Colossal Blunder by Houdini 6.03

[AdminX]

[d]8/4K3/2B5/q2Pb3/8/8/1k6/8 b - - 0 65

In the following position with about 3 minutes left on it's clock, Houdini played 65. ... Qxd5??, I can only imagine that this is a bug.

Code: Select all

 
[Event "15 Minutes/Game"]
[Site "Engine Match"]
[Date "2018.06.29"]
[Round "2"]
[White "Komodo 12.1.1 64-bit  \[N3\]"]
[Black "Houdini 6.03 Pro x64-pext \[3\]"]
[Result "1/2-1/2"]
[ECO "D38"]
[PlyCount "136"]
[EventDate "2018.??.??"]

1. d4 {book 0s} d5 {book 0s} 2. Nf3 {book 0s} Nf6 {book 0s} 3. c4 {book 0s} e6
{book 0s} 4. Nc3 {book 0s} Bb4 {book 0s} 5. cxd5 {book 0s} exd5 {book 0s} 6.
Bf4 {book 0s} Ne4 {0.18/22 37} 7. Rc1 {0.41/22 25} Nc6 {-0.02/22 12 (O-O)} 8.
Nd2 {0.57/25 32 (h3)} g5 {-0.20/21 21 (Bd6)} 9. Be3 {0.33/24 52 (Bg3)} Nxc3 {
-0.13/23 14} 10. bxc3 {0.36/23 5} Ba3 {-0.18/23 10 (Bd6)} 11. Rb1 {0.46/22 20}
f5 {0.06/24 18} 12. g3 {0.24/23 5} Bd6 {-0.01/22 8} 13. Nf3 {0.15/25 15} f4 {
-0.08/23 6} 14. Bd2 {0.37/25 41} O-O {-0.32/22 0 (Qf6)} 15. gxf4 {0.17/23 28}
gxf4 {-0.25/22 5} 16. c4 {0.17/23 20 (Rg1+)} Qf6 {-0.28/23 17} 17. Rg1+ {
0.04/23 26 (Rb5)} Kh8 {-0.26/25 23} 18. Rb5 {0.23/24 7} Bf5 {-0.31/24 23 (Nxd4)
} 19. cxd5 {0.07/22 19} Ne7 {-0.25/23 17 (Nxd4)} 20. Qc1 {-0.04/23 20} a6 {
-0.22/24 31} 21. Rxb7 {-0.09/27 0} Rab8 {-0.31/23 15} 22. Rb3 {-0.06/24 8} Nxd5
{-0.44/24 12} 23. Bg2 {-0.21/26 30 (Qd1)} a5 {-0.53/23 43 (Be6)} 24. Kf1 {
-0.27/22 20} a4 {-0.57/24 28} 25. Rxb8 {-0.17/27 0} Rxb8 {-0.40/26 40} 26. Ne1
{-0.30/27 0} Qe6 {-0.42/25 30} 27. Qc6 {-0.18/26 9 (Bf3)} Bg4 {-0.59/23 20 
(Re8)} 28. Bf3 {-0.57/24 22 (Bxd5)} Bh3+ {-0.67/24 18} 29. Bg2 {-0.60/26 22} f3
{-0.56/26 9} 30. Nxf3 {-0.85/25 12} Nf4 {-0.62/26 13} 31. Bxf4 {-0.91/25 10}
Bxf4 {-0.81/26 0} 32. Qc3 {-1.31/27 57} Bg4 {-0.91/29 0} 33. d5+ {-1.35/27 16}
Be5 {-1.30/26 0} 34. Qc1 {-1.34/27 43 (Qd3)} Qf5 {-1.27/23 31} 35. Nd2 {
-1.51/27 0} Qg6 {-1.53/24 10} 36. h4 {-1.53/24 20 (h3)} Rb2 {-1.67/23 9 (Bd4)}
37. Bf3 {-1.53/25 36} Bh3+ {-1.94/27 0} 38. Bg2 {-1.53/26 13} Qf5 {-1.98/27 1}
39. e4 {-1.49/27 14} Qf4 {-1.82/28 0} 40. Qc6 {-1.49/28 14} Rxd2 {-2.14/27 0}
41. Qe8+ {-1.49/27 10} Kg7 {-2.18/26 8} 42. Qe7+ {-1.46/28 7} Kh6 {-2.36/28 9}
43. Qg5+ {-1.46/30 4} Qxg5 {-2.36/26 0} 44. hxg5+ {-1.46/29 6} Kg7 {-2.37/27 20
} 45. Bxh3 {-1.46/32 0} Rxa2 {-2.35/24 2} 46. d6 {-1.46/30 4 (Kg2)} cxd6 {
-2.52/26 0} 47. Be6 {-1.46/32 11} Rd2 {-2.60/27 0} 48. Rg2 {-1.46/33 8} a3 {
-2.88/28 0} 49. f3 {-4.09/32 53} Rxg2 {-4.46/34 0} 50. Kxg2 {-12.12/25 8} Kg6 {
-5.24/24 0} 51. Ba2 {-250.00/25 3 (Kf1)} Kxg5 {-6.59/26 10} 52. Kf1 {-250.00/
54 0 (Be6)} Kf4 {-8.04/24 20} 53. Kf2 {-250.00/60 0 (Kg2)} h5 {-148.82/27 7}
54. Kg2 {-250.00/50 8} h4 {-148.86/39 0} 55. Bd5 {-250.00/50 0 (Kh3)} h3+ {
-148.89/42 10} 56. Kxh3 {-250.00/54 0} Kxf3 {-148.90/53 1} 57. Kh4 {-250.00/50
0} Ke3 {-148.91/57 24} 58. Kg5 {-250.00/57 0} Kd3 {-148.92/56 2 (Kd4)} 59. Kg6
{-250.00/44 0 (Kf5)} Kc2 {-148.93/56 12} 60. Kf7 {-250.00/50 0 (Bg8)} Kb2 {
-148.94/57 11} 61. Ke7 {-250.00/49 0 (Bc4)} a2 {-148.95/57 23} 62. Ke6 {
-250.00/48 0 (Kd8)} a1=Q {-148.96/55 12} 63. Ke7 {-250.00/48 0} Qa5 {-148.97/
55 1} 64. Bc6 {-250.00/47 12} d5 {-148.98/60 0} 65. exd5 {-250.00/82 0} Qxd5 {
-148.95/99 0} 66. Bxd5 {0.00/5 0} Kc3 {0.00/25 20 (Bd4)} 67. Bh1 {0.00/40 3 
(Ke6)} Kd4 {0.00/24 10 (Bd4)} 68. Ke6 {0.00/36 3} Bf4 {0.00/23 0 (Bc7)} 1/2-1/2

[Guenther]

[d]8/4K3/2B5/q2Pb3/8/8/1k6/8 b - - 0 65

In the following position with about 3 minutes left on it's clock, Houdini played 65. ... Qxd5??, I can only imagine that this is a bug.

As we already have 6men on the board it seems to depend on your tablebase setup.
(speculating about incomplete ones...or interfering Nalimov/Syzygy?)

[AdminX]

[d]8/4K3/2B5/q2Pb3/8/8/1k6/8 b - - 0 65

In the following position with about 3 minutes left on it's clock, Houdini played 65. ... Qxd5??, I can only imagine that this is a bug.

As we already have 6men on the board it seems to depend on your tablebase setup.
(speculating about incomplete ones...or interfering Nalimov/Syzygy?)

I was wondering about that myself. I have the complete set of SYZYGY, so I was wondering if Nalimov was was the problem. Checking my Shredder GUI I found that GUI boxes were checked to use tablebases (Nalimov and Shredderbase), so I unchecked them and restarted the match. Hopefully it is/ was just a GUI configuration issue.

[Guenther]

I was wondering about that myself. I have the complete set of SYZYGY, so I was wondering if Nalimov was was the problem. Checking my Shredder GUI I found that GUI boxes were checked to use tablebases (Nalimov and Shredderbase), so I unchecked them and restarted the match. Hopefully it is/ was just a GUI configuration issue.

I have only 5men here, because full 3-6men WDL+DTZ are >150GB and I need already a lot of space for other things.
You could try if you can reproduce it with various settings.

[Henrik Dinesen]

Haven't you checked with manual anaylysis?

[Henrik Dinesen]

I have only 5men here, because full 3-6men WDL+DTZ are >150GB and I need already a lot of space for other things.
You could try if you can reproduce it with various settings.

When you can/will find the room, that 150Gb is well worth it, Gunther. With a "high" endgame hash setting the speeed is _noticiable_ when I analyse endgames, or near-endgames.

Henrik