 (+  8,=  1,- 11), 42.5 %
Thinker v4.5b                 :  20 (+  6,=  9,-  5), 52.5 %
SmarThink v0.17a              :  20 (+  7,=  8,-  5), 55.0 %
Crafty v17.14DC               :  20 (+  7,=  6,-  7), 50.0 %
Ktulu v5.0                    :  19 (+  2,=  6,- 11), 26.3 %
Crafty v19.06DCntb            :  20 (+  6,=  8,-  6), 50.0 %
Delfi v4.3                    :  20 (+  9,=  4,-  7), 55.0 %
El Chinito v3.25              :  20 (+  6,=  8,-  6), 50.0 %
Crafty-MPC v18.15DC           :  20 (+  5,=  6,-  9), 40.0 %
Little Goliath 2000 v3.9      :  20 (+  7,=  2,- 11), 40.0 %
Pepito v1.59 profile          :  20 (+  6,=  2,- 12), 35.0 %
Yace Paderborn                :  20 (+  6,=  7,-  7), 47.5 %
Green Light Chess v3.00       :  20 (+  5,=  9,-  6), 47.5 %
Tao v5.6                      :  20 (+ 13,=  3,-  4), 72.5 %
Anmon v5.30                   :  20 (+  7,=  7,-  6), 52.5 %
LambChop v10.99               :  20 (+ 13,=  4,-  3), 75.0 %
Gromit v3.8.2                 :  20 (+  8,=  5,-  7), 52.5 %
KnightDreamer v3.2            :  20 (+ 10,=  7,-  3), 67.5 %
Comet B44-2                   :  2&#4

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 01 July 2004 12:00:40:
Als Antwort auf:/In reply to: YABRL: Fruit 1.5, the "bombshell" in action ... geschrieben von:/posted by: Robert Allgeuer at 01 July 2004 11:48:54:

I think everyone must have expected Fruit 1.5 to score very well, but I was then still a bit surprised, when under my conditions Fruit scored at the same level as engines such as Yace, SoS and Pepito.
For exact conditions, please refer to the link below. YABRL will go now into a break, because I will leave on holiday for 4 weeks ...
Robert

>
>    Program                     Elo    +   -   Games   Score   Av.Op.  Draws
> 01 Chess Tiger 15.0 normal   : 2718   17  27   834    73.0 %   2545   27.3 %
> 02 Ruffian v2.1.0            : 2676   16  24   958    69.4 %   2534   25.3 %
> 03 List v5.12                : 2674   18  25   838    68.4 %   2540   27.0 %
> 04 Ruffian v2.0.0            : 2673   17  25   900    70.1 %   2524   27.3 %
> 05 Ruffian v2.0.2dvbk        : 2672   17  24   894    68.4 %   2538   27.4 %
> 06 Ruffian v1.0.1            : 2649   16  22  1015    67.7 %   2520   27.9 %
> 07 DeepSjeng v1.6ntb         : 2624   18  22   938    62.9 %   2532   24.7 %
> 08 Rebel v12.00.01           : 2617   21  22   818    59.9 %   2547   25.2 %
> 09 Aristarch v4.50           : 2616   19  22   898    62.0 %   2531   25.6 %
> 10 Patriot v1.2.3cpbk        : 2599   20  21   897    59.6 %   2531   25.6 %
> 11 SmarThink v0.17a          : 2593   19  19  1019    58.6 %   2532   26.3 %
> 12 Ktulu v4.2                : 2590   19  20  1012    58.6 %   2529   23.8 %
> 13 Crafty v17.14DC           : 2584   18  16  1183    58.2 %   2527   33.1 %
> 14 Ktulu v5.0                : 2582   21  22   874    57.2 %   2532   19.9 %
> 15 Thinker v4.5b             : 2581   20  17   976    56.9 %   2533   33.6 %
> 16 Aristarch v4.37           : 2581   20  17   959    56.7 %   2534   36.4 %
> 17 Aristarch v4.21           : 2580   18  17  1177    57.2 %   2529   25.2 %
> 18 Crafty v19.06DCntb        : 2578   19  17  1059    56.6 %   2532   30.8 %
> 19 Delfi v4.3                : 2564   19  17  1140    55.2 %   2528   24.8 %
> 20 El Chinito v3.25          : 2563   20  18   980    54.7 %   2530   27.3 %
> 21 Delfi v4.2                : 2561   25  25   580    58.1 %   2504   27.2 %
> 22 Crafty-MPC v18.15DC       : 2560   19  17  1121    54.6 %   2528   27.8 %
> 23 Little Goliath 2000 v3.9  : 2559   18  16  1220    54.5 %   2528   26.5 %
> 24 SmarThink v0.16b++        : 2558   21  21   836    58.0 %   2503   24.3 %
> 25 SoS 3                     : 2557   18  17  1219    54.2 %   2528   23.3 %
> 26 Pepito v1.59 profile      : 2556   19  16  1220    54.1 %   2528   24.3 %
> 27 Crafty v18.15DC           : 2556   22  22   741    59.0 %   2493   29.0 %
> 28 Fruit v1.5                : 2556   24  20   739    52.6 %   2538   25.8 %
> 29 Yace Paderborn            : 2553   19  16  1200    53.9 %   2526   26.6 %
> 30 Aristarch v4.4            : 2547   36  34   319    54.1 %   2518   20.4 %
> 31 SoS 4                     : 2547   21  18   999    52.3 %   2531   23.7 %
> 32 Yace v0.99.56             : 2541   33  29   380    54.3 %   2511   26.6 %
> 33 Little Goliath 2000 v3.5  : 2537   31  25   440    53.6 %   2511   30.9 %
> 34 Green Light Chess v3.00   : 2536   20  15  1220    51.1 %   2528   26.0 %
> 35 Tao v5.6                  : 2525   18  22   935    48.7 %   2534   24.4 %
> 36 Anmon v5.30               : 2524   17  22   979    49.0 %   2531   26.5 %
> 37 Amyan v1.59               : 2510   17  21  1030    48.6 %   2520   26.2 %
> 38 Pharaon v2.62             : 2509   17  20  1139    48.3 %   2521   23.7 %
> 39 Crafty v19.01DC           : 2499   24  19   815    50.4 %   2496   25.5 %
> 40 LambChop v10.99           : 2495   17  18  1217    45.2 %   2529   22.5 %
> 41 Gromit v3.8.2             : 2491   17  18  1192    44.6 %   2529   23.9 %
> 42 Ktulu v3.9                : 2490   19  24   779    48.6 %   2500   26.1 %
> 43 SlowChess v2.89b          : 2488   18  20  1013    45.1 %   2522   24.3 %
> 44 KnightDreamer v3.2        : 2484   18  19  1120    43.5 %   2529   24.0 %
> 45 Anmon v5.22               : 2483   19  22   899    46.4 %   2508   26.7 %
> 46 Comet B44-2               : 2481   18  18  1140    43.0 %   2529   25.4 %
> 47 SoS v11-99                : 2476   33  34   359    46.0 %   2505   17.3 %
> 48 Tao v5.4                  : 2475   19  19  1039    44.1 %   2516   21.0 %
> 49 Amy v0.8.3                : 2472   19  17  1193    42.0 %   2528   19.4 %
> 50 Dragon v4.4.3             : 2466   19  18  1081    41.2 %   2528   25.7 %
> 51 Comet B62-3               : 2461   19  18  1119    40.3 %   2529   25.3 %
> 52 Francesca M.0.0.9         : 2444   19  16  1218    37.9 %   2530   24.4 %
> 53 PostModernist v1.007      : 2441   20  17  1138    37.5 %   2530   23.9 %
> 54 Comet B60                 : 2434   22  21   780    41.2 %   2496   25.6 %
> 55 Leila v0.53h              : 2425   22  17  1098    36.0 %   2525   20.0 %
> 56 Tcb v0045                 : 2421   21  16  1139    34.7 %   2531   23.9 %
> 57 Resp v0.19                : 2402   22  16  1120    32.4 %   2530   22.9 %
> 58 Arasan v7.4               : 2398   27  18   808    31.9 %   2529   23.0 %
> 59 Nejmet v3.07              : 2388   25  18   876    33.2 %   2510   22.3 %
> 60 SlowChess v2.78           : 2378   27  19   790    33.5 %   2497   19.6 %
> 61 Exchess v4.03             : 2324   29  14  1119    23.3 %   2532   20.6 %
> 62 Beowulf v2.2              : 2306   32  14  1120    22.2 %   2524   17.2 %
>Games        :  29929 (finished)
>White Wins   :  12241 (40.9 %)
>Black Wins   :  10167 (34.0 %)
>Draws        :   7521 (25.1 %)
>Unfinished   :      0
>White Perf.  : 53.5 %
>Black Perf.  : 46.5 %
>ECO A =   4553 Games (15.2 %)
>ECO B =   8937 Games (29.9 %)
>ECO C =   5730 Games (19.1 %)
>ECO D =   5359 Games (17.9 %)
>ECO E =   5350 Games (17.9 %)
>(28) Fruit v1.5                : 739 (+293,=191,-255), 52.6 %
>Ruffian v2.1.0                :  20 (+  2,=  7,- 11), 27.5 %
>Chess Tiger 15.0 normal       :  20 (+  5,=  3,- 12), 32.5 %
>Rebel v12.00.01               :  20 (+  4,=  5,- 11), 32.5 %
>List v5.12                    :  20 (+  4,=  5,- 11), 32.5 %
>SoS 3                         :  20 (+  6,=  3,- 11), 37.5 %
>SoS 4                         :  20 (+  7,=  3,- 10), 42.5 %
>DeepSjeng v1.6ntb             :  20 (+  7,=  7,-  6), 52.5 %
>Aristarch v4.50               :  20 (+  5,=  4,- 11), 35.0 %
>Patriot v1.2.3cpbk            :  20 (+  7,=  6,-  7), 50.0 %
>Ktulu v4.2                    :  20 (+  8,=  1,- 11), 42.5 %
>Thinker v4.5b                 :  20 (+  6,=  9,-  5), 52.5 %
>SmarThink v0.17a              :  20 (+  7,=  8,-  5), 55.0 %
>Crafty v17.14DC               :  20 (+  7,=  6,-  7), 50.0 %
>Ktulu v5.0                    :  19 (+  2,=  6,- 11), 26.3 %
>Crafty v19.06DCntb            :  20 (+  6,=  8,-  6), 50.0 %
>Delfi v4.3                    :  20 (+  9,=  4,-  7), 55.0 %
>El Chinito v3.25              :  20 (+  6,=  8,-  6), 50.0 %
>Crafty-MPC v18.15DC           :  20 (+  5,=  6,-  9), 40.0 %
>Little Goliath 2000 v3.9      :  20 (+  7,=  2,- 11), 40.0 %
>Pepito v1.59 profile          :  20 (+  6,=  2,- 12), 35.0 %
>Yace Paderborn                :  20 (+  6,=  7,-  7), 47.5 %
>Green Light Chess v3.00       :  20 (+  5,=  9,-  6), 47.5 %
>Tao v5.6                      :  20 (+ 13,=  3,-  4), 72.5 %
>Anmon v5.30                   :  20 (+  7,=  7,-  6), 52.5 %
>LambChop v10.99               :  20 (+ 13,=  4,-  3), 75.0 %
>Gromit v3.8.2                 :  20 (&

[Robert Allgeuer]

Geschrieben von:/Posted by: Robert Allgeuer at 01 July 2004 12:49:35:
Als Antwort auf:/In reply to: Re: YABRL: Fruit 1.5, the "bombshell" in action ... geschrieben von:/posted by: Heinz van Kempen at 01 July 2004 12:00:40:

Hello Robert,
thanks for the interesting tests. At least I am not surprised by the results for Fruit 1.5 and now finally know that I am not a victim of a statistical anomaly. It really has improved like hell in such a short time.
If you want to test more improved engines after your holidays I would recommend WildCat 4.0 and the new Gothmog beta. Maybe also the new Yace 0.99.87.
Best Regards
Heinz

Thanks Heinz,
Wildcat 4 is one of the next, that is for sure. I wanted to test Gothmog for a long time already, but for some reasons it did not run on my system (the feature done=0 related problem). I have not got my hands on any of these betas, but if Gothmog runs on my machine, I will test it of course.
Is Yace 0.99.87 something like an unofficial version?
Robert


YABRL (Yet Another Blitz Rating List)

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 01 July 2004 13:04:46:
Als Antwort auf:/In reply to: Re: YABRL: Fruit 1.5, the "bombshell" in action ... geschrieben von:/posted by: Robert Allgeuer at 01 July 2004 12:49:35:

Thanks Heinz,
Wildcat 4 is one of the next, that is for sure. I wanted to test Gothmog for a long time already, but for some reasons it did not run on my system (the feature done=0 related problem). I have not got my hands on any of these betas, but if Gothmog runs on my machine, I will test it of course.
Is Yace 0.99.87 something like an unofficial version?
Robert

Hi Robert,
it is a pity that you can´t run Gothmog. It plays spectacular chess. Tord is working on Gothmog 1.0 and has further plans for something completely new and even stronger. Maybe those will run on your machine.
For Yace indeed I read somewhere that the version is unofficial and it is not an Dieter´s homepage. Somewhere I found a link where it could be downloaded. Someone posted a disappointed comment after only a few games, but here it plays very well with first three Nunn positions. I will upload this new tournament in about two hours. Also involved Delfi 4.5 and Little Goliath Revival will also be interesting and it will need some time to see if they are improved or not.
Best Regards
Heinz

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 01 July 2004 13:22:16:
Als Antwort auf:/In reply to: Re: YABRL: Fruit 1.5, the "bombshell" in action ... geschrieben von:/posted by: Heinz van Kempen at 01 July 2004 12:00:40:

I think everyone must have expected Fruit 1.5 to score very well, but I was then still a bit surprised, when under my conditions Fruit scored at the same level as engines such as Yace, SoS and Pepito.
For exact conditions, please refer to the link below. YABRL will go now into a break, because I will leave on holiday for 4 weeks ...
Robert

Hello Robert,
thanks for the interesting tests. At least I am not surprised by the results for Fruit 1.5 and now finally know that I am not a victim of a statistical anomaly.

>>
>>    Program                     Elo    +   -   Games   Score   Av.Op.  Draws
>> 01 Chess Tiger 15.0 normal   : 2718   17  27   834    73.0 %   2545   27.3 %
>> 02 Ruffian v2.1.0            : 2676   16  24   958    69.4 %   2534   25.3 %
>> 03 List v5.12                : 2674   18  25   838    68.4 %   2540   27.0 %
>> 04 Ruffian v2.0.0            : 2673   17  25   900    70.1 %   2524   27.3 %
>> 05 Ruffian v2.0.2dvbk        : 2672   17  24   894    68.4 %   2538   27.4 %
>> 06 Ruffian v1.0.1            : 2649   16  22  1015    67.7 %   2520   27.9 %
>> 07 DeepSjeng v1.6ntb         : 2624   18  22   938    62.9 %   2532   24.7 %
>> 08 Rebel v12.00.01           : 2617   21  22   818    59.9 %   2547   25.2 %
>> 09 Aristarch v4.50           : 2616   19  22   898    62.0 %   2531   25.6 %
>> 10 Patriot v1.2.3cpbk        : 2599   20  21   897    59.6 %   2531   25.6 %
>> 11 SmarThink v0.17a          : 2593   19  19  1019    58.6 %   2532   26.3 %
>> 12 Ktulu v4.2                : 2590   19  20  1012    58.6 %   2529   23.8 %
>> 13 Crafty v17.14DC           : 2584   18  16  1183    58.2 %   2527   33.1 %
>> 14 Ktulu v5.0                : 2582   21  22   874    57.2 %   2532   19.9 %
>> 15 Thinker v4.5b             : 2581   20  17   976    56.9 %   2533   33.6 %
>> 16 Aristarch v4.37           : 2581   20  17   959    56.7 %   2534   36.4 %
>> 17 Aristarch v4.21           : 2580   18  17  1177    57.2 %   2529   25.2 %
>> 18 Crafty v19.06DCntb        : 2578   19  17  1059    56.6 %   2532   30.8 %
>> 19 Delfi v4.3                : 2564   19  17  1140    55.2 %   2528   24.8 %
>> 20 El Chinito v3.25          : 2563   20  18   980    54.7 %   2530   27.3 %
>> 21 Delfi v4.2                : 2561   25  25   580    58.1 %   2504   27.2 %
>> 22 Crafty-MPC v18.15DC       : 2560   19  17  1121    54.6 %   2528   27.8 %
>> 23 Little Goliath 2000 v3.9  : 2559   18  16  1220    54.5 %   2528   26.5 %
>> 24 SmarThink v0.16b++        : 2558   21  21   836    58.0 %   2503   24.3 %
>> 25 SoS 3                     : 2557   18  17  1219    54.2 %   2528   23.3 %
>> 26 Pepito v1.59 profile      : 2556   19  16  1220    54.1 %   2528   24.3 %
>> 27 Crafty v18.15DC           : 2556   22  22   741    59.0 %   2493   29.0 %
>> 28 Fruit v1.5                : 2556   24  20   739    52.6 %   2538   25.8 %
>> 29 Yace Paderborn            : 2553   19  16  1200    53.9 %   2526   26.6 %
>> 30 Aristarch v4.4            : 2547   36  34   319    54.1 %   2518   20.4 %
>> 31 SoS 4                     : 2547   21  18   999    52.3 %   2531   23.7 %
>> 32 Yace v0.99.56             : 2541   33  29   380    54.3 %   2511   26.6 %
>> 33 Little Goliath 2000 v3.5  : 2537   31  25   440    53.6 %   2511   30.9 %
>> 34 Green Light Chess v3.00   : 2536   20  15  1220    51.1 %   2528   26.0 %
>> 35 Tao v5.6                  : 2525   18  22   935    48.7 %   2534   24.4 %
>> 36 Anmon v5.30               : 2524   17  22   979    49.0 %   2531   26.5 %
>> 37 Amyan v1.59               : 2510   17  21  1030    48.6 %   2520   26.2 %
>> 38 Pharaon v2.62             : 2509   17  20  1139    48.3 %   2521   23.7 %
>> 39 Crafty v19.01DC           : 2499   24  19   815    50.4 %   2496   25.5 %
>> 40 LambChop v10.99           : 2495   17  18  1217    45.2 %   2529   22.5 %
>> 41 Gromit v3.8.2             : 2491   17  18  1192    44.6 %   2529   23.9 %
>> 42 Ktulu v3.9                : 2490   19  24   779    48.6 %   2500   26.1 %
>> 43 SlowChess v2.89b          : 2488   18  20  1013    45.1 %   2522   24.3 %
>> 44 KnightDreamer v3.2        : 2484   18  19  1120    43.5 %   2529   24.0 %
>> 45 Anmon v5.22               : 2483   19  22   899    46.4 %   2508   26.7 %
>> 46 Comet B44-2               : 2481   18  18  1140    43.0 %   2529   25.4 %
>> 47 SoS v11-99                : 2476   33  34   359    46.0 %   2505   17.3 %
>> 48 Tao v5.4                  : 2475   19  19  1039    44.1 %   2516   21.0 %
>> 49 Amy v0.8.3                : 2472   19  17  1193    42.0 %   2528   19.4 %
>> 50 Dragon v4.4.3             : 2466   19  18  1081    41.2 %   2528   25.7 %
>> 51 Comet B62-3               : 2461   19  18  1119    40.3 %   2529   25.3 %
>> 52 Francesca M.0.0.9         : 2444   19  16  1218    37.9 %   2530   24.4 %
>> 53 PostModernist v1.007      : 2441   20  17  1138    37.5 %   2530   23.9 %
>> 54 Comet B60                 : 2434   22  21   780    41.2 %   2496   25.6 %
>> 55 Leila v0.53h              : 2425   22  17  1098    36.0 %   2525   20.0 %
>> 56 Tcb v0045                 : 2421   21  16  1139    34.7 %   2531   23.9 %
>> 57 Resp v0.19                : 2402   22  16  1120    32.4 %   2530   22.9 %
>> 58 Arasan v7.4               : 2398   27  18   808    31.9 %   2529   23.0 %
>> 59 Nejmet v3.07              : 2388   25  18   876    33.2 %   2510   22.3 %
>> 60 SlowChess v2.78           : 2378   27  19   790    33.5 %   2497   19.6 %
>> 61 Exchess v4.03             : 2324   29  14  1119    23.3 %   2532   20.6 %
>> 62 Beowulf v2.2              : 2306   32  14  1120    22.2 %   2524   17.2 %
>>Games        :  29929 (finished)
>>White Wins   :  12241 (40.9 %)
>>Black Wins   :  10167 (34.0 %)
>>Draws        :   7521 (25.1 %)
>>Unfinished   :      0
>>White Perf.  : 53.5 %
>>Black Perf.  : 46.5 %
>>ECO A =   4553 Games (15.2 %)
>>ECO B =   8937 Games (29.9 %)
>>ECO C =   5730 Games (19.1 %)
>>ECO D =   5359 Games (17.9 %)
>>ECO E =   5350 Games (17.9 %)
>>(28) Fruit v1.5                : 739 (+293,=191,-255), 52.6 %
>>Ruffian v2.1.0                :  20 (+  2,=  7,- 11), 27.5 %
>>Chess Tiger 15.0 normal       :  20 (+  5,=  3,- 12), 32.5 %
>>Rebel v12.00.01               :  20 (+  4,=  5,- 11), 32.5 %
>>List v5.12                    :  20 (+  4,=  5,- 11), 32.5 %
>>SoS 3                         :  20 (+  6,=  3,- 11), 37.5 %
>>SoS 4                         :  20 (+  7,=  3,- 10), 42.5 %
>>DeepSjeng v1.6ntb             :  20 (+  7,=  7,-  6), 52.5 %
>>Aristarch v4.50               :  20 (+  5,=  4,- 11), 35.0 %
>>Patriot v1.2.3cpbk            :  20 (+  7,=  6,-  7), 50.0 %
>>Ktulu v4.2                    :  20 (+  8,=  1,- 11), 42.5 %
>>Thinker v4.5b                 :  20 (+  6,=  9,-  5), 52.5 %
>>SmarThink v0.17a              :  20 (+  7,=  8,-  5), 55.0 %
>>Crafty v17.14DC               :  20 (+  7,=  6,-  7), 50.0 %
>>Ktulu v5.0                    :  19 (+  2,=  6,- 11), 26.3 %
>>Crafty v19.06DCntb            :  20 (+  6,=  8,-  6), 50.0 %
>>Delfi v4.3                    :  20 (+  9,=  4,-  7), 55.0 %
>>El Chinito v3.25              :  20 (+  6,=  8,-  6), 50.0 %
>>Crafty-MPC v18.15DC           :  20 (+  5,=  6,-  9), 40.0 %
>>Little Goliath 2000 v3.9      :  20 (+  7,=  2,- 11), 40.0 %
>>Pepito v1.59 profile          :  20 (+  6,=  2,- 12), 35.0 %
>>Yace Paderborn                :  20 (+  6,=  7,-  7), 47.5 %
>>Green Light Chess v3.00       :  20 (+  5,=  9,-  6), 47.5 %
>>Tao v5.6                      :  20 (+ 13,=  3,-  4), 72.5 %
>>Anmon v5.30                   :  20 (+  7

[Robert Allgeuer]

Geschrieben von:/Posted by: Robert Allgeuer at 01 July 2004 13:28:39:
Als Antwort auf:/In reply to: Re: YABRL: Fruit 1.5, the "bombshell" in action ... geschrieben von:/posted by: Heinz van Kempen at 01 July 2004 13:04:46:

Thanks Heinz,
Wildcat 4 is one of the next, that is for sure. I wanted to test Gothmog for a long time already, but for some reasons it did not run on my system (the feature done=0 related problem). I have not got my hands on any of these betas, but if Gothmog runs on my machine, I will test it of course.
Is Yace 0.99.87 something like an unofficial version?
Robert

Hi Robert,
it is a pity that you can´t run Gothmog. It plays spectacular chess. Tord is working on Gothmog 1.0 and has further plans for something completely new and even stronger. Maybe those will run on your machine.
For Yace indeed I read somewhere that the version is unofficial and it is not an Dieter´s homepage. Somewhere I found a link where it could be downloaded. Someone posted a disappointed comment after only a few games, but here it plays very well with first three Nunn positions. I will upload this new tournament in about two hours. Also involved Delfi 4.5 and Little Goliath Revival will also be interesting and it will need some time to see if they are improved or not.
Best Regards
Heinz

yes many new interesting engine versions out at the moment, more than what can be tested in fact.


YABRL (Yet Another Blitz Rating List)

[Fabien Letouzey]

Geschrieben von:/Posted by: Fabien Letouzey at 01 July 2004 18:03:39:
Als Antwort auf:/In reply to: Re: YABRL: Fruit 1.5, the geschrieben von:/posted by: Uri Blass at 01 July 2004 13:22:16:

I am also not surprised.
After reading your results and after knowing that Fruit still has no king safety and no special search tricks and after knowing that Fabian plans not only to implement them but he is also busy in adding endgame knowledge to fruit I will be surprised if Fruit is not going to be stronger than Shredder8 in some years(I predict that it will be stronger than shredder8 in 2007 or earlier if you do not consider the opening book and do not consider parallel search).
Uri

Hi Uri,
In order to avoid science-fiction speculation, here is my "plan":
1) Make the evaluation "complete" (= contain all major components). This implies adding some form of king safety and passed-pawn knowledge.
2) Try classical ideas in move generation/ordering: incremental move generation and SEE-sorted captures.
3) Try to improve the quiescence search by adding more "potentially-good" moves, like more checks and some passer moves.
I consider none of these points as "tricks", and I certainly have no plans to try any. Forward pruning in the main search is a separate topic that I don't intend to address now. If I ever do, I expect it will require years of work (same for a proper move ordering).
I would be very happy with just having a "clean" engine that uses only well-known techniques, which I would then apply to games other than chess. I am confident however that the level of Shredder cannot be reached by such a naive approach.
It would be interesting to know which is the strongest engine using only documented forward-pruning schemes. Unless it is List, the distance to the top engines seems rather large though.
Fabien.

[Tord Romstad]

Geschrieben von:/Posted by: Tord Romstad at 01 July 2004 19:18:06:
Als Antwort auf:/In reply to: Re: YABRL: Fruit 1.5 geschrieben von:/posted by: Fabien Letouzey at 01 July 2004 18:03:39:

It would be interesting to know which is the strongest engine using only documented forward-pruning schemes. Unless it is List, the distance to the top engines seems rather large though.

The King, perhaps? Or Ferret?
Tord

[Tord Romstad]

Geschrieben von:/Posted by: Tord Romstad at 01 July 2004 19:19:54:
Als Antwort auf:/In reply to: Re: YABRL: Fruit 1.5, the "bombshell" in action ... geschrieben von:/posted by: Robert Allgeuer at 01 July 2004 12:49:35:

Thanks Heinz,
Wildcat 4 is one of the next, that is for sure. I wanted to test Gothmog for a long time already, but for some reasons it did not run on my system (the feature done=0 related problem). I have not got my hands on any of these betas, but if Gothmog runs on my machine, I will test it of course.

Don't bother to download Gothmog 1.0 beta 7. It almost certainly behaves
exactly like 0.4.8 in this respect. I'll try to address the problem in the
next version.
Tord

[Robert Allgeuer]

Geschrieben von:/Posted by: Robert Allgeuer at 01 July 2004 20:05:57:
Als Antwort auf:/In reply to: Re: YABRL: Fruit 1.5, the "bombshell" in action ... geschrieben von:/posted by: Tord Romstad at 01 July 2004 19:19:54:

Thanks Heinz,
Wildcat 4 is one of the next, that is for sure. I wanted to test Gothmog for a long time already, but for some reasons it did not run on my system (the feature done=0 related problem). I have not got my hands on any of these betas, but if Gothmog runs on my machine, I will test it of course.

Don't bother to download Gothmog 1.0 beta 7. It almost certainly behaves
exactly like 0.4.8 in this respect. I'll try to address the problem in the
next version.
Tord

Thanks
Robert

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 01 July 2004 20:44:53:
Als Antwort auf:/In reply to: Re: YABRL: Fruit 1.5 geschrieben von:/posted by: Fabien Letouzey at 01 July 2004 18:03:39:

I am also not surprised.
After reading your results and after knowing that Fruit still has no king safety and no special search tricks and after knowing that Fabian plans not only to implement them but he is also busy in adding endgame knowledge to fruit I will be surprised if Fruit is not going to be stronger than Shredder8 in some years(I predict that it will be stronger than shredder8 in 2007 or earlier if you do not consider the opening book and do not consider parallel search).
Uri

Hi Uri,
In order to avoid science-fiction speculation, here is my "plan":
1) Make the evaluation "complete" (= contain all major components). This implies adding some form of king safety and passed-pawn knowledge.
2) Try classical ideas in move generation/ordering: incremental move generation and SEE-sorted captures.
3) Try to improve the quiescence search by adding more "potentially-good" moves, like more checks and some passer moves.
I consider none of these points as "tricks", and I certainly have no plans to try any. Forward pruning in the main search is a separate topic that I don't intend to address now. If I ever do, I expect it will require years of work (same for a proper move ordering).
I would be very happy with just having a "clean" engine that uses only well-known techniques, which I would then apply to games other than chess.

considering the fact that fruit is already not weaker than old commercial programs like Genius3 and Mchess8 it seems that it already has not a bad evaluation.
I suspect that when you are going to have your evaluation complete then your evaluation may be better than Shredder8's evaluation.
I think that forward pruning in the main search is something that is possible to generalize to games that are not chess so I am surprised to read that you consider it as something that will require years.
You only need some good if conditions.
I do not use IID or efficient hash tables and I also believe that I do not have good order of moves but I believe that I compensate for it by some good if's so movei's search in the middle game is probably at similiar level to fruit search.
I also believe that my forward pruning can be improved significantly.

Forward pruning can be used also to games that are different than chess.
I think that the idea of history based pruning is not an idea that is only for chess.
history based pruning is probably more important for games that have more possibilities than chess.
I think also that there are a lot of techniques that are not known that it is possible to use not only for chess.
Uri

[Robert Allgeuer]

Geschrieben von:/Posted by: Robert Allgeuer at 01 July 2004 23:41:54:
Als Antwort auf:/In reply to: Re: YABRL: Fruit 1.5 geschrieben von:/posted by: Fabien Letouzey at 01 July 2004 18:03:39:

,

In order to avoid science-fiction speculation, here is my "plan":
1) Make the evaluation "complete" (= contain all major components). This implies adding some form of king safety and passed-pawn knowledge.
2) Try classical ideas in move generation/ordering: incremental move generation and SEE-sorted captures.
3) Try to improve the quiescence search by adding more "potentially-good" moves, like more checks and some passer moves.
I consider none of these points as "tricks", and I certainly have no plans to try any. Forward pruning in the main search is a separate topic that I don't intend to address now. If I ever do, I expect it will require years of work (same for a proper move ordering).
I would be very happy with just having a "clean" engine that uses only well-known techniques, which I would then apply to games other than chess. I am confident however that the level of Shredder cannot be reached by such a naive approach.
It would be interesting to know which is the strongest engine using only documented forward-pruning schemes. Unless it is List, the distance to the top engines seems rather large though.
Fabien.

This is a very interesting "experiment" to watch; essentially how far can an engine get using only publically available, "open" techniques. Already looking forward to it ....
Robert

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 01 July 2004 23:50:11:
Als Antwort auf:/In reply to: Re: YABRL: Fruit 1.5 geschrieben von:/posted by: Robert Allgeuer at 01 July 2004 23:41:54:

,

In order to avoid science-fiction speculation, here is my "plan":
1) Make the evaluation "complete" (= contain all major components). This implies adding some form of king safety and passed-pawn knowledge.
2) Try classical ideas in move generation/ordering: incremental move generation and SEE-sorted captures.
3) Try to improve the quiescence search by adding more "potentially-good" moves, like more checks and some passer moves.
I consider none of these points as "tricks", and I certainly have no plans to try any. Forward pruning in the main search is a separate topic that I don't intend to address now. If I ever do, I expect it will require years of work (same for a proper move ordering).
I would be very happy with just having a "clean" engine that uses only well-known techniques, which I would then apply to games other than chess. I am confident however that the level of Shredder cannot be reached by such a naive approach.

This is a very interesting "experiment" to watch; essentially how far can an engine get using only publically available, "open" techniques. Already looking forward to it ....
Robert

The point is that the available open techniques get better so there is no limit.
I do not use part of the publically avaiable open techniques and I consider understanding and using correctly publically available "open" techniques as harder task than inventing new productive techniques.
Uri