DoubleCheck 2.3 64-bit Gauntlets (CCRL 40/40)

[Aser Huerga]

DoubleCheck 2.3 64-bit Gauntlets

Hardware: Intel Q8400 2.66GHz 4GB RAM
Operating system: Microsoft Windows XP 64 Bit Professional
All engines 1 CPU (SSE versions when available)
Arena 2.0.1 GUI
128MB Hash Tables
3-4-5 EGTBs in SSD
Ponder off
Book ECO_500_Most_Played (from the last 30 years, based on Norm Pollock collections), each pair plays the same position with switched colours.
40 moves in 28 minutes repeating (adapted to the CCRL 40/40 standard)

Code: Select all

1   DoubleCheck 2.3 64-bit -2300- 109.0/216
2   KMTChess 1.2.1 64-bit   2320  15.5/24
3   NagaSkaki 5.12          2265  14.0/24
4   Myrddin 0.85 64-bit     2344  13.5/24
5   OBender 3.2.4.2         2282  13.0/24
6   Neurosis 2.5            2300  11.5/24
7   Kurt 0.9.2.2 64-bit     2294  11.0/24
8   Beowulf 2.4a 64-bit     2352  10.0/24
9   ECE 11.01               2278  9.5/24
10  KnockOut 0.7.1          2250  9.0/24

(All ELOs referred to the last Chess Computer Rating List when tournament starts, except -XXXX- one, wich is referred to performance).

Games: http://kirill-kryukov.com/chess/discuss ... p?id=25037

[lucasart]

DoubleCheck 2.3 64-bit Gauntlets

Hardware: Intel Q8400 2.66GHz 4GB RAM
Operating system: Microsoft Windows XP 64 Bit Professional
All engines 1 CPU (SSE versions when available)
Arena 2.0.1 GUI
128MB Hash Tables
3-4-5 EGTBs in SSD
Ponder off
Book ECO_500_Most_Played (from the last 30 years, based on Norm Pollock collections), each pair plays the same position with switched colours.
40 moves in 28 minutes repeating (adapted to the CCRL 40/40 standard)

Code: Select all

1   DoubleCheck 2.3 64-bit -2300- 109.0/216
2   KMTChess 1.2.1 64-bit   2320  15.5/24
3   NagaSkaki 5.12          2265  14.0/24
4   Myrddin 0.85 64-bit     2344  13.5/24
5   OBender 3.2.4.2         2282  13.0/24
6   Neurosis 2.5            2300  11.5/24
7   Kurt 0.9.2.2 64-bit     2294  11.0/24
8   Beowulf 2.4a 64-bit     2352  10.0/24
9   ECE 11.01               2278  9.5/24
10  KnockOut 0.7.1          2250  9.0/24

(All ELOs referred to the last Chess Computer Rating List when tournament starts, except -XXXX- one, wich is referred to performance).

Games: http://kirill-kryukov.com/chess/discuss ... p?id=25037

Thank you!

It confirms my intuition that DoubleCheck performs somewhat worse at long time control than in blitz. And the reason is that, compared to its "peers", it has a more efficient search but very little chess knowledge.

I'll certainly have a closer look at the games. I will probably find some some chess knowledge to improve its play, especially in the endgame, where it's not very good.

[lucasart]

I'll certainly have a closer look at the games. I will probably find some some chess knowledge to improve its play, especially in the endgame, where it's not very good.

This one is a particularly instructive game:

Code: Select all

[Event "DoubleCheck 2.3 64-bit uno"]

[Site "PUESTO3"]

[Date "2011.12.19"]

[Round "2"]

[White "KnockOut 0.7.1"]

[Black "DoubleCheck 2.3 64-bit"]

[Result "1-0"]

[ECO "E60"]

[WhiteElo "2250"]

[BlackElo "1"]

[PlyCount "148"]

[TimeControl "40/1680:40/1680:40/1680"]



1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. Nf3 d6 5. g3 O-O 6. Bg2 Nbd7 7. O-O {

+0.19/10 42} e5 {-0.23/15 41} 8. dxe5 {+0.60/10 46} dxe5 {-0.21/16 40} 9. Qb3 {

+0.66/9 27} Re8 {-0.11/15 39} 10. Rd1 {+0.57/10 47} c6 {-0.12/15 38} 11. Ng5 {

+0.81/9 34} Qc7 {-0.05/16 37} 12. Be3 {+0.87/9 48} Ng4 {+0.03/16 37} 13. Nce4 {

+0.75/9 33} Nxe3 {+0.34/16 36} 14. Qxe3 {+0.63/9 26} h6 {+0.37/15 35} 15. Nd6 {

+0.58/11 50} Re7 {+0.20/17 34} 16. Nge4 {+0.85/10 36} f5 {+0.39/16 33} 17. Nxc8

{+0.76/11 50} Qxc8 {+0.28/17 32} 18. Nd6 {+0.71/11 51} Qc7 {+0.21/16 31} 19.

Rd2 {+0.70/10 39} Nf6 {+0.27/15 31} 20. Rad1 {+0.64/9 32} e4 {+0.33/15 30} 21.

Qb3 {+0.69/10 52} h5 {+0.32/15 29} 22. e3 {+0.91/10 52} Ng4 {+0.36/15 28} 23.

Bf1 {+0.89/9 51} Kh7 {+0.42/15 28} 24. c5 {+0.93/9 33} b5 {+0.45/15 27} 25. Be2

{+0.79/9 27} a6 {+0.41/15 26} 26. Qc2 {+0.75/10 56} Nf6 {+0.44/14 25} 27. h3 {

+0.32/10 56} Nd5 {+0.44/14 25} 28. g4 {+0.01/10 55} Be5 {+0.44/15 24} 29. gxf5

{+0.08/11 35} Bxd6 {+0.46/15 24} 30. cxd6 {+0.38/11 50} Qxd6 {+0.27/16 23} 31.

Rc1 {+0.39/10 25} Rc8 {+0.17/15 22} 32. Qc5 {+0.17/10 38} Qxc5 {0.00/15 22} 33.

fxg6+ {+1.47/11 21} Kxg6 {-0.22/17 21} 34. Rxc5 {+1.69/12 25} Rcc7 {-0.22/14 21

} 35. Bd1 {+2.39/11 70} Rg7 {-0.35/15 20} 36. Kh1 {+2.95/12 86} Rge7 {

-0.49/16 20} 37. Rd4 {+3.14/12 45} a5 {-0.49/15 19} 38. Bc2 {+3.46/13 63} Nf6 {

-0.76/16 19} 39. Rd6 {+3.65/12 52} a4 {-1.00/17 18} 40. Bd1 {+3.74/14 197} Rh7

{-1.20/17 18} 41. f4 {+3.91/12 79} Rcd7 {-1.28/17 59} 42. Rg5+ {+3.90/14 35}

Kf7 {-1.41/20 58} 43. Rxd7+ {+3.87/14 29} Nxd7 {-1.31/19 56} 44. Bxh5+ {

+3.87/13 39} Ke6 {-1.37/19 55} 45. Bg4+ {+4.09/12 29} Kd6 {-1.30/18 53} 46.

Rg6+ {+4.01/12 18} Kc7 {-1.33/19 52} 47. Bf5 {+4.11/13 22} Re7 {-1.25/19 51}

48. Re6 {+4.07/13 24} Rxe6 {-1.12/23 50} 49. Bxe6 {+3.83/16 11} a3 {-1.14/22 48

} 50. bxa3 {+4.34/17 28} Nf6 {-1.35/23 47} 51. Bb3 {+4.35/17 22} Kd6 {

-1.47/23 46} 52. Kg2 {+4.53/17 44} c5 {-1.63/23 45} 53. a4 {+4.62/18 44} c4 {

-1.89/23 44} 54. Bc2 {+4.62/17 39} bxa4 {-1.96/22 42} 55. Bxa4 {+4.92/17 40}

Ke6 {-1.97/22 41} 56. Kf2 {+5.05/16 33} Kd6 {-1.98/22 40} 57. Bc2 {+5.16/16 34}

Kd5 {-1.99/21 39} 58. Ke1 {+5.34/16 24} Kc5 {-2.22/22 38} 59. Kd2 {+5.41/17 45}

Kd5 {-2.27/22 37} 60. a4 {+6.17/17 34} Ne8 {-3.10/24 36} 61. a5 {+6.15/17 37}

Nc7 {-3.43/23 36} 62. h4 {+6.76/16 48} Kc5 {-4.64/22 35} 63. Bxe4 {+7.29/16 35}

Ne8 {-4.89/23 34} 64. Bg6 {+7.29/16 34} Nf6 {-5.09/22 33} 65. h5 {+7.58/16 41}

Ng4 {-5.05/21 32} 66. Bf5 {+8.10/16 26} Nh6 {-5.04/19 31} 67. Bd7 {+8.37/18 42}

Ng8 {-5.58/22 30} 68. a6 {+8.37/19 54} Kb6 {-5.90/23 30} 69. Bb5 {+8.37/18 31}

Nh6 {-6.41/23 29} 70. Bxc4 {+8.37/18 38} Kc6 {-7.19/21 28} 71. Kc3 {+9.16/17 58

} Kb6 {-8.07/21 27} 72. Kd4 {+9.88/16 32} Ng4 {-10.04/21 27} 73. Be2 {

+10.96/17 62} Nh6 {-11.58/21 26} 74. Ke5 {+12.44/17 62} Nf7+ {

-16.21/22 25  1-0 Arena Adjudication} 1-0

Out of the many bad moves that DoubleCheck playes, I would see 2 typical patterns that it wasn't able to recognize:
1/ doubled rooks on an open file. This helped me spot a bad bug in my mobility evaluation function, where the mobility count is not accounting for squares that each rook sees through the other rook.
2/ knight outposts. A knight that is defended by a pawn, where the pawn cannot be exchanged by another pawn, often becomes a real thorn in the enemy flank; especially when the knight is on the 6-th rank.