The wizardry of Spark 0.3

[Mike S.]

The following results are from a D945 cpu only (~dual P4), from one core at 3.4 GHz, 128 MB hash. These are among the most difficult test positions in my collection:

158: Peter Güssow - #15
[d]5r1n/3pp3/3qp3/1Rpp1NK1/3bkpp1/4pb2/6P1/3RQ2N w - - 0 1

Analysis by spark-0.3:

1.Nxd6+ exd6 2.Nf2+ Ke5 3.Nd3+ Ke4 4.gxf3+ gxf3 5.Rb7 Nf7+ 6.Kg6 Ne5+ 7.Nxe5 dxe5 8.Rxd7
+- (3.57) Depth: 1 00:00:00
+- (2.70) Depth: 6/26 00:00:00 73kN
1.Nf2+ Ke5 2.Nd3+ Ke4 3.gxf3+ gxf3 4.Nf2+ Ke5 5.Ng4+ Ke4 6.Rxd4+ cxd4 7.Nxd6+ exd6 8.Qb1+ d3 9.Qb4+ d4 10.Re5+ dxe5 11.Qb7+ d5 12.Qh7+ Ng6 13.Qxg6+ Rf5+ 14.Qxf5+ exf5 15.Nf6#
+- (2.98) Depth: 6/30 00:00:00 112kN
+- (#15) Depth: 8/30 00:00:01 677kN

175: Hiarcs 12.239 - Jonny, Thüringen
[d]2k2r1r/pp1bppbn/6n1/q1pN2Pp/2P1PB2/P2PQ2P/4N1B1/1R3RK1 w - - 0 1

Analysis by spark-0.3:

20.Qc1 Nxf4 21.Qxf4
+/- (1.16) Depth: 1 00:00:00
20.Rb5 Bxb5 21.Qxc5+ Kd7 22.Bd2 Qxd2 23.Qc7+ Ke6 24.Nef4+ Qxf4 25.Rxf4 Rc8 26.Qxb7 Bd4+ 27.Kh1 Nxg5 28.Qxb5 Nxf4 29.Nxf4+ Kf6
+/- (1.27) Depth: 1 00:00:00
+/- (1.09) Depth: 10/20 00:00:01 868kN
20.Rxb7 e5 21.Rfb1 Kd8 22.R1b5 Qxa3 23.Rb8+ Bc8 24.h4 Ne7 25.Bh3 f5 26.exf5 Qa1+ 27.Kh2 Qa3 28.Rxc5 Qa1 29.Nxe7 Kxe7 30.Bxe5
+/- (1.37) Depth: 10/23 00:00:02 1149kN
+- (5.85) Depth: 15/38 00:00:58 28566kN

1: Von_Witonsti,G - [+4884.76d1g8], tb13 1913
[d]1n1r1rk1/ppq2ppp/3p2b1/3B1NP1/4PB1R/bP2P2P/P1P5/3KQ1R1 w - - 0 1

Analysis by spark-0.3:

(...)
1.c4 Nc6 2.Qc3 Bxf5 3.exf5 Bb4 4.Qa1 Ne5 5.g6 hxg6 6.Bxe5 dxe5 7.Rxg6
+/- (1.24) Depth: 7/10 00:00:00 36kN
+/- (1.29) Depth: 10/16 00:00:00 197kN
1.Qc3 Qxc3 2.Ne7+ Kh8 3.Nxg6+ fxg6 4.Rxh7+ Kxh7 5.Rg4 Rxf4 6.Rxf4 Qa1+ 7.Ke2 Qf1+ 8.Kxf1 Rf8 9.Rxf8 Nd7 10.Rf4 Nf6 11.gxf6 gxf6 12.Rxf6 b5 13.Rf7+ Kh6 14.Rxa7 Bc5 15.Rd7 Kg5 16.c3
+/- (1.39) Depth: 10/17 00:00:00 271kN
+- (9.55) Depth: 21/48 00:01:15 50187kN

7: Karpov,A - Shirov,A, Biel 1992
[d]r5k1/Rb4p1/2q2pBp/1pp5/1b4QN/1P2P2P/5PP1/6K1 w - - 0 1

Analysis by spark-0.3:

(...)
30.Rxa8+ Bxa8 31.Bd3 g5 32.Nf5 Kf8 33.Nxh6 Be1 34.f3 Bb7 35.Qh5 Qe6
+/- (1.06) Depth: 8/18 00:00:00 101kN
+/- (1.01) Depth: 11/22 00:00:00 472kN
30.Rxb7 Qxb7 31.Qe6+ Kh8 32.Be4 Qxe4 33.Qxe4 Rb8 34.Ng6+ Kg8 35.Qd5+ Kh7 36.Ne7 Ba3 37.Nf5 Rg8 38.Qd3 Rb8 39.Qd7 Rg8 40.Qxb5 Bb4 41.Qd7 Kh8 42.Kh2
+/- (1.36) Depth: 11/25 00:00:00 568kN
+- (7.20) Depth: 19/42 00:01:09 54020kN

Of course we can also find positions where Spark 0.3 cannot solve that fast, within the first few seconds. But the examples above show a great potential IMO.

[Vinvin]

Great, when I saw that, I tried one of my set ( http://home.scarlet.be/vincentlejeune/hardsets.html )
I found no engine before that find Rxg7!! in a reasonnable time, Spark 0.3 found it in 8m19s on an AMD 4200+(2.2 Ghz)
This spark is very special, as Bright but may be more

13: Pos - Attack the castled king, hard CCC 2008
1r1rb1k1/5ppp/4p3/1p1p3P/1q2P2Q/pN3P2/PPP4P/1K1R2R1 w - - 0 1

Analysis by spark-0.3:
...
1.Df6 Df8 2.bxa3 Tbc8 3.De5 Fc6 4.Cd4 Fe8 5.Df6 Ta8 6.Td3 Tac8 7.h6
² (0.30) Profondeur: 13/31 00:00:20 14621kN
1.Df6
² (0.55) Profondeur: 14/31 00:00:26 18152kN
1.Df6 Df8 2.exd5 exd5 3.bxa3 Ta8 4.Tde1 Fd7 5.Te7 Ff5 6.Cd4 Fg6 7.De5 Te8 8.Txe8 Txe8 9.Dxd5
² (0.56) Profondeur: 14/31 00:00:26 18669kN
1.Df6 Df8 2.exd5 exd5 3.bxa3 Tbc8 4.Cd4 Rh8 5.Df5 Fd7 6.De5 h6 7.Dxd5 Tc5 8.Db7 Txh5
² (0.44) Profondeur: 15/31 00:00:35 25185kN
1.Df6 Df8 2.exd5 exd5 3.bxa3 Tbc8 4.Cd4 Rh8 5.Df5 Fd7 6.Dxd5 Tc5 7.Db7 Txh5 8.Cc6 Fxc6 9.Txd8 Dxd8 10.Dxc6
² (0.41) Profondeur: 16/37 00:00:58 42573kN
1.Df6 Df8 2.exd5 exd5 3.bxa3 Tbc8 4.Cd4 Rh8 5.Df5 Fd7 6.Dxd5 Tc5 7.Db7 Txh5 8.Cc6 Fxc6 9.Txd8 Dxd8 10.Dxc6
² (0.41) Profondeur: 17/40 00:01:49 79625kN
1.Df6 Df8 2.exd5 exd5 3.bxa3 Rh8 4.Dd4 Td6 5.Tg5 f6 6.Txd5 Txd5 7.Dxd5 Ff7 8.Dc5 Dxc5 9.Cxc5 Fxh5 10.Td3 Ta8 11.Ce6
² (0.48) Profondeur: 18/47 00:03:45 157mN
1.Df6 Df8 2.exd5 exd5 3.bxa3 Ta8 4.Tde1 Rh8 5.De7 Dxe7 6.Txe7 Txa3 7.Tge1 Taa8 8.Cd4 Tdc8 9.T1e5 b4 10.Txd5 f6 11.Ce6
² (0.41) Profondeur: 19/48 00:07:46 296mN
1.Txg7+
± (0.73) Profondeur: 19/48 00:08:19 316mN
1.Txg7+
± (1.23) Profondeur: 19/48 00:08:29 321mN
1.Txg7+
+- (1.98) Profondeur: 19/48 00:08:39 326mN
1.Txg7+
+- (3.98) Profondeur: 19/48 00:09:02 342mN
1.Txg7+ Rxg7 2.h6+ Rf8 3.Df6 Fd7 4.bxa3 Dd6 5.Tg1 Re8 6.Tg8+ Df8 7.Txf8+ Rxf8 8.Cc5 Re8 9.Cd3 Tdc8 10.Ce5 Tb7 11.exd5 exd5 12.Dxf7+ Rd8 13.Dxh7 b4 14.Dg8+ Rc7 15.h7 bxa3+ 16.Rc1 Tbb8
+- (4.92) Profondeur: 19/48 00:09:31 359mN
1.Txg7+
+- (5.17) Profondeur: 20/48 00:09:51 373mN
1.Txg7+
+- (5.67) Profondeur: 20/48 00:09:58 376mN
1.Txg7+
+- (6.42) Profondeur: 20/48 00:10:20 389mN
1.Txg7+ Rxg7 2.h6+ Rf8 3.Df6 Fd7 4.bxa3 Dxa3 5.Tg1 Re8 6.Tg8+ Df8 7.Txf8+ Rxf8 8.Cc5 dxe4 9.Cxe4 Tdc8 10.Cg5 Re8 11.Cxf7 Tb6 12.Ce5 Td6 13.Dh8+ Re7 14.Dxh7+ Rd8 15.Dg8+ Rc7 16.h7 Td2
+- (6.22) Profondeur: 20/50 00:11:01 418mN
1.Txg7+
+- (6.47) Profondeur: 21/50 00:11:45 445mN
1.Txg7+ Rxg7 2.h6+ Rf8 3.Df6 Fd7 4.bxa3 Dd6 5.Tg1 Re8 6.Tg8+ Df8 7.Txf8+ Rxf8 8.Cc5 dxe4 9.Cxe4 Tdc8 10.Cg5 Re8 11.Cxf7 Tb6 12.Ce5 Td6 13.Dh8+ Re7 14.Dxh7+ Rd8 15.Dg8+ Rc7 16.h7 Td2
+- (6.94) Profondeur: 21/50 00:13:34 515mN
1.Txg7+
+- (7.19) Profondeur: 22/50 00:14:59 565mN

[Dr.Wael Deeb]

Wow,thanks Mike
The last three positions are simply awesome....to describe them with the word wizardy is an underestimation to say the least....
Dr.D