Incredible hard testposition Zukertort -Steinitz 1872

[Dr.Ex]

[d] r2q4/pp1b2pp/5pk1/3Q2B1/8/8/PP4PP/4R2K b - - 0 24

Steinitz played 24...Bc6!. Can any engine find this move?

[Event "London m"]
[Site "London"]
[Date "1872.08.06"]
[Round "1"]
[White "Zukertort, Johannes Hermann"]
[Black "Steinitz, William"]
[Result "0-1"]
[ECO "C56"]
[PlyCount "108"]
[EventDate "1872.08.06"]
[EventType "match"]
[EventRounds "12"]
[EventCountry "ENG"]

1. e4 e5 2. Nf3 Nc6 3. Bc4 Bc5 4. c3 Nf6 5. d4 exd4 6. cxd4 Bb6 7. O-O Nxe4 8.
Re1 O-O 9. Rxe4 d5 10. Bxd5 Qxd5 11. Nc3 Qd8 12. d5 Ne7 13. Bg5 f6 14. Qb3 Rf7
15. Rae1 Kf8 16. d6 cxd6 17. Nd5 Nxd5 18. Qxd5 Bd7 19. Qxd6+ Kg8 20. Ne5 Bxf2+
21. Kh1 Bxe1 22. Nxf7 Kxf7 23. Qd5+ Kg6 24. Rxe1 Bc6 (24... fxg5 25. Rd1 Rc8
26. h4 $1 Rc7 (26... Rc6 27. Qd3+ Kh6 28. Qxd7 Qxd7 29. hxg5+ Kxg5 30. Rxd7 Rb6
31. Rxg7+ Kh6 32. Rd7 Rxb2 33. a4 $11) 27. Qd3+ Kh6 28. hxg5+ Kxg5 29. Qd2+ $11
) 25. Qxd8 Rxd8 26. Be3 Kf7 27. Kg1 g5 28. Re2 a6 29. Rd2 Re8 30. Kf2 Bxg2 31.
Bxg5 fxg5 32. Kxg2 Kf6 33. Kf3 h5 34. h4 gxh4 35. Rd4 Kg5 36. Rd5+ Kg6 37. Rd6+
Kf5 38. Rd4 h3 39. Rh4 Kg5 40. Rxh3 h4 41. Kf2 Rc8 42. Rb3 Rc2+ 43. Kg1 b5 44.
a4 bxa4 45. Rb4 Rc1+ 46. Kh2 Rb1 47. Rxa4 Rxb2+ 48. Kh3 Rb3+ 49. Kh2 Rb6 50.
Kh3 Kf5 51. Kxh4 Rg6 52. Kh5 Re6 53. Ra5+ Ke4 54. Kg5 $4 Re5+ {
Deutsche Schachzeitung 1872, p. 309} 0-1

[Albert Silver]

[d] r2q4/pp1b2pp/5pk1/3Q2B1/8/8/PP4PP/4R2K b - - 0 24

Steinitz played 24...Bc6!. Can any engine find this move?

[-1.14] d=14 24...Bc6 25.Qxd8 Rxd8 26.Be3 b6 27.Kg1 Rd7 28.Rc1 Be4 29.Kf2 Kf5 30.b3 g5 (0:00.45)
[-1.14] d=13 24...Bc6 25.Qxd8 Rxd8 26.Be3 b6 27.Kg1 Rd7 28.Rc1 Be4 29.Kf2 Kf5 30.b3 g5 (0:00.39)
[-1.14] d=12 24...Bc6 25.Qxd8 Rxd8 26.Be3 b6 27.Kg1 Rd7 28.Rc1 Be4 29.Kf2 Kf5 30.b3 g5 (0:00.34)
[-1.09] d=11 24...Bc6 25.Qxd8 (0:00.31)
[³/-0.87] d=11 24...fxg5 25.Qd3 Kh6 26.Rd1 (0:00.30)
[-1.53] d=10 24...fxg5 25.Rd1 Rc8 26.Qd3 Kh6 (0:00.07)
[µ/-2.21] d=9 24...fxg5 25.Rd1 Rc8 26.Qd6 Kh5 27.g4 Kxg4 28.h3 Kh5 29.Qxd7 Qxd7 30.Rxd7 Rc1 31.Kh2 Rc2 32.Kg3 Rxb2 33.Rxg7 h6 34.Rd7 (0:00.04)
[µ/-2.10] d=8 24...fxg5 25.Rd1 Rc8 26.Qd6 Kh5 27.g4 Kxg4 28.h3 Kh5 29.Qxd7 Qxd7 30.Rxd7 Rc1 31.Kh2 Rc2 32.Kg3 Rxb2 33.Rxg7 h6 34.Rd7 (0:00.02)
[µ/-2.09] d=7 24...fxg5 25.Rd1 Rc8 26.Qd6 Kh5 27.g4 (0:00.01)

Rybka 3 32-bit takes 31 seconds on my Athlon64 2.2 GHz.

[Dr.Ex]

Great, any other engine would probably take hours to solve that one.

[AdminX]

Great, any other engine would probably take hours to solve that one.

r2q4/pp1b2pp/5pk1/3Q2B1/8/8/PP4PP/4R2K b - - 0 1

Analysis by Deep Junior 10.1:

1...fxg5 2.Rd1 Rc8 3.Qxd7 Qxd7 4.Rxd7
-+ (-1.55) Depth: 3 00:00:00 1kN
= (0.07) Depth: 6 00:00:00 21kN

1...Bc6 2.Qxd8 Rxd8 3.Be3 a6 4.Kg1 Kf5 5.Kf2 Rd5
-/+ (-0.87) Depth: 6 00:00:00 26kN
-+ (-1.54) Depth: 12 00:00:00 1449kN

1...fxg5 2.Rd1 Rc8 3.h3 Rc7 4.Qe5 Kh6 5.a3 Rc6 6.g4
-+ (-1.60) Depth: 12 00:00:00 3447kN
-+ (-1.69) Depth: 15 00:00:02 21054kN

[George Tsavdaris]

[d] r2q4/pp1b2pp/5pk1/3Q2B1/8/8/PP4PP/4R2K b - - 0 24

Steinitz played 24...Bc6!. Can any engine find this move?

[-1.14] d=14 24...Bc6 25.Qxd8 Rxd8 26.Be3 b6 27.Kg1 Rd7 28.Rc1 Be4 29.Kf2 Kf5 30.b3 g5 (0:00.45)
[-1.14] d=13 24...Bc6 25.Qxd8 Rxd8 26.Be3 b6 27.Kg1 Rd7 28.Rc1 Be4 29.Kf2 Kf5 30.b3 g5 (0:00.39)
[-1.14] d=12 24...Bc6 25.Qxd8 Rxd8 26.Be3 b6 27.Kg1 Rd7 28.Rc1 Be4 29.Kf2 Kf5 30.b3 g5 (0:00.34)
[-1.09] d=11 24...Bc6 25.Qxd8 (0:00.31)
[³/-0.87] d=11 24...fxg5 25.Qd3 Kh6 26.Rd1 (0:00.30)
[-1.53] d=10 24...fxg5 25.Rd1 Rc8 26.Qd3 Kh6 (0:00.07)
[µ/-2.21] d=9 24...fxg5 25.Rd1 Rc8 26.Qd6 Kh5 27.g4 Kxg4 28.h3 Kh5 29.Qxd7 Qxd7 30.Rxd7 Rc1 31.Kh2 Rc2 32.Kg3 Rxb2 33.Rxg7 h6 34.Rd7 (0:00.04)
[µ/-2.10] d=8 24...fxg5 25.Rd1 Rc8 26.Qd6 Kh5 27.g4 Kxg4 28.h3 Kh5 29.Qxd7 Qxd7 30.Rxd7 Rc1 31.Kh2 Rc2 32.Kg3 Rxb2 33.Rxg7 h6 34.Rd7 (0:00.02)
[µ/-2.09] d=7 24...fxg5 25.Rd1 Rc8 26.Qd6 Kh5 27.g4 (0:00.01)

Rybka 3 32-bit takes 31 seconds on my Athlon64 2.2 GHz.

This is strange since i get the following analysis in the exact same computer with 256 MB hash tables(in short: i get the move been found in 8 seconds. Perhaps the 256 MB number of hash tables is the critical here for the time difference):

r2q4/pp1b2pp/5pk1/3Q2B1/8/8/PP4PP/4R2K b - - 0 1

Analysis by Rybka 3 1-cpu 32-bit :

1...Bd7-c6
-+ (-1.43) Depth: 2 00:00:00
1...Bd7-c6
-+ (-1.42) Depth: 3 00:00:00
1...Bd7-c6
-/+ (-1.36) Depth: 4 00:00:00
1...Bd7-c6 2.Qd5xd8
-/+ (-1.29) Depth: 5 00:00:01
1...Bd7-c6 2.Qd5xd8 Ra8xd8 3.Bg5-e3 b7-b6 4.Kh1-g1
-/+ (-1.17) Depth: 6 00:00:01 4kN
1...Bd7-f5 2.Qd5xd8 Ra8xd8 3.Bg5-e3 b7-b6
-/+ (-1.18) Depth: 6 00:00:01 5kN
1...Bd7-f5 2.Qd5xd8 Ra8xd8 3.Bg5-e3 b7-b6 4.Kh1-g1
-/+ (-1.13) Depth: 7 00:00:01 7kN
1...Bd7-c6 2.Qd5xd8 Ra8xd8 3.Bg5-e3 b7-b6 4.Kh1-g1 Kg6-f5 5.Kg1-f2
-/+ (-1.17) Depth: 7 00:00:01 10kN
1...f6xg5
-/+ (-1.38) Depth: 7 00:00:01 16kN
1...f6xg5
-+ (-1.58) Depth: 7 00:00:01 19kN
1...f6xg5
-+ (-1.98) Depth: 7 00:00:01 23kN
1...f6xg5 2.Re1-d1 Ra8-c8 3.Qd5-d6+ Kg6-h5 4.g2-g4+ Kh5xg4 5.Qd6-d4+ Kg4-h5 6.Qd4xd7
-+ (-2.10) Depth: 7 00:00:01 37kN
1...f6xg5 2.Re1-d1 Ra8-c8 3.Qd5-d6+ Kg6-h5
-+ (-2.11) Depth: 8 00:00:02 59kN
1...f6xg5 2.Re1-d1 Ra8-c8 3.Qd5-d6+ Kg6-h5
-+ (-2.26) Depth: 9 00:00:03 107kN
1...f6xg5 2.Qd5-d3+ Kg6-h6 3.Re1-d1
-/+ (-0.99) Depth: 10 00:00:07 290kN
1...Bd7-c6 2.Qd5xd8
-/+ (-1.20) Depth: 10 00:00:08 316kN
1...Bd7-c6 2.Qd5xd8 Ra8xd8 3.Bg5-e3 b7-b6 4.Kh1-g1 Rd8-d7 5.Re1-e2 Rd7-e7 6.Kg1-f2 Kg6-f5 7.b2-b3
-/+ (-1.20) Depth: 11 00:00:09 353kN
1...Bd7-c6 2.Qd5xd8 Ra8xd8 3.Bg5-e3 b7-b6 4.Kh1-g1 Rd8-d7 5.Re1-e2 Rd7-e7 6.Kg1-f2 Kg6-f5 7.b2-b3
-/+ (-1.20) Depth: 12 00:00:10 416kN
1...Bd7-c6 2.Qd5xd8 Ra8xd8 3.Bg5-e3 b7-b6 4.Kh1-g1 Rd8-d7 5.Re1-e2 Kg6-f5 6.Kg1-f2 h7-h5 7.b2-b3 Bc6-e4
-/+ (-1.18) Depth: 13 00:00:13 548kN
1...Bd7-c6 2.Qd5xd8 Ra8xd8 3.Bg5-e3 b7-b6 4.Kh1-g1 Rd8-d7 5.Re1-e2 Kg6-f5 6.Kg1-f2 h7-h5 7.b2-b3 Bc6-e4 8.Re2-d2
-/+ (-1.17) Depth: 14 00:00:21 882kN

[Ovyron]

This is strange since i get the following analysis in the exact same computer with 256 MB hash tables

Rybka 3 on Aquarium is known to show different results than in other GUIs.

[Dann Corbit]

This is strange since i get the following analysis in the exact same computer with 256 MB hash tables

Rybka 3 on Aquarium is known to show different results than in other GUIs.

I get different results every time I analyze a position. I guess that this is for two reasons:
1. SMP causes some non-determinism
2. Permanent hash will remember effort from before.

[Mike S.]

What's so special about 24...Bc6? I think 24...fxg5 wins too, and the difference between these moves is small. This is not a good test position.

[Dr.Ex]

What's so special about 24...Bc6? I think 24...fxg5 wins too, and the difference between these moves is small. This is not a good test position.

fxg5 is a sure draw. I gave the variations.

[Eelco de Groot]

In the variation with 24... fxg5, the direct 26. h4 is hard to find for Naum but 26. Qd3+ seems to draw also:


[d]2rq4/pp1b2pp/6k1/3Q2p1/8/8/PP4PP/3R3K w - -

Engine: Naum 3.1 (256 MB)
by Aleksandar Naumov


11/29 0:00 -1.84 26.Qd6+ Kh5 27.g4+ Kxg4 28.Qd4+ Kh5
29.Qxd7 Qxd7 30.Rxd7 Rc1+ 31.Kg2 (770.189) 1026

12/27 0:01 -2.00 26.Qd6+ Kh5 27.g4+ Kxg4 28.Qd4+ Kh5
29.Qxd7 Qxd7 30.Rxd7 Rc1+ 31.Kg2 Rc2+ (1.260.862) 1008

13/40 0:04 -1.79 26.Qd6+ Kh5 27.g4+ Kxg4 28.h3+ Kh4
29.Qd4+ Kh5 30.Qxd7 Qxd7 31.Rxd7 Rc1+
32.Kg2 (4.065.471) 974

14/33 0:05 -1.83 26.Qd6+ Kh5 27.g4+ Kxg4 28.h3+ Kh4
29.Qd4+ Kh5 30.Qxd7 Qxd7 31.Rxd7 Rc1+
32.Kg2 Rc2+ (5.807.903) 996

15/32 0:10 -1.77 26.Qd6+ Kh5 27.g4+ Kxg4 28.h3+ Kh4
29.Qd4+ Kh5 30.Qxd7 Qxd7 31.Rxd7 Rc1+
32.Kg2 Rc2+ 33.Kf3 (10.282.959) 1007

16/35 0:16 -1.90 26.Qd6+ Kh5 27.g4+ Kxg4 28.h3+ Kh4
29.Qd4+ Kh5 30.Qxd7 Qxd7 31.Rxd7 Rc1+
32.Kg2 Rc2+ 33.Kf3 Rxb2 (16.923.245) 1017 TB:1

17/42 0:31 -1.90 26.Qd6+ Kh5 27.g4+ Kxg4 28.h3+ Kh4
29.Qd4+ Kh5 30.Qxd7 Qxd7 31.Rxd7 Rc1+
32.Kg2 Rc2+ 33.Kf3 Rxb2 34.Rxg7 (32.537.275) 1018 TB:2

18/41 1:03 -1.97 26.Qd6+ Kh5 27.g4+ Kxg4 28.h3+ Kh4
29.Qd4+ Kh5 30.Qxd7 Qxd7 31.Rxd7 Rc1+
32.Kg2 Rc2+ 33.Kf3 Rxb2 34.Rxg7 h6 (64.833.789) 1022 TB:7

18/46 5:54 -0.77 26.Qd3+ Kh6 27.h4 gxh4 28.Qe3+ Kg6
29.Qd3+ Kf6 30.Qd6+ Kf7 31.Qxd7+ Qxd7
32.Rxd7+ Kf6 33.Rxb7 Ra8 34.b4 a6 (338.608.246) 956 TB:83

19/43 6:21 -0.77 26.Qd3+ Kh6 27.h4 gxh4 28.Qe3+ Kg6
29.Qd3+ Kf6 30.Qd6+ Kf7 31.Qxd7+ Qxd7
32.Rxd7+ Kf6 33.Rxb7 Ra8 34.b4 a6
35.a4 (366.038.656) 960 TB:102

20/41 7:40 -0.78 26.Qd3+ Kh6 27.h4 Rc6 28.Qxd7 Qxd7
29.hxg5+ Kxg5 30.Rxd7 Rh6+ 31.Kg1 Rb6
32.b3 Kg6 33.Kh2 Ra6 34.Rxb7 Rxa2
35.Kg3 Kf6 (444.640.996) 965 TB:473

21/41 12:01 -0.77 26.Qd3+ Kh6 27.h4 Rc6 28.Qxd7 Qxd7
29.hxg5+ Kxg5 30.Rxd7 Rh6+ 31.Kg1 Rb6
32.Rxg7+ Kh6 33.Rd7 Rxb2 34.a4 Kg6
35.Kh2 Ra2 36.Rxb7 (700.668.706) 971
TB:1.310

22/42 14:52 -0.77 26.Qd3+ Kh6 27.h4 Rc6 28.Qxd7 Qxd7
29.hxg5+ Kxg5 30.Rxd7 Rh6+ 31.Kg1 Rb6
32.Rxg7+ Kh6 33.Rd7 Rxb2 34.a4 Kg6
35.Kh2 Ra2 36.Rd6+ Kg5 (870.408.940) 975
TB:1.983

23/46 21:36 -0.79 26.Qd3+ Kh6 27.h4 Rc6 28.Qxd7 Qxd7
29.hxg5+ Kxg5 30.Rxd7 Rb6 31.Rxg7+ Kh6
32.Rd7 Rxb2 33.a4 Ra2 34.Rd4 Kg5
35.Kh2 b6 36.Kg3 Ra3+ 37.Kf2 (1.277.406.394) 985
TB:3.707

24/59 53:12 -0.79 26.Qd3+ Kh6 27.h4 Rc6 28.hxg5+ Qxg5
29.Qxd7 Qh4+ 30.Kg1 Qf4 31.Kh1 Rc1
32.b3 Rxd1+ 33.Qxd1 b6 34.Qd7 Qh4+
35.Kg1 Qe1+ 36.Kh2 Qe5+ 37.Kh3 Qe3+ (3.058.688.061) 958
TB:7.250

best move: Qd5-d3 time: 78:04.406 min n/s: 958.207 nodes: 3.769.207.582 TB: 7.250

After 26. h4 there is also no result with a 0.00 score, but as there is also no progress I would declare it a draw:

After 24... fxg5 25. Rd1 Rc8 26. h4

[d]2rq4/pp1b2pp/6k1/3Q2p1/7P/8/PP4P1/3R3K b - -

Engine: Naum 3.1 (256 MB)
by Aleksandar Naumov

16/46 0:21 0.00 26...Rc7 27.Qd3+ Kh6 28.hxg5+ Kxg5
29.Qg3+ Kh6 30.Qh2+ Kg6 31.Qd6+ Kh5
32.Qe5+ g5 33.Rd6 Rc1+ 34.Kh2 (21.733.888) 1024 TB:1

16/46 0:37 +1.02 26...gxh4 27.Qxd7 Qxd7 28.Rxd7 b5
29.Rxa7 Rc2 30.b4 Re2 31.a4 bxa4
32.Rxa4 Rb2 33.b5 Rxb5 34.Rg4+ (37.762.217) 1020 TB:52

17/41 1:02 +0.82 26...gxh4 27.Qd3+ Kh6 28.Qe3+ Qg5
29.Qxg5+ Kxg5 30.Rxd7 Kf6 31.Rxb7 Rc2
32.Rxa7 Rxb2 33.Kg1 Re2 34.a4 Re7 (63.210.500) 1014 TB:176

18/39 2:06 +0.68 26...gxh4 27.Qd3+ Kf6 28.Qd6+ Kf7
29.Qxd7+ Qxd7 30.Rxd7+ Kf6 31.Rxb7 Ra8
32.Kh2 h5 33.b4 a6 34.a4 g5 35.Rb6+ (126.689.154) 1000 TB:388

18/39 3:07 +0.71 26...Rc6 27.Qd3+ Kh6 28.Qxd7 Qxd7
29.hxg5+ Kxg5 30.Rxd7 Rb6 31.Rxg7+ Kh6
32.Rd7 Rxb2 33.a4 Kg6 34.Kh2 h6
35.Kg3 (183.125.804) 975 TB:612

19/39 3:35 +0.71 26...Rc6 27.Qd3+ Kh6 28.Qxd7 Qxd7
29.hxg5+ Kxg5 30.Rxd7 Rb6 31.Rxg7+ Kh6
32.Rd7 Rxb2 33.a4 Kg6 34.Kh2 h6
35.Kg3 Kf5 (211.720.988) 981 TB:640

20/39 5:33 +0.74 26...Rc6 27.Qd3+ Kh6 28.Qxd7 Qxd7
29.hxg5+ Kxg5 30.Rxd7 Rb6 31.Rxg7+ Kh6
32.Rd7 Rxb2 33.a4 Kg6 34.Kh2 Ra2
35.Rxb7 Rxa4 36.Kg3 (321.996.301) 965 TB:838

21/51 13:09 +0.79 26...Rc6 27.Qd3+ Kh6 28.hxg5+ Qxg5
29.Qxd7 Qh4+ 30.Kg1 Qf4 31.Qd3 Rc1
32.b3 Rxd1+ 33.Qxd1 Qe3+ 34.Kh1 Qe7
35.Qd4 b5 36.Kh2 b4 (766.931.901) 971
TB:1.814

22/49 22:56 +0.80 26...Rc6 27.Qd3+ Kh6 28.hxg5+ Qxg5
29.Qxd7 Qh4+ 30.Kg1 Qf4 31.Qd3 b6
32.b3 Rf6 33.Qh3+ Kg6 34.Qd7 Qe3+
35.Kh2 Qe5+ 36.Kg1 a5 37.Qg4+ (1.365.378.667) 992
TB:2.629

23/52 40:10 +0.82 26...Rc6 27.Qd3+ Kh6 28.hxg5+ Qxg5
29.Qxd7 Qh4+ 30.Kg1 Qf4 31.Qd3 b6
32.b3 Rf6 33.Rd2 a5 34.g3 Qf5
35.Qe3+ g5 36.Kg2 Re6 37.Qf3 Qxf3+ (2.321.935.043) 963
TB:4.447

best move: Rc8-c6 time: 42:57.250 min n/s: 963.146 nodes: 2.458.249.113 TB: 4.447

Three plies deeper after 26... Rc6 27. Qd3 Kh6

[d]3q4/pp1b2pp/2r4k/6p1/7P/3Q4/PP4P1/3R3K w - -

Engine: Naum 3.1 (256 MB)
by Aleksandar Naumov

16/32 0:11 -0.80 28.Qxd7 Qxd7 29.hxg5+ Kxg5 30.Rxd7 Rb6
31.Rxg7+ Kh6 32.Rc7 Rxb2 33.a4 Kg6
34.Kh2 h5 35.Kg3 a5 (12.903.197) 1085 TB:51

17/34 0:26 -0.82 28.Qxd7 Qxd7 29.hxg5+ Kxg5 30.Rxd7 Rb6
31.Rxg7+ Kh6 32.Rc7 Rxb2 33.a4 Kg6
34.Kh2 a5 35.Kg3 h5 36.Kf3 (28.451.845) 1092 TB:121

18/35 1:01 -0.79 28.Qxd7 Qxd7 29.hxg5+ Kxg5 30.Rxd7 Rb6
31.Rxg7+ Kh6 32.Rc7 Rxb2 33.a4 Kg6
34.Kh2 a5 35.Kg3 h5 36.Kf3 b6 (66.690.246) 1093 TB:221

19/37 1:49 -0.89 28.Qxd7 Qxd7 29.hxg5+ Kxg5 30.Rxd7 Rb6
31.b3 Kf6 32.Kh2 h5 33.Kg3 g5 34.Kf3 Ra6
35.Rxb7 Rxa2 36.g3 a6 37.Rc7 (119.226.545) 1093 TB:572

20/37 7:29 -0.89 28.Qxd7 Qxd7 29.hxg5+ Kxg5 30.Rxd7 Rb6
31.b3 Kf6 32.Kh2 h5 33.Kg3 g5 34.Kf3 a6
35.g3 Rb4 36.Rd8 a5 37.Rd7 b5 (492.348.081) 1095 TB:936

21/41 10:12 -0.86 28.Qxd7 Qxd7 29.hxg5+ Kxg5 30.Rxd7 Rb6
31.Rxg7+ Kh6 32.Rd7 Rxb2 33.a4 Ra2
34.Rd6+ Kg5 35.Rd5+ Kg4 36.Rd4+ Kf5
37.Kh2 b6 38.Kg3 (671.911.776) 1097
TB:2.429

22/41 16:13 -0.83 28.Qxd7 Qxd7 29.hxg5+ Kxg5 30.Rxd7 Rb6
31.Rxg7+ Kh6 32.Rd7 Rxb2 33.a4 Ra2
34.Rd4 b6 35.Kh2 Ra3 36.Rg4 a6 37.g3 Ra2+
38.Kh3 Kh5 (1.067.064.677) 1096
TB:5.415

23/43 27:38 -0.82 28.Qxd7 Qxd7 29.hxg5+ Kxg5 30.Rxd7 Rb6
31.Rxg7+ Kh6 32.Rd7 Rxb2 33.a4 Ra2
34.Rd4 Kg5 35.Kh2 Ra1 36.Kg3 Ra3+
37.Kf2 b6 38.Rd5+ Kg6 39.Rd6+ (1.817.058.964) 1095
TB:11.702

24/45 49:03 -0.85 28.Qxd7 Qxd7 29.hxg5+ Kxg5 30.Rxd7 Rb6
31.Rxg7+ Kh6 32.Rd7 Rxb2 33.a4 Ra2
34.Rd4 Kg5 35.Kh2 Ra1 36.Kg3 Ra3+
37.Kf2 b6 38.Rd5+ Kg6 39.Rd4 h6 (3.205.233.169) 1088
TB:22.197

25/46 118:52 -0.82 28.Qxd7 Qxd7 29.hxg5+ Kxg5 30.Rxd7 Rb6
31.Rxg7+ Kh6 32.Rd7 Rxb2 33.a4 Ra2
34.Rd4 Kg5 35.Kh2 Ra3 36.Rd5+ Kg4
37.Rd4+ Kf5 38.Rd7 h5 39.Rxb7 Rxa4
40.Kg3 (7.489.449.385) 1050
TB:47.228

26/48 203:38 -0.80 28.Qxd7 Qxd7 29.hxg5+ Kxg5 30.Rxd7 Rb6
31.Rxg7+ Kh6 32.Rd7 Rxb2 33.a4 Ra2
34.Rd4 Kg5 35.Kh2 Ra3 36.Rd5+ Kg4
37.Rd4+ Kf5 38.Rd7 h5 39.Rxb7 Rxa4
40.Kg3 a6 (12.550.629.728) 1027
TB:114.330

best move: Qd3xd7 time: 207:09.000 min n/s: 1.027.199 nodes: 12.550.629.728 TB: 114.330

An Ancalagon testversion I tried was too optimistic about 26... fxg5. :

r2q4/pp1b2pp/5pk1/3Q2B1/8/8/PP4PP/4R2K b - -

Engine: Ancalagon 1.08 Testversion 003 (64 MB)
by Tord Romstad

19.01 34:02 +3.00 24...fxg5 25.Rd1 Rc8 26.Qd3+ Kh6
27.h4 Rc7 28.hxg5+ Kxg5 29.Qg3+ Kh6
30.Qh2+ Kg6 31.Qf4 h6 32.Qe4+ Kh5
33.Qf3+ Bg4 34.Rxd8 Bxf3 35.gxf3 Rc1+
36.Kh2 Rc2+ 37.Kg3 Rxb2 (990.182.739) 484

27. h4 Rc7 leads to a draw and the engine can see that after 27. h4 but at this point that was too deep to see for this version.