New engine releases 2024 ... 6+3

[Frank Quisinsky]

Hi there,

Code: Select all

Conditions:
Operating system: Windows 11
GUI:              Shredder 12 by Stefan Meyer-Kahlen
Processor:        Intel® Core™ i7-1185g7 = 4 cores
Ghz, RAM:         3.0Ghz overclocked, DDR4 3200Mhz
Elo:              Wasp 6.50 NN is playing with ~3300 Elo (reference)
Tournament type:  Everyone against everyone, 50 games per match
Time control:     6 Minutes/Game + 3 Seconds/Move
Endgame 5-man:    syzygy
Opening book:     feobos-6m-v2.1.bkt for Shredder GUI, based on 3.698 lines
Cores:            1 core, Hyperthreading = off, Large-Pages = off
Hashtables:       128Mb
Contempt:         Default engine settings changed to contempt = 0 (if possible)
Other settings:   resign = off, ponder = off, learning = off
Replay games:     Every draw less than 20 moves

[Frank Quisinsky]

Releases 2024:
01. Akimbo 0.8.0 NN
02. Starzix 3.0 NN
03. Lizard 10.0 NN
04. Clarity 4.1.0 NN
05. Caissa 1.16 NN ... still running

Code: Select all

         Name                 Games     Win     Draw     Lose       Pts         S-B         %    wins-55m  lost-55m  AV-wins  AV-draws

01. Akimbo 0.8.0 NN        :   250  :   60+  :  169=  :   21-  :   144.5  :  17369.25  :  57.8%      2         2       90        91
02. Starzix 3.0 NN         :   250  :   44+  :  181=  :   25-  :   134.5  :  16364.00  :  53.8%      1         4       93        92
03. Wasp 6.63 NN dev       :   250  :   49+  :  171=  :   30-  :   134.5  :  16342.25  :  53.8%      6         1       80        84
04. Wasp 6.50 NN           :   250  :   36+  :  177=  :   37-  :   124.5  :  15361.50  :  49.8%      5         0       79        79
05. Lizard 10.0 NN         :   250  :   20+  :  176=  :   54-  :   108.0  :  13708.25  :  43.2%      3         6       92        99
06. Clarity 4.1.0 NN       :   250  :   17+  :  174=  :   59-  :   104.0  :  13313.75  :  41.6%      0         4       87        86

Code: Select all

   # Player              :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Akimbo 0.8.0 NN     :  3347.99    250    57.8   60   169    21   144.5   67.6  19.86  3291.80  19.34    5.0
   2 Wasp 6.63 NN dev    :  3323.89    250    53.8   49   171    30   134.5   68.4  19.64  3296.62  19.39    5.0
   3 Starzix 3.0 NN      :  3323.89    250    53.8   44   181    25   134.5   72.4  18.91  3296.62  19.54    5.0
   4 Wasp 6.50 NN        :  3300.00    250    49.8   36   177    37   124.5   70.8  19.27  3301.40  19.46    5.0
   5 Lizard 10.0 NN      :  3260.45    250    43.2   20   176    54   108.0   70.4  18.99  3309.30  19.52    5.0
   6 Clarity 4.1.0 NN    :  3250.75    250    41.6   17   174    59   104.0   69.6  19.92  3311.24  19.33    5.0

White advantage = 37.46 +/- 6.71
Draw rate (equal opponents) = 74.22 % +/- 1.73

[Frank Quisinsky]

Hi there,

an idea on my Dell Ultrabook (only 18w power when running 4 cores at 4x 3.0Ghz).
If I have more to do, I can use my Intel i9-10900k with 10x 4.4Ghz with 4+2 (same in green, compare to i7 with 6+3).
In this case I tested without AVX-512 because the i9-10900 doesn't support it.

Can't test all engines, only stronger ones around 3150 Elo if Wasp 6.50 NN = 3300 Elo.


Wasp 6.63 NN dev is playing.
+30 but for me only to see in blitz.
Not important if 2x Wasp is in the field.

Best
Frank

[Frank Quisinsky]

Hi there,

two people asked for a download file ...
I added a link in the replay-zone from my 66+6 tournament.
Should be enough!

Replay-Zone:
https://www.amateurschach.de/fling/index.html

At the moment Counter 5.5 NN und Caissa 1.16 NN are still running.

Later ...

Best
Frank

[Frank Quisinsky]

05. Caissa 1.16 NN
06. Counter 5.5 NN

Code: Select all

         Name                 Games     Win     Draw     Lose       Pts         S-B         %    wins-55m  lost-55m  AV-wins  AV-draws

01. Caissa 1.16 NN         :   350  :  193+  :  157=  :    0-  :   271.5  :  43186.50  :  77.57%    10         0       86       83
02. Akimbo 0.8.0 NN        :   350  :   79+  :  226=  :   45-  :   192.0  :  31175.75  :  54.86%     3         6       92       91
03. Starzix 3.0 NN         :   350  :   69+  :  236=  :   45-  :   187.0  :  30296.25  :  53.43%     1         4       94       93
04. Wasp 6.63 NN dev       :   350  :   68+  :  226=  :   56-  :   181.0  :  29461.50  :  51.71%     8         1       80       85
05. Wasp 6.50 NN           :   350  :   59+  :  223=  :   68-  :   170.5  :  27671.00  :  48.71%    11         0       79       79
06. Lizard 10.0 NN         :   350  :   33+  :  230=  :   87-  :   148.0  :  24769.00  :  42.29%     3         8       94       99
07. Clarity 4.1.0 NN       :   350  :   31+  :  222=  :   97-  :   142.0  :  23763.50  :  40.57%     0         7       90       86
08. Counter 5.5 NN         :   350  :   19+  :  178=  :  153-  :   108.0  :  18568.00  :  30.86%     0        10       94       99

White Wins =    360 ( 25.71% )
Draws      =    849 ( 60.64% )
Black Wins =    191 ( 13.64% )
Average    = 176.95 ( 88,47 moves )

Code: Select all

   # Player              :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Caissa 1.16 NN      :  3507.34    350    77.6  193   157     0   271.5   44.9  24.72  3281.76  18.44    7.0
   2 Akimbo 0.8.0 NN     :  3340.71    350    54.9   79   226    45   192.0   64.6  18.09  3305.57  19.39    7.0
   3 Starzix 3.0 NN      :  3331.21    350    53.4   69   236    45   187.0   67.4  17.90  3306.93  19.42    7.0
   4 Wasp 6.63 NN dev    :  3319.84    350    51.7   68   226    56   181.0   64.6  18.42  3308.55  19.34    7.0
   5 Wasp 6.50 NN        :  3300.00    350    48.7   59   223    68   170.5   63.7  18.33  3311.38  19.36    7.0
   6 Lizard 10.0 NN      :  3257.27    350    42.3   33   230    87   148.0   65.7  18.24  3317.49  19.37    7.0
   7 Clarity 4.1.0 NN    :  3245.73    350    40.6   31   222    97   142.0   63.4  17.82  3319.14  19.43    7.0
   8 Counter 5.5 NN      :  3177.59    350    30.9   19   178   153   108.0   50.9  20.30  3328.87  19.07    7.0

White advantage = 48.23 +/- 5.47
Draw rate (equal opponents) = 74.77 % +/- 1.54

Best
Frank

[Frank Quisinsky]

I am not sure ...

But I think the idea is bad, means to add the stronger new releases in a tournament.
Now we can see that Caissa is on rank 1, but what should we do with the information without any results to compare Caissa with other strong engines? Ok, with time other stronger engines will be available.

But all in all ... typical Frank, not every idea is good.

Let us look in Pawn 3.0 NN now.
With time all will be wonderful.

Later ...

Best
Frank

[Frank Quisinsky]

07. Pawn 3.0 NN
08. Texel 1.11 NN ... still running on i9-10900k, 4.4Ghz with 4+2
09. RubiChess 20240112 NN ... soon
10. Avalanche 2.1.0 NN ... still running on i7-1185g7, 3.0Ghz with 6+3

Code: Select all

         Name                 Games     Win     Draw     Lose       Pts         S-B         %    wins-55m  lost-55m  AV-wins  AV-draws  AV-all

01. Caissa 1.16 NN         :   400  :  215+  :  183=  :    2-  :   306.5  :  56458.75  :  76.63%    12        0        86        84       85
02. Pawn 3.0 NN            :   400  :   86+  :  265=  :   49-  :   218.5  :  41014.00  :  54.63%     3        4        83        89       88
03. Akimbo 0.8.0 NN        :   400  :   84+  :  261=  :   55-  :   214.5  :  40209.75  :  53.63%     3        6        93        92       90
04. Starzix 3.0 NN         :   400  :   76+  :  272=  :   52-  :   212.0  :  39688.50  :  53.00%     1        5        94        92       91
05. Wasp 6.63 NN dev       :   400  :   75+  :  260=  :   65-  :   205.0  :  38536.50  :  51.25%     8        1        81        83       84
06. Wasp 6.50 NN           :   400  :   64+  :  260=  :   76-  :   194.0  :  36391.25  :  48.50%    12        0        80        80       83
07. Lizard 10.0 NN         :   400  :   34+  :  264=  :  102-  :   166.0  :  31960.25  :  41.50%     3        9        94       100       95
08. Clarity 4.1.0 NN       :   400  :   32+  :  253=  :  115-  :   158.5  :  30515.50  :  39.63%     0        8        90        84       85
09. Counter 5.5 NN         :   400  :   20+  :  210=  :  170-  :   125.0  :  24622.50  :  31.25%     1       10        92       100       95


White Wins =    447 ( 24.83% )
Draws      =  1.114 ( 61.89% )
Black Wins =    239 ( 13.28% )
Average    = 176.59 ( 88,29 moves )
 

Code: Select all

   # Player              :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Caissa 1.16 NN      :  3502.60    400    76.6  215   183     2   306.5   45.8  22.63  3287.17  17.53    8.0
   2 Pawn 3.0 NN         :  3340.88    400    54.6   86   265    49   218.5   66.3  17.93  3307.39  18.12    8.0
   3 Akimbo 0.8.0 NN     :  3334.18    400    53.6   84   261    55   214.5   65.3  17.37  3308.23  18.19    8.0
   4 Starzix 3.0 NN      :  3330.00    400    53.0   76   272    52   212.0   68.0  16.96  3308.75  18.24    8.0
   5 Wasp 6.63 NN dev    :  3318.32    400    51.3   75   260    65   205.0   65.0  17.12  3310.21  18.22    8.0
   6 Wasp 6.50 NN        :  3300.00    400    48.5   64   260    76   194.0   65.0  16.82  3312.50  18.26    8.0
   7 Lizard 10.0 NN      :  3253.03    400    41.5   34   264   102   166.0   66.0  16.94  3318.37  18.24    8.0
   8 Clarity 4.1.0 NN    :  3240.23    400    39.6   32   253   115   158.5   63.3  17.01  3319.97  18.23    8.0
   9 Counter 5.5 NN      :  3180.74    400    31.3   20   210   170   125.0   52.5  20.10  3327.40  17.85    8.0

White advantage = 45.43 +/- 4.89
Draw rate (equal opponents) = 74.58 % +/- 1.37

A lot to do here, like that!
Thanks to the programmers for all the new engines!!

Best
Frank

[Frank Quisinsky]

08. Texel 1.11 NN
09. RubiChess 20240112 NN ... still running
10. Avalanche 2.1.0 NN ... still running

Code: Select all

         Name                 Games     Win     Draw     Lose       Pts         S-B         %    wins-55m  lost-55m  AV-wins  AV-draws  AV-all

01. Caissa 1.16 NN         :   450  :  239+  :  209=  :    2-  :   343.5  :  71876.00  :  76.33%    12         0        86       84        85
02. Pawn 3.0 NN            :   450  :  100+  :  297=  :   53-  :   248.5  :  52844.00  :  55.22%     3         5        83       90        88
03. Akimbo 0.8.0 NN        :   450  :   89+  :  303=  :   58-  :   240.5  :  51095.75  :  53.44%     3         8        93       93        91
04. Starzix 3.0 NN         :   450  :   79+  :  312=  :   59-  :   235.0  :  49877.50  :  52.22%     1         5        94       94        92
05. Texel 1.11 NN          :   450  :   82+  :  301=  :   67-  :   232.5  :  49132.50  :  51.67%     9         1        76       90        87
06. Wasp 6.63 NN dev       :   450  :   80+  :  295=  :   75-  :   227.5  :  48465.00  :  50.56%     9         1        82       82        84
07. Wasp 6.50 NN           :   450  :   72+  :  297=  :   81-  :   220.5  :  46856.25  :  49.00%    12         0        81       80        83
08. Lizard 10.0 NN         :   450  :   37+  :  297=  :  116-  :   185.5  :  40362.75  :  41.22%     3        11        94      100        94
09. Clarity 4.1.0 NN       :   450  :   34+  :  288=  :  128-  :   178.0  :  38701.00  :  39.56%     0        10        90       84        84
10. Counter 5.5 NN         :   450  :   23+  :  231=  :  196-  :   138.5  :  30770.25  :  30.78%     1        12        89       98        93


White Wins =    547 ( 24.31% )
Draws      =  1.415 ( 62.89% )
Black Wins =    288 ( 12.80% )
Average    = 175.88 ( 87,94 moves )

Code: Select all

   # Player              :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Caissa 1.16 NN      :  3498.74    450    76.3  239   209     2   343.5   46.4  20.60  3286.41  16.46    9.0
   2 Pawn 3.0 NN         :  3341.90    450    55.2  100   297    53   248.5   66.0  15.49  3303.83  17.03    9.0
   3 Akimbo 0.8.0 NN     :  3329.88    450    53.4   89   303    58   240.5   67.3  16.23  3305.17  16.94    9.0
   4 Starzix 3.0 NN      :  3321.64    450    52.2   79   312    59   235.0   69.3  16.46  3306.08  16.92    9.0
   5 Texel 1.11 NN       :  3317.91    450    51.7   82   301    67   232.5   66.9  16.42  3306.50  16.92    9.0
   6 Wasp 6.63 NN dev    :  3310.44    450    50.6   80   295    75   227.5   65.6  16.31  3307.33  16.94    9.0
   7 Wasp 6.50 NN        :  3300.00    450    49.0   72   297    81   220.5   66.0  15.93  3308.49  16.98    9.0
   8 Lizard 10.0 NN      :  3247.37    450    41.2   37   297   116   185.5   66.0  16.78  3314.34  16.88    9.0
   9 Clarity 4.1.0 NN    :  3235.87    450    39.6   34   288   128   178.0   64.0  16.42  3315.61  16.92    9.0
  10 Counter 5.5 NN      :  3172.66    450    30.8   23   231   196   138.5   51.3  18.09  3322.64  16.74    9.0

White advantage = 44.82 +/- 4.29
Draw rate (equal opponents) = 74.84 % +/- 1.18

76 move average for wins is just fantastic, but all this isn't new.
Now the move-average for draws looks much better for Texel.

May Velvet 6.00 can beat Texel's move-average wins.
But I only test new releases here.

Best
Frank

PS: Akimbo 0.8.0 NN is just great.
Just a very aggressive style in the late mid-game and move-average wins seems to be clearly improved ... very, very nice!
And I forgot to Texel ... it seems that with Pawn 3.0 NN ... Texel 1.11 NN has a "Fearful opponent".

[Frank Quisinsky]

09. RubiChess 20240112 NN
10. Avalanche 2.1.0 NN ... still running
11. Lizard 10.1 NN ... still running

Code: Select all

              Name                 Games     Win     Draw     Lose       Pts         S-B         %    wins-55m  lost-55m  AV-wins  AV-draws  AV-all

01. RubiChess 20240112 NN       :   500  :  254+  :  243=  :    3-  :   375.5  :  86504.25  :  75.1%     26         0       81        75       78
02. Caissa 1.16 NN              :   500  :  242+  :  255=  :    3-  :   369.5  :  85344.00  :  73.9%     12         0       86        83       85
03. Pawn 3.0 NN                 :   500  :  100+  :  325=  :   75-  :   262.5  :  60922.25  :  52.5%      3         7       83        89       87
04. Akimbo 0.8.0 NN             :   500  :   89+  :  327=  :   84-  :   252.5  :  58359.25  :  50.5%      3        11       93        92       90
05. Starzix 3.0 NN              :   500  :   79+  :  337=  :   84-  :   247.5  :  57262.25  :  49.5%      1        10       94        93       90
06. Texel 1.11 NN               :   500  :   82+  :  325=  :   93-  :   244.5  :  56279.50  :  48.9%      9         1       76        89       86
07. Wasp 6.63 NN dev            :   500  :   80+  :  321=  :   99-  :   240.5  :  55908.50  :  48.1%      9         2       82        81       83
08. Wasp 6.50 NN                :   500  :   72+  :  316=  :  112-  :   230.0  :  52980.25  :  46.0%     12         0       81        80       83
09. Lizard 10.0 NN              :   500  :   37+  :  313=  :  150-  :   193.5  :  45565.50  :  38.7%      3        15       94        99       93
10. Clarity 4.1.0 NN            :   500  :   34+  :  310=  :  156-  :   189.0  :  44859.50  :  37.8%      0        14       90        82       83
11. Counter 5.5 NN              :   500  :   23+  :  244=  :  233-  :   145.0  :  34856.75  :  29.0%      1        19       89        97       92


White Wins =    718 ( 26.11% )
Draws      =  1.658 ( 60.29% )
Black Wins =    374 ( 13.60% )
Average    = 172.18 ( 86,09 moves )

Code: Select all

   # Player                   :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 RubiChess 20240112 NN    :  3516.36    500    75.1  254   243     3   375.5   48.6  19.97  3310.72  16.26   10.0
   2 Caissa 1.16 NN           :  3506.00    500    73.9  242   255     3   369.5   51.0  19.35  3311.75  16.32   10.0
   3 Pawn 3.0 NN              :  3345.26    500    52.5  100   325    75   262.5   65.0  15.86  3327.83  16.67   10.0
   4 Akimbo 0.8.0 NN          :  3331.33    500    50.5   89   327    84   252.5   65.4  15.45  3329.22  16.71   10.0
   5 Starzix 3.0 NN           :  3324.37    500    49.5   79   337    84   247.5   67.4  15.91  3329.92  16.67   10.0
   6 Texel 1.11 NN            :  3320.20    500    48.9   82   325    93   244.5   65.0  15.38  3330.33  16.72   10.0
   7 Wasp 6.63 NN dev         :  3314.63    500    48.1   80   321    99   240.5   64.2  15.25  3330.89  16.73   10.0
   8 Wasp 6.50 NN             :  3300.00    500    46.0   72   316   112   230.0   63.2  15.64  3332.35  16.69   10.0
   9 Lizard 10.0 NN           :  3248.37    500    38.7   37   313   150   193.5   62.6  15.80  3337.52  16.68   10.0
  10 Clarity 4.1.0 NN         :  3241.86    500    37.8   34   310   156   189.0   62.0  15.82  3338.17  16.68   10.0
  11 Counter 5.5 NN           :  3175.16    500    29.0   23   244   233   145.0   48.8  18.14  3344.84  16.44   10.0

White advantage = 50.82 +/- 3.90
Draw rate (equal opponents) = 76.63 % +/- 1.15

Rubi on rank 1, but lost the match vs. Caissa.
Move-average stats and average won-games are more as strong only!

Best
Frank

[Frank Quisinsky]

10. Avalanche 2.1.0 NN ... still running on Wasp-2
11. Lizard 10.1 NN ... ready
12. Minic 3.40 NN ... still running on Wasp-1

Code: Select all

              Name                 Games     Win     Draw     Lose       Pts         S-B         %    wins-55m  lost-55m  AV-wins  AV-draws  AV-all

01. RubiChess 20240112 NN       :   500  :  245+  :  251=  :    4-  :   370.5  :  86861.50  :  74.10%    23         0       81        75       78
02. Caissa 1.16 NN              :   500  :  236+  :  261=  :    3-  :   366.5  :  85950.00  :  73.30%    12         0       87        82       84
03. Lizard 10.1 NN              :   550  :   91+  :  354=  :  105-  :   268.0  :  55540.50  :  48.73%     5        10       89        95       91
04. Pawn 3.0 NN                 :   500  :   96+  :  325=  :   79-  :   258.5  :  61169.50  :  51.70%     2         7       83        87       86
05. Akimbo 0.8.0 NN             :   500  :   89+  :  324=  :   87-  :   251.0  :  59124.50  :  50.20%     4        10       95        92       90
06. Starzix 3.0 NN              :   500  :   79+  :  337=  :   84-  :   247.5  :  58286.75  :  49.50%     1         9       95        91       89
07. Texel 1.11 NN               :   500  :   72+  :  327=  :  101-  :   235.5  :  55278.50  :  47.10%     8         1       78        90       88
08. Wasp 6.63 NN dev            :   500  :   69+  :  331=  :  100-  :   234.5  :  55889.75  :  46.90%     8         2       83        80       82
09. Wasp 6.50 NN                :   500  :   72+  :  316=  :  112-  :   230.0  :  53872.00  :  46.00%    12         0       81        80       83
10. Clarity 4.1.0 NN            :   500  :   29+  :  304=  :  167-  :   181.0  :  43865.50  :  36.20%     0        13       90        82       83
11. Counter 5.5 NN              :   500  :   19+  :  239=  :  242-  :   138.5  :  34146.50  :  27.70%     1        20       88        98       92
12. Avalanche 2.1.0 NN          :    50  :    4+  :   29=  :   17-  :    18.5  :   4958.00  :  37.00%     0         4       78        95       87


White Wins =    720 ( 25.71% )
Draws      =  1.699 ( 60.68% )
Black Wins =    381 ( 13.61% )
Average    = 171.82 ( 85,91 moves )

Code: Select all

   # Player                   :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 RubiChess 20240112 NN    :  3506.47    500    74.1  245   251     4   370.5   50.2  20.11  3311.10  17.20   10.0
   2 Caissa 1.16 NN           :  3499.71    500    73.3  236   261     3   366.5   52.2  20.60  3311.78  17.15   10.0
   3 Pawn 3.0 NN              :  3339.48    500    51.7   96   325    79   258.5   65.0  16.95  3327.80  17.51   10.0
   4 Akimbo 0.8.0 NN          :  3329.09    500    50.2   89   324    87   251.0   64.8  16.38  3328.84  17.57   10.0
   5 Starzix 3.0 NN           :  3324.25    500    49.5   79   337    84   247.5   67.4  16.37  3329.33  17.57   10.0
   6 Lizard 10.1 NN           :  3309.02    550    48.7   91   354   105   268.0   64.4  14.86  3320.22  20.97   11.0
   7 Texel 1.11 NN            :  3307.63    500    47.1   72   327   101   235.5   65.4  16.25  3330.99  17.58   10.0
   8 Wasp 6.63 NN dev         :  3306.25    500    46.9   69   331   100   234.5   66.2  17.08  3331.13  17.50   10.0
   9 Wasp 6.50 NN             :  3300.00    500    46.0   72   316   112   230.0   63.2  16.42  3331.75  17.56   10.0
  10 Clarity 4.1.0 NN         :  3230.47    500    36.2   29   304   167   181.0   60.8  17.64  3338.70  17.44   10.0
  11 Avalanche 2.1.0 NN       :  3213.98     50    37.0    4    29    17    18.5   58.0  53.42  3309.02  14.86    1.0
  12 Counter 5.5 NN           :  3165.14    500    27.7   19   239   242   138.5   47.8  19.41  3345.24  17.27   10.0

White advantage = 48.75 +/- 3.80
Draw rate (equal opponents) = 76.00 % +/- 1.13

10.
Run-10 need a longer time (on Ultrabook, 4 cores with 6+3)

11.
Run-11, Lizard 10.0 NN tested before!
Version 10.1 NN is 61 Elo stronger after the 500 games I have.

I think I have tomorrow the results from Avalanche 2.1.0 NN and Minic 3.40 NN.

Best
Frank

Again, you can download what I do here ...
https://www.amateurschach.de/fling/index.html (point 8 in Replay-Zone from my 66+6 tourney).