Kayra 1.1

[Eduard]

Are these super bullet results with 4 to 10 book moves as the default? Stockfish is maximally trained for this, and consequently any other engine will not be better than Stockfish dev. Me and some other people, we test differently. A few days ago I played 100 games 5 + 0 with Stockfish dev. played on the server and I would rather take Kayra in a tournament. On fast computers at 5 + 0, the engines sometimes achieve up to 60 plies in the middlegame (not the endgame), so some things look different than with the superbullet.

[Guenther]

...

Me and some other people, we test differently. A few days ago I played 100 games 5 + 0 with Stockfish dev. played on the server and I would rather take Kayra in a tournament.

...

Yes, you don't test Kayra, that's the difference. Instead you test your book and your hardware vs. complelety different often unknown entities.
Should be self explaining really, but we live in times of unfounded belief and hysteria.

Also it will be a selffulfilling prophecy to hint to slower time controls any time the results are not what the believers think.
This can be extended beyond our lifetimes...

Instead of all of this, we simply look into the sources and see nothing, but random void and a funny mixture of older and newer SF code.
I would call it monkey number changes.

[Eduard]

Of course I test with my book. That's important to me. I play in engine tournaments and also in FreestyleChess.

This is the practice! You only test in the laboratory.

It is important to me to know how an engine can handle openings that I want to play myself. The ability to analyze is also important. What do I care about a superbullet result!?

[mehmet123]

Instead of all of this, we simply look into the sources and see nothing, but random void and a funny mixture of older and newer SF code.
I would call it monkey number changes.

Can you look at the Kayra 1.1 codes. Did you really not notice any difference, especially in the codes at the search section?

[mehmet123]

Hmm? It obviously plays worse

Code: Select all

Score of stockfish_dev vs kayra_1_1: 1210 - 901 - 1889 [0.539]
...      stockfish_dev playing White: 1189 - 3 - 808  [0.796] 2000
...      stockfish_dev playing Black: 21 - 898 - 1081  [0.281] 2000
...      White vs Black: 2087 - 24 - 1889  [0.758] 4000
Elo difference: 26.9 +/- 7.8, LOS: 100.0 %, DrawRatio: 47.2 %
4000 of 10000 games finished.

I wrote this last week at Outskirt Chess Forum
"In my tests 10 sec+ 0.2 sec and 1 min + 0.5 sec tc Stockfish is a little stronger than Kayra but at 6 cores 1 min+ 0.5 sec tests Kayra is stronger than Stockfish dev in my tests. I had many tests at these condition ( 6 cores 1 min+ 0.5 sec) but only published one test"

I have never claimed that Kayra is stronger than Stockfish, as I know that different test conditions may produce different results. Kayra is a very new engine and I do not have a such a goal for now. I had to publish Kayra 1.0 very early. My original plan was to release Kayra 1.0 around February 2022. But due to private tests, some testers started to receive intense criticism, so I had to publish Kayra much earlier than I planned.

[Sopel]

Hmm? It obviously plays worse

Code: Select all

Score of stockfish_dev vs kayra_1_1: 1210 - 901 - 1889 [0.539]
...      stockfish_dev playing White: 1189 - 3 - 808  [0.796] 2000
...      stockfish_dev playing Black: 21 - 898 - 1081  [0.281] 2000
...      White vs Black: 2087 - 24 - 1889  [0.758] 4000
Elo difference: 26.9 +/- 7.8, LOS: 100.0 %, DrawRatio: 47.2 %
4000 of 10000 games finished.

I wrote this last week at Outskirt Chess Forum
"In my tests 10 sec+ 0.2 sec and 1 min + 0.5 sec tc Stockfish is a little stronger than Kayra but at 6 cores 1 min+ 0.5 sec tests Kayra is stronger than Stockfish dev in my tests. I had many tests at these condition ( 6 cores 1 min+ 0.5 sec) but only published one test"

I have never claimed that Kayra is stronger than Stockfish, as I know that different test conditions may produce different results. Kayra is a very new engine and I do not have a such a goal for now. I had to publish Kayra 1.0 very early. My original plan was to release Kayra 1.0 around February 2022. But due to private tests, some testers started to receive intense criticism, so I had to publish Kayra much earlier than I planned.

what tests are you referring to?

[mehmet123]

what tests are you referring to?

I talk about my tests.

Please Tomasz make a minimum 5 min/ 10 min (1 core ) test and publish your results. Use any opening book you want. Check if there is a significant difference between in your two tests and inform us.

[mehmet123]

As I wrote before; "I have making aggressive changes especially in the search section in Kayra for Kayra 1.1 version. I' m not worried if these changes will cause loss of power in Kayra. For me, the goal isn't yet to be at the top at least until Kayra 2 and there is still a long way to this version.(Kayra will continue Kayra 1.1, Kayra 1.2,.....)"

I have working on a full-time job and I have been involved in a ridicilous project for some time, such as AI project that can predict stock prices based on fundamental and technical analysis.
Since I love computer chess, I try to do something by reducing my sleeping hours. If I can find time, I will continue my projects related to computer chess. But from now on, I will no longer take part in any discussion about Kayra.

[Dicaste]

It has a great score in Marathon test suite in chessbase.


Also it plays well against FF2

4min+2sec 6Thread 2048MB Hashtable without books.

[pgn][Event "Machine, Blitz 4min+2sec"]
[Site "Machine"]
[Date "2021.12.27"]
[Round "6"]
[White "Kayra 1.1 bmi2"]
[Black "Fat Fritz 2"]
[Result "1-0"]
[ECO "B33"]
[Annotator "0.51;0.31"]
[PlyCount "111"]
[TimeControl "240+2"]

{Intel(R) Core(TM) i7-8700 CPU @ 3.20GHz 3192 MHz W=30.8 plies; 7.052kN/s; 3.
273.882 TBAs B=30.7 plies; 7.899kN/s; 2.309.734 TBAs} 1. e4 {0.51/31 9 Both
last book move} c5 {0.31/32 15} 2. Nf3 {0.54/29 3} Nc6 {0.37/30 4} 3. d4 {
0.80/29 4 (Bb5)} cxd4 {0.21/32 3} 4. Nxd4 {0.64/31 7} Nf6 {0.18/31 3} 5. Nc3 {
0.44/31 5} e5 {0.27/31 3} 6. Ndb5 {0.57/33 4} d6 {0.38/31 3} 7. Nd5 {0.55/32 4}
Nxd5 {0.27/31 4} 8. exd5 {0.64/34 10} Nb8 {0.29/32 5} 9. a4 {0.56/32 4 (c4)}
Be7 {0.37/31 8} 10. Be2 {0.50/32 7} O-O {0.16/33 17 (Nd7)} 11. Bd2 {0.50/33 9
(0-0)} f5 {0.04/28 3} 12. a5 {0.50/37 26} Nd7 {0.21/33 26 (f4)} 13. O-O {
0.65/31 4} a6 {0.26/28 3} 14. Na3 {0.58/32 6} f4 {0.25/27 7 (e4)} 15. Nc4 {
0.74/27 5} Rf6 {0.09/27 5 (Rb8)} 16. Re1 {0.72/39 18 (Bg4)} Rh6 {0.07/27 5} 17.
Bb4 {0.64/29 4 (Ra3)} Rb8 {0.13/31 24} 18. Bd3 {0.75/27 4 (Bg4)} Qf8 {0.20/25 2
} 19. Ra3 {0.84/33 22 (h3)} b5 {0.13/33 12} 20. axb6 {0.75/31 3} Nxb6 {0.20/31
2} 21. Nxb6 {0.90/29 4} Rxb6 {0.34/34 6} 22. Re4 {0.65/33 8 (Bd2)} g5 {0.43/29
26 (Bd8)} 23. Rc3 {1.06/27 6 (Rc4)} Bf5 {0.76/31 12} 24. Rec4 {1.32/29 4} g4 {
0.78/28 2 (Rb8)} 25. g3 {1.48/27 4} fxg3 {0.62/28 3} 26. hxg3 {1.45/30 4} Rg6 {
0.71/31 7} 27. Ba3 {1.58/32 8 (Bxf5)} Rb8 {0.70/29 3} 28. Bxf5 {1.75/30 3} Qxf5
{0.96/27 4} 29. Qd3 {1.65/31 5} Qg5 {0.90/33 18 (Rf8)} 30. Rc8+ {1.80/29 6}
Rxc8 {1.12/31 2} 31. Rxc8+ {1.73/31 5} Kg7 {1.08/29 4} 32. Rc7 {1.98/29 4} a5 {
1.25/34 11} 33. b3 {2.17/30 5} Kf8 {1.28/32 2} 34. b4 {2.71/32 5} axb4 {
1.35/33 4} 35. Bxb4 {2.90/33 4} Qf6 {1.88/35 9} 36. Bd2 {2.97/35 5} Kg8 {
1.60/31 3} 37. c4 {3.09/35 7} Bf8 {1.71/34 3} 38. c5 {3.25/31 4} dxc5 {1.79/36
3} 39. Rc6 {3.25/33 4} Qf7 {1.82/33 4} 40. Qe4 {3.36/29 4} Bd6 {1.96/32 4} 41.
Be3 {3.36/31 5} Qf3 {2.00/35 3} 42. Rc8+ {3.63/31 7} Kg7 {2.04/33 2} 43. Qa4 {
4.10/32 6} Qxd5 {2.12/33 3} 44. Rd8 {4.27/32 4 (Qe8)} Qe6 {2.75/34 10} 45. Qa7+
{4.33/29 5} Kf6 {2.91/32 5} 46. Qxh7 {4.62/33 4} Qe7 {3.04/31 4} 47. Qh8+ {
4.72/33 5 (Qh1)} Kf5 {3.08/31 2} 48. Qh5+ {5.04/32 4} Kf6 {2.75/28 2} 49. Rh8 {
5.23/34 11} c4 {3.23/36 17 (Qa7)} 50. Rh7 {7.15/25 2 (Qh4+)} Qf8 {5.46/28 3
(Qe8)} 51. f4 {7.57/24 3} c3 {5.69/26 1} 52. f5 {7.55/30 3} Rg8 {6.47/28 2} 53.
Rh6+ {10.10/25 4} Ke7 {7.25/26 2} 54. f6+ {65.63/27 3} Qxf6 {7.83/26 2} 55.
Rxf6 {66.37/30 3} Kxf6 {8.73/26 2} 56. Qh6+ {69.83/29 3} 1-0

[/pgn]

[pgn][Event "Machine, Blitz 4min+2sec"]
[Site "Machine"]
[Date "2021.12.27"]
[Round "8"]
[White "Kayra 1.1 bmi2"]
[Black "Fat Fritz 2"]
[Result "1-0"]
[ECO "C67"]
[Annotator "0.60;0.27"]
[PlyCount "185"]
[TimeControl "240+2"]

{Intel(R) Core(TM) i7-8700 CPU @ 3.20GHz 3192 MHz W=32.6 plies; 7.048kN/s; 2.
853.406 TBAs B=32.6 plies; 8.023kN/s; 2.736.772 TBAs} 1. e4 {0.60/33 7 Both
last book move} e5 {0.27/33 16} 2. Nf3 {0.59/34 3} Nc6 {0.44/32 4} 3. Bb5 {
0.48/35 3} Nf6 {0.23/30 3} 4. O-O {0.37/36 7} Nxe4 {0.28/33 3} 5. Re1 {0.52/38
4} Nd6 {0.33/35 4} 6. Nxe5 {0.43/38 5} Nxe5 {0.12/38 12 (Be7)} 7. Rxe5+ {
0.39/37 6} Be7 {0.12/34 2} 8. Bf1 {0.35/40 7} O-O {0.21/33 3} 9. d4 {0.38/37 6}
Ne8 {0.14/32 3 (Bf6)} 10. c4 {0.42/32 8} Bf6 {0.23/33 4} 11. Re1 {0.48/31 3} d5
{0.27/33 4} 12. cxd5 {0.61/33 6} Qxd5 {0.14/31 4} 13. Be3 {0.63/33 6} Be6 {
0.19/34 14} 14. Nc3 {0.48/31 6} Qa5 {0.14/32 3 (Qd7)} 15. Ne4 {0.52/32 12 (a3)}
Bd5 {0.16/32 13 (Rd8)} 16. b4 {0.71/35 25 (Nxf6+)} Qa3 {0.13/30 5 (Qb6)} 17.
Nc5 {0.62/34 3} Rd8 {0.18/28 4} 18. Qc2 {0.65/36 6} a5 {0.28/29 4} 19. b5 {
0.65/36 5} b6 {0.16/27 4} 20. Nd3 {0.56/39 8 (Ne4)} Be4 {0.37/32 6} 21. Rad1 {
0.36/37 13} Bf5 {0.43/33 10} 22. Bc1 {0.35/37 6} Qd6 {0.20/31 4} 23. Qe2 {
0.46/36 5} Qd5 {0.22/34 18 (Bxd3)} 24. Ne5 {0.60/29 11} Nd6 {0.37/30 6} 25. Bf4
{0.95/29 4} Rfe8 {0.52/31 34 (Rde8)} 26. Qb2 {0.59/33 13} Rc8 {0.41/30 14 (Re7)
} 27. h3 {1.04/28 10} Be6 {0.49/29 10} 28. Rd2 {1.03/27 5} h6 {0.67/29 6} 29.
Qc1 {0.93/31 19 (Qc2)} a4 {0.72/28 12 (Bf5)} 30. Qd1 {1.19/32 25} Bf5 {0.49/25
2} 31. Re3 {1.29/28 4 (a3)} Be4 {0.74/29 21 (Re6)} 32. Rg3 {1.45/26 4 (a3)} Kh8
{0.99/29 12 (Qe6)} 33. Ra3 {1.42/27 5} Ra8 {1.00/25 2} 34. Rc3 {1.27/31 5} Bd8
{0.85/25 4 (Kg8)} 35. a3 {1.63/29 9 (Re3)} Kg8 {1.15/32 28} 36. Qc1 {1.81/27 4}
Rc8 {1.10/26 4 (Re6)} 37. Rb2 {1.65/30 18} Ra8 {1.00/23 2 (Be7)} 38. Re2 {
2.06/26 1} Bf6 {1.38/26 10} 39. Re1 {2.62/25 2} Rac8 {1.18/22 0} 40. Rxc7 {
2.92/27 2 (Qd1)} Bf5 {1.53/24 1} 41. Qc6 {2.92/29 2} Qxd4 {1.76/31 1} 42. Nf3 {
3.11/29 2} Qxf4 {1.66/30 1} 43. g3 {2.98/35 4} Be4 {1.72/31 2} 44. Rxc8 {
2.93/39 2} Bxc6 {1.82/32 1} 45. Rcxe8+ {2.93/39 2} Bxe8 {1.82/32 1} 46. gxf4 {
2.92/43 3} Bb2 {1.75/29 1} 47. Nd2 {3.05/37 4 (Ne5)} Kf8 {1.76/34 2} 48. Nc4 {
3.05/38 1} Nxc4 {1.73/36 1} 49. Bxc4 {3.03/45 2} f5 {1.73/40 1} 50. Bd5 {
3.49/41 13 (Re6)} Bd4 {2.02/33 2 (Bxa3)} 51. Bc6 {3.64/32 1} Bf7 {2.21/34 2}
52. Bd7 {3.79/33 1} Bc5 {2.09/33 1} 53. Rc1 {3.71/31 2 (Bxf5)} Bxa3 {2.70/39 11
} 54. Ra1 {3.63/32 1} Bc5 {2.34/37 1} 55. Rxa4 {3.33/36 1} g6 {2.31/41 0} 56.
Ra8+ {3.74/34 1 (Ra7)} Ke7 {2.29/38 1} 57. Ra7 {3.96/35 1} Kf8 {2.24/39 1} 58.
Rc7 {3.70/34 2 (h4)} Bd4 {2.24/43 1 (Bd5)} 59. Bc6 {4.02/37 4 (h4)} h5 {
2.32/44 1 (Bc5)} 60. Bd5 {5.66/30 3 (h4)} Be8 {3.79/43 10} 61. Bc4 {5.73/26 1}
Bf6 {3.83/33 1} 62. Rb7 {5.71/33 1} Bd4 {3.71/36 1} 63. Kf1 {5.74/31 1} h4 {
3.59/34 1} 64. Ke2 {5.79/34 2 (Ra7)} Bc5 {3.91/35 1} 65. Rh7 {6.21/34 2 (Kf3)}
Ba3 {4.00/35 1 (Bd4)} 66. Rh8+ {6.70/31 1 (Kf3)} Ke7 {3.25/31 1} 67. Rxh4 {
6.67/31 3} Bc5 {3.58/44 1 (Bb4)} 68. Rh7+ {6.72/32 1} Kf6 {3.66/44 1 (Kf8)} 69.
Kf3 {5.61/39 4 (Rb7)} Bd6 {3.66/40 1 (Bd4)} 70. Rb7 {5.60/39 3 (Kg2)} Bc5 {
3.66/45 1} 71. Kg2 {6.30/37 1 (Ra7)} Bd4 {4.84/42 10} 72. Ra7 {6.75/39 6 (h4)}
Bb2 {4.89/34 2 (Bc3)} 73. Ra6 {6.87/30 1 (Rb7)} Bd4 {4.09/29 0} 74. Ra8 {
6.82/33 1 (Ra7)} Ke7 {4.75/32 2} 75. Ra2 {6.86/32 1 (h4)} Bc5 {4.92/34 2} 76.
Re2+ {6.86/38 1} Kd8 {4.93/35 0} 77. h4 {6.82/34 1 (Kg3)} Bb4 {4.91/35 1 (Bd6)}
78. Re6 {6.91/34 2 (Re5)} Bc5 {5.00/36 1} 79. Bd5 {7.07/29 2 (Re5)} Bf7 {
5.00/31 1} 80. Re5 {7.10/31 2} Be8 {5.05/40 2} 81. Bc6 {7.00/32 3 (Bc4)} Bf7 {
5.08/39 2} 82. h5 {7.23/34 9 (Bd5)} gxh5 {5.22/28 3 (Bf8)} 83. Rxf5 {7.42/28 0}
Ke7 {5.27/29 2} 84. Bd5 {7.30/29 1} Be8 {5.35/27 1} 85. Re5+ {7.63/25 1} Kd8 {
5.58/30 3} 86. Bc6 {7.80/26 1 (Bc4)} Bf7 {5.85/30 2} 87. f5 {7.91/27 2 (Bd5)}
Be7 {6.00/30 7 (Bd4)} 88. Re3 {8.12/26 2} Bg5 {6.18/30 2 (Bc4)} 89. Rg3 {
8.23/27 1} Bf6 {6.40/29 4} 90. Rd3+ {8.34/26 1} Kc8 {6.75/26 1} 91. f4 {
8.51/29 2} Ba2 {6.84/26 1 (Bc4)} 92. Rd6 {9.10/24 1 (Rd2)} Be7 {7.21/27 1 (Bc3)
} 93. Rh6 {10.25/25 1} 1-0

[/pgn]

[AlexChess]

Looks strictly worse than stockfish. What's the point? Does it have any useful features?

Tomasz, in most tests it doesn't seem strictly worse than Stockfish (or FT161221) and Mehmet has already told us that he will release the source code soon. All derivatives are a big resource for all.

Best regards, Alex