Komodo 10.4 C0 vs. Houdini 5.01, 4CPU, TC 600+10

[Jeroen]

Hi Andrey,

Thanks for playing these matches, using my TCEC 9 super final testset.

In the current match Houdini loses 4 games because of illegal moves: see games 21, 28, 71 and 76. Also the behaviour towards the end of game 21 is very strange.

Any idea what causes this? Is it a flaw in Houdini, or is it the GUI that is causing a bug?

Kind regards, Jeroen

[majortom]

Code: Select all

     Program                          Elo    +   -   Games   Score   Av.Op.  Draws

  1 Komodo 10.4 64-bit             :   42   34  32   100    56.0 %      0   76.0 %
  2 Houdini 5.01 Pro x64-pext      :    0   32  34   100    44.0 %     42   76.0 %

  1 Komodo 10.4 64-bit             :   42  100 (+ 18,= 76,-  6), 56.0 %
  2 Houdini 5.01 Pro x64-pext      :    0  100 (+  6,= 76,- 18), 44.0 %

[majortom]

Hi Andrey,

Thanks for playing these matches, using my TCEC 9 super final testset.

In the current match Houdini loses 4 games because of illegal moves: see games 21, 28, 71 and 76. Also the behaviour towards the end of game 21 is very strange.

Any idea what causes this? Is it a flaw in Houdini, or is it the GUI that is causing a bug?

Kind regards, Jeroen

Thanks Jeroen for these findings.
I need to check it.
All the error games will be finished from positions before the illegal moves.

[majortom]

I found I have used Cutechess-cli 0.6.0 instead of 0.7.0 version.
The illegal moves was made during the endgames.

[majortom]

All the error games will be finished from positions before the illegal moves.

#21 from 8/5K2/8/8/1k6/8/8/7q b - - 0 175

#28 from 6k1/6B1/8/7P/6QK/8/8/8 w - - 1 88

#71 from 8/8/R7/1k6/8/8/8/1K6 b - - 0 60

#76 from 3R4/1k6/8/8/8/5K2/8/8 w - - 0 86

[majortom]

Fixed:

Code: Select all

    Program                          Elo    +   -   Games   Score   Av.Op.  Draws

  1 Komodo 10.4 64-bit             :   18   33  32   100    52.5 %      0   77.0 %
  2 Houdini 5.01 Pro x64-pext      :    0   32  33   100    47.5 %     18   77.0 %

  1 Komodo 10.4 64-bit             :   18  100 (+ 14,= 77,-  9), 52.5 %
  2 Houdini 5.01 Pro x64-pext      :    0  100 (+  9,= 77,- 14), 47.5 %

http://treu.ru/2017/k104_h501_600_10_c0.pgn

[Jeroen]

Thanks for the correction.

BTW, lots of great games in your matches. I enjoy replaying them!

Regards, Jeroen

[majortom]

Thank you for the kind words!
Have you got any plans for the new opening set?

[Nordlandia]

TC: 45m+45s
Hash: 2048 Mb
CPU: i5-5200U
EGTB: 6-men Syzygy on SSD
GUI: Fritz 15 [Last Patch]
OS: Windows 10

Opening: King's Indian Defense, Kozul Gambit.

[pgn][Event "Lenova-Laptop #2, Slow 45m+45s"]
[Site "Lenova-Laptop #2"]
[Date "2017.03.30"]
[Round "1"]
[White "Stockfish 280317 64 POPCNT"]
[Black "Houdini 5.01 Pro x64-pext"]
[Result "0-1"]
[Annotator "0.22;0.60"]
[SetUp "1"]
[FEN "r2q1rk1/Nppb2bp/n2p2n1/3Pp1p1/1P2Pp2/5P2/P3BBPP/2RQNRK1 b - - 0 1"]
[PlyCount "81"]
[TimeControl "2700+45"]

{Intel(R) Core(TM) i5-5200U CPU @ 2.20GHz 2195 MHz W=36.8 plies; 2 704kN/s;
67 826 381 TBAs B=27.9 plies; 3 508kN/s; 175 555 TBAs} 1... h5 {0.60/24 259
Both last book move} 2. a4 {0.22/26 63} Nxb4 {0.27/24 64} 3. Qb3 {0.29/30 92}
g4 {0.56/26 136} 4. Nc6 {0.43/29 16} Qg5 {0.42/24 58} 5. Kh1 {0.40/31 96} g3 {
0.49/26 106} 6. Bg1 {0.41/31 17} Na6 {0.53/26 128} 7. Qxb7 {0.59/34 98} Nc5 {
0.46/25 64} 8. Qxc7 {0.54/26 24} Rfc8 {0.50/24 240} 9. Qb6 {0.80/29 68} Rxa4 {
0.55/26 127} 10. Bb5 {0.56/27 15} Ra3 {0.35/24 49} 11. Nd4 {0.18/32 435} gxh2 {
0.64/24 148} 12. Bf2 {0.22/32 104} Nxe4 {0.40/24 48} 13. Rxc8+ {0.16/29 24}
Bxc8 {0.30/25 17} 14. fxe4 {0.48/30 87} exd4 {0.28/26 74} 15. Qxd6 {0.08/29 19}
Rc3 {0.10/27 51} 16. Qb8 {0.08/34 84} Qg4 {0.18/26 39 (Nf8)} 17. d6 {0.09/31
143} Nf8 {0.10/24 0} 18. Qa7 {0.00/32 127} Bd7 {0.10/28 82} 19. Qa6 {0.00/33
93 (Bxd4)} f3 {0.10/24 88 (Be6)} 20. Nxf3 {0.00/35 87 (gxf3)} Rxf3 {-0.10/28 63
} 21. Be2 {0.00/39 38} Rxf2 {-0.10/28 0} 22. Bxg4 {0.00/44 79} Rxf1+ {-0.19/31
82} 23. Qxf1 {0.00/45 18} Bxg4 {-0.19/31 17} 24. Qc4+ {0.00/52 123 (Qb5)} Kh7 {
-0.10/33 159} 25. Kxh2 {0.00/50 50} Nd7 {-0.10/31 32 (Be5+)} 26. Qf7 {0.00/51
85 (Qd5)} d3 {-1.45/29 59 (Ne5)} 27. Qd5 {-1.27/44 195} Ne5 {-1.45/27 0} 28.
Kg3 {-0.73/36 42 (Kh1)} Bf6 {-1.52/30 111 (Bd7)} 29. Kf2 {-0.73/45 109} h4 {
-1.54/34 91 (Bd7)} 30. Kg1 {-1.74/43 183} Kg6 {-1.54/34 41 (Bd7)} 31. d7 {
-1.86/45 128 (Kh2)} Bxd7 {-1.76/30 77} 32. Qg8+ {-2.68/42 152} Kh6 {-1.88/31 70
} 33. Qf8+ {-2.97/46 251} Bg7 {-1.94/33 81} 34. Qg8 {-4.54/46 580} Bg4 {
-2.01/36 120} 35. Qd8 {-5.39/39 144} Kg6 {-2.15/29 71} 36. Qe8+ {-6.44/40 132}
Nf7 {-2.13/28 0} 37. Qc6+ {-6.46/42 128} Bf6 {-2.60/29 65} 38. Qd5 {-6.11/32 10
} Ne5 {-2.69/29 103} 39. Kf2 {-7.26/38 203} Kg7 {-3.14/29 72 (Kh5)} 40. Qd6 {
-7.62/36 106} Bh5 {-3.66/29 67 (Bd7)} 41. Kf1 {-8.32/35 54} Bf7 {-4.56/27 68
(Bg4)} 0-1

[/pgn]

Lichess analysis & full PGN - https://nb.lichess.org/cMLoyhN9

http://www.viewchess.com/cbreader/2017/ ... 59046.html

[Jeroen]

Hi Andrey,

No plans for a new testsuite at the moment. But in the future: probably yes .

Kind regards, Jeroen