Test position: GM attacking lessons

[Albert Silver]

Though I have no doubt some engine will end up finding it quite quickly, it is a tough shot nonetheless. In the Aeroflot Open, Anton Korobov (who already found a move last year that no engine could find in 10 mins) once again showed his attacking skills:

[D]r1b2rk1/pp4pp/q3pp2/n1n1N2Q/2p1P3/P1N3P1/1P3PBP/3R1RK1 w - - 0 18

18.Rd8!!

18...fxe5 19.Rxf8+ Kxf8 20.f4! Ke7 21.Qxe5 and White regains the piece with interest.

[GARGAMEL]

Hello

NN - NN, Td8
r1b2rk1/pp4pp/q3pp2/n1n1N2Q/2p1P3/P1N3P1/1P3PBP/3R1RK1 w - - 0 1

Analysis by Critter 0.90 64-bit:

1.Sg6 hxg6 2.Dxc5 b6 3.Db4 e5 4.Td6 Sb7 5.Tc6 Tf7 6.Sb5 Sc5 7.Td1 Da4 8.Td8+ Tf8 9.Txf8+ Kxf8 10.Dd2 Sd3 11.Txf6+ gxf6 12.Dh6+ Kf7 13.Dh7+ Kf8 14.Dh8+ Kf7
=/+ (-0.33) Tiefe: 6/8 00:00:00 3kN
= (0.00) Tiefe: 26/55 00:04:22 1280mN
1.Td8!! fxe5 2.Txf8+
= (0.11) Tiefe: 26/56 00:09:03 2661mN
+- (1.64) Tiefe: 26/98 00:30:25 8869mN

regards Anton

[GARGAMEL]

Hello

NN - NN, Td8
Analysis by Spark-1.0:

1.Sg6 hxg6 2.Dxc5 b6 3.Db4 Lb7 4.Lh3 Tfe8 5.Sb5 Sc6 6.Dxc4 Se5 7.Lxe6+ Kh8 8.Db3 Lxe4 9.Sd4 Tad8 10.f4 Ld3 11.Tfe1 Lc4 12.Lxc4 Dxc4
+/= (0.36) Tiefe: 6/8 00:00:00 5kN
+/= (0.36) Tiefe: 22/64 00:01:12 1266mN
1.Td8!! fxe5 2.Txf8+ Kxf8 3.f4 Ke7 4.Dxe5 Ke8 5.Dh5+ g6 6.Dxc5 Db6 7.Dxb6 axb6 8.Td1 Ld7 9.e5 Ke7 10.Se4 Sc6 11.Sf6 Le8 12.Sxh7 c3 13.bxc3 Txa3
+/= (0.61) Tiefe: 22/64 00:01:35 1628mN
+/- (1.39) Tiefe: 22/64 00:02:08 2155mN

regards Anton

[zullil]

Anton Korobov (who already found a move last year that no engine could find in 10 mins)

Can you please post that position from last year, or provide a link? Thanks.

[EDIT] Do you mean this http://www.talkchess.com/forum/viewtopi ... ht=korobov

[D]r1b2r2/p5kp/1p1q1p2/4nP2/3p1N2/3B4/P2Q2PP/4RRK1 w - - 0 21

[yanquis1972]

a quick & dirty custom 'attacking' junior 11 -- (no new line after ~10 mins)

[+0.78] d=21 18.Rd8 (0:01:10) 227129kN
[+0.28] d=21 18.Ng6 hxg6 19.Qxc5 Qb6 20.Qxb6 axb6 21.Rd6 (0:00:08) 49618kN
[+0.30] d=18 18.Ng6 hxg6 (0:00:00) 299kN
[+0.30] d=15 18.Ng6 hxg6 (0:00:00) 37kN
[+0.30] d=12 18.Ng6 hxg6 (0:00:00) 10kN
[+0.30] d=9 18.Ng6 hxg6 (0:00:00) 7kN
[+0.30] d=6 18.Ng6 hxg6 (0:00:00) 6kN
[+0.12] d=6 18.Ng4 Qb6 19.Bh3 Qxb2 20.Qxc5 Qxc3 (0:00:00) 5kN
[-0.63] d=6 18.Nf3 e5 19.Nh4 (0:00:00) 1kN
[+0.30] d=3 18.Nf3 Nd3 (0:00:00) 0k

of course, maybe i should've just tried default settings first :/

DJ 11.2 --

[+1.02] d=19 18.Rd8 fxe5 19.Rxf8+ Kxf8 20.f4 Ke7 21.fxe5 Kd8 22.Qf7 Qb6 23.Kh1 Bd7 24.Qf8+ Kc7 25.Qxa8 Qxb2 26.Qf8 b6 27.Rf3 (0:00:33) 259663kN
[+1.11] d=17 18.Rd8 fxe5 19.Rxf8+ Kxf8 20.f4 Ke7 21.fxe5 Kd8 22.Qf7 Nab3 23.Kh1 Rb8 (0:00:14) 103835kN
[+0.87] d=15 18.Rd8 fxe5 19.Rxf8+ Kxf8 20.f4 Kg8 21.Qe8# (0:00:03) 20102kN
[+0.86] d=15 18.Rd8 (0:00:01) 7308kN
[+0.45] d=15 18.Ng6 hxg6 19.Qxc5 Qb6 20.Qxb6 axb6 21.Rd6 Nc6 22.Bf3 Ra5 23.Be2 b5 24.Rfd1 Ne5 (0:00:00) 2907kN
[+0.38] d=12 18.Ng6 hxg6 19.Qxc5 Qb6 20.Qxb6 axb6 21.Rd6 Ra6 22.Bh3 Nb3 23.Rfd1 e5 (0:00:00) 429kN
[+0.45] d=9 18.Ng6 hxg6 19.Qxc5 Qb6 20.Qb5 Nc6 21.Bh3 Qxb5 22.Nxb5 Ne5 (0:00:00) 37kN
[+0.76] d=6 18.Ng6 hxg6 19.Qxc5 Nb3 20.Qe7 Qb6 (0:00:00) 1kN
[+1.37] d=3 18.Ng6 hxg6 19.Qxc5 (0:00:00) 0kN
[+0.92] d=3 18.Nf3 Nab3 19.Bh3 (0:00:00) 0kN

(triple checked, that was w/ clear hash btw. can anyone beat junior??)

[yanquis1972]

heh -- junior 11 'won' that won too...toiletgate II? (or maybe just a great speculative engine..)

[Albert Silver]

Anton Korobov (who already found a move last year that no engine could find in 10 mins)

Can you please post that position from last year, or provide a link? Thanks.

[EDIT] Do you mean this http://www.talkchess.com/forum/viewtopi ... ht=korobov

[D]r1b2r2/p5kp/1p1q1p2/4nP2/3p1N2/3B4/P2Q2PP/4RRK1 w - - 0 21

Yes, that is the one: Rxe5!!

[peter]

(triple checked, that was w/ clear hash btw. can anyone beat junior??)

Hi!
Less than one second isn't easy to get, but I have found another two oldies goldies:


r1b2rk1/pp4pp/q3pp2/n1n1N2Q/2p1P3/P1N3P1/1P3PBP/3R1RK1 w - - 0 1

Analysis by ProDeo 1.6Q3:

1.Sg6 hxg6 2.Dxc5 e5 3.De7 b5
+/- (0.95) Tiefe: 6 00:00:00 39kN
...
1.Td8
+/- (0.98) Tiefe: 10 00:00:01 2291kN
1.Td8 Txd8 2.Df7+ Kh8 3.De7 Sc6 4.Sxc6 Tg8 5.Sd8 h6 6.Sf7+ Kh7 7.Dxc5
+- (2.94) Tiefe: 10 00:00:01 2898kN
1.Td8 Txd8 2.Df7+ Kh8 3.De7 Sc6 4.Sxc6 Tg8 5.Sd8 h6 6.Sf7+ Kh7 7.Dxc5 e5
+- (1.86) Tiefe: 11 00:00:01 3365kN
1.Td8 Txd8 2.Df7+ Kh8 3.De7 Sc6 4.Sxc6 Tg8 5.Sd8 h6 6.Sf7+ Kh7 7.Dxc5 Db6 8.Dxc4 Dxb2
+- (2.19) Tiefe: 12 00:00:03 6288kN
1.Td8 Txd8 2.Df7+ Kh8 3.De7 Sc6 4.Sxc6 Tg8 5.Sd8 h6 6.Sf7+ Kh7 7.Dxc5 Db6 8.Dxc4 Dxb2
+- (2.19) Tiefe: 13 00:00:04 10751kN
1.Td8 Txd8 2.Df7+ Kh8 3.De7 Sc6 4.Sxc6 Tg8 5.Sd8 h6 6.Sf7+ Kh7 7.Dxc5 e5 8.Sb5 Le6 9.Sc7
+- (2.09) Tiefe: 14 00:00:19 46222kN


r1b2rk1/pp4pp/q3pp2/n1n1N2Q/2p1P3/P1N3P1/1P3PBP/3R1RK1 w - - 0 1

Analysis by Deep Fritz 10:

1.Sg6 hxg6 2.Dxc5 Sb3 3.Dc7 Dc6 4.Dxc6 bxc6 5.e5 Lb7 6.Tfe1 Tad8 7.Txd8 Txd8 8.exf6 gxf6 9.Txe6
= (-0.09) Tiefe: 9/24 00:00:00 174kN
...
1.Td8
= (-0.25) Tiefe: 14/34 00:00:01 6825kN
1.Td8
= (-0.09) Tiefe: 14/34 00:00:01 6890kN
1.Td8 fxe5 2.Txf8+ Kxf8 3.f4 Ke7 4.Dxe5 Sd3 5.Dxg7+ Kd8 6.Kh1 Ld7 7.a4 Db6 8.Df8+ Le8 9.e5 Kc7 10.De7+ Ld7 11.Sb5+ Kc8 12.Sd6+
= (-0.04) Tiefe: 15/39 00:00:04 16630kN
1.Td8 fxe5 2.Txf8+ Kxf8 3.f4 Ke7 4.Dxe5 Sd3 5.Dxg7+ Kd8 6.Kh1 Ld7 7.a4 Db6 8.Df8+ Le8 9.e5 Kc7 10.De7+ Ld7 11.Sb5+ Kc8 12.Sd6+
= (-0.04) Tiefe: 16/44 00:00:07 30558kN
1.Td8 fxe5 2.Txf8+ Kxf8 3.f4 Sd7 4.Dxh7 exf4 5.e5 Ke7 6.Dh4+ Ke8 7.Dh8+ Ke7 8.Dh4+
= (0.00) Tiefe: 17/41 00:00:15 70562kN
1.Td8
+/= (0.28) Tiefe: 18/45 00:00:30 140mN
1.Td8
+/= (0.56) Tiefe: 19/43 00:01:03 293mN
1.Td8
+/- (0.84) Tiefe: 20/51 00:02:13 619mN
1.Td8
+/- (1.13) Tiefe: 20/51 00:02:20 652mN


And then mind the hardware: simple Dualcore of which ProDeo uses one core only.

[yanquis1972]

wow, crazy -- pro deo @ depth 10? realize it's a personality, but it's a true shame (for us) ed gave up programming.

i would love to compare DF12 vs DF10 on this position. any volunteers? remember to limit to 2 cores for at least a roughly parallel result.

[peter]

wow, crazy -- pro deo @ depth 10? realize it's a personality, but it's a true shame (for us) ed gave up programming.

i would love to compare DF12 vs DF10 on this position. any volunteers? remember to limit to 2 cores for at least a roughly parallel result.

Yes, it's the legendary Q3 tactical engine of the legendary ProDeo 1.6 and yes, Ed Schröder really did a great job.

But no, you need not try DF12 on this, not on 2, not on 4 and not on 12 cores, unless you have really much time to leave your machine alone. At least I didn't see the solution with DF12 up to depth 30, which is reached in about 8 min. on a good machine.
Tried DF11 too same time on quad, didn't find it neither.
F10 surely was something special as for tactics, bad enough it doesn't run on my Win7 64bit anymore.