test postion: Pinkovetsky-Fischer, WC/27 final ICCF 2011

[jdart]

[D] 6rk/2r2pp1/p2q3p/1pb1pPPN/3pQ2P/3P4/PPP3R1/5RK1 w - - 0 30

bm f6. Stockfish 4 takes about 80 seconds for this on my quad.

If that is too easy for you, try this position from earlier in the game:

[D] 3q1r1k/2r2pp1/p6p/1pbppP1N/3pP1PP/3P1Q2/PPP4R/5RK1 w - - 0 27

bm g5

The game:

[pgn]
[Event "WC27/final"]
[Site "ICCF"]
[Date "2011.06.10"]
[Round "?"]
[White "Pinkovetsky, Semen Usherovich"]
[Black "Fischer, Wolfgang"]
[Result "1-0"]
[ECO "B30"]
[WhiteElo "2549"]
[BlackElo "2520"]
[PlyCount "73"]
[EventDate "2011.??.??"]

1. e4 c5 2. Nf3 Nc6 3. Nc3 Nf6 4. Bb5 Qc7 5. d3 e6 6. O-O Nd4 7. Nxd4 cxd4 8.
Ne2 a6 9. Ba4 b5 10. Bb3 Bc5 11. Bf4 d6 12. Bg5 Qe7 13. Ng3 h6 14. Bxf6 Qxf6
15. Nh5 Qg6 16. f4 O-O 17. Qf3 e5 18. f5 Qg5 19. Rf2 Kh8 20. Bd5 Ra7 21. Raf1
Rc7 22. g3 Qd8 23. h4 Bb7 24. Bxb7 Rxb7 25. Rh2 d5 26. g4 Rc7 27. g5 dxe4 28.
Qxe4 Rg8 29. Rg2 Qd6 30. f6 g6 31. Ng7 h5 32. Nxh5 Qc6 33. Qe1 Qd6 34. Ng3 Bb4
35. Qf2 e4 36. h5 Rf8 37. Nxe4 1-0
[/pgn]

[zullil]

If that is too easy for you, try this position from earlier in the game:

[D] 3q1r1k/2r2pp1/p6p/1pbppP1N/3pP1PP/3P1Q2/PPP4R/5RK1 w - - 0 27

bm g5

Stockfish gets it, but takes a while:

Code: Select all

info depth 36 seldepth 52 score cp 141 lowerbound nodes 1702984183 nps 16620803 time 102461 multipv 1 pv g4g5 d5e4 f3e4 c5d6 f5f6 g7g6 h5g3 f8g8 h4h5 d6f8 f1f2 d8d6 h5g6 g8g6 h2h5 g6g5 h5g5 h6g5 f2h2

[Dan Astrachan]

On my computer:

SF4 (latest build): 27 seconds
Fire 3.0: 73 seconds
Critter 1.6a: 9 seconds

[zullil]

On my computer:

SF4 (latest build): 27 seconds
Fire 3.0: 73 seconds
Critter 1.6a: 9 seconds

Which position?

The latest SF solves the first in 2 seconds here:

Code: Select all

info depth 20 seldepth 35 score cp 188 nodes 11276385 nps 5137305 time 2195 multipv 1 pv f5f6 g7g6 h5g7 d6c6 e4e2 c5f8 g5h6 f8d6 h4h5 g6h5 e2h5 c6d5 a2a3 h8h7 h5f5 h7h8 f5g4 h8h7 g4h3 e5e4 g7f5 e4e3 g2g7 h7h8

[Dan Astrachan]

On my computer:

SF4 (latest build): 27 seconds
Fire 3.0: 73 seconds
Critter 1.6a: 9 seconds

Which position?

The latest SF solves the first in 2 seconds here:

Code: Select all

info depth 20 seldepth 35 score cp 188 nodes 11276385 nps 5137305 time 2195 multipv 1 pv f5f6 g7g6 h5g7 d6c6 e4e2 c5f8 g5h6 f8d6 h4h5 g6h5 e2h5 c6d5 a2a3 h8h7 h5f5 h7h8 f5g4 h8h7 g4h3 e5e4 g7f5 e4e3 g2g7 h7h8

The first position posted by the OP

[Eelco de Groot]

If that is too easy for you, try this position from earlier in the game:

[D] 3q1r1k/2r2pp1/p6p/1pbppP1N/3pP1PP/3P1Q2/PPP4R/5RK1 w - - 0 27

bm g5

Stockfish gets it, but takes a while:

Code: Select all

info depth 36 seldepth 52 score cp 141 lowerbound nodes 1702984183 nps 16620803 time 102461 multipv 1 pv g4g5 d5e4 f3e4 c5d6 f5f6 g7g6 h5g3 f8g8 h4h5 d6f8 f1f2 d8d6 h5g6 g8g6 h2h5 g6g5 h5g5 h6g5 f2h2

I think that was probably an unlucky run or probably the tactics of the position is solved better with smaller hash. With Stockfish 20140216 which codebase is functionally the same as the the latest dev version and no further changes except for a small change in ProbCut:


3q1r1k/2r2pp1/p6p/1pbppP1N/3pP1PP/3P1Q2/PPP4R/5RK1 w - -

Engine: Sf20140216_004 MP (Q6700, 32 bit, 4 threads, 512 MB)
Codechange: ProbCut only if Nullmove or Verification search fails low
by Tord Romstad, Marco Costalba and Joona Kiiski

.
.
.
.

29/45 0:52 +1.20 27.f6 g6 28.Ng7 Qd7 29.h5 g5 30.Nf5 Kh7
31.exd5 Rd8 32.d6 Bxd6 33.Qe4 Bf8
34.Re1 Kh8 35.Qxe5 Rc5 36.Qe4 Qc7
37.Ree2 Rc6 38.Rhf2 Re6 39.Qf3 Rxf6
40.Re4 (128.427.793) 2462

30/45 0:56 +1.12-- 27.f6 g6 28.Ng7 Qd7 29.h5 g5 30.Nf5 Kh7
31.exd5 Rd8 32.d6 Bxd6 33.Qe4 Bf8
34.Re1 Kg8 35.Qxe5 Rc6 36.Rhe2 Re6
37.Qh2 Rxf6 38.Qe5 Re6 39.Qg3 Rxe2
40.Rxe2 (140.411.085) 2473

30/45 1:02 +1.04-- 27.f6 g6 28.Ng7 Qd7 29.h5 g5 (154.815.914) 2496

30/45 1:08 +0.92-- 27.f6 g6 28.Ng7 Qd7 29.h5 g5 30.Nf5 Kh7
31.a3 Rd8 32.Re2 Bf8 33.exd5 Qxd5
34.Qxd5 Rxd5 35.Rfe1 Bd6 36.b4 Rb7
37.Rf1 Bc7 38.Ne7 Rd6 39.Rxe5 Rxf6
40.Rxf6 (174.312.068) 2526

30/45 1:15 +1.09 27.f6 g6 28.Ng7 Qd7 29.h5 g5 30.Nf5 Kh7
31.Re2 Rd8 32.Kg2 Qc6 33.exd5 Qxd5
34.Qxd5 Rxd5 35.Rfe1 Bb6 36.b4 Rd8
37.Kf3 Rc6 38.Rxe5 Rxf6 39.Re8 Rxe8
40.Rxe8 (191.664.708) 2549

31/45 1:18 +1.17++ 27.f6 g6 28.Ng7 Qd7 29.h5 g5 30.Nf5 Kh7
31.Re2 Rd8 32.Kg2 Qc6 33.exd5 Qxd5 (201.421.337) 2559

31/45 1:23 +1.01-- 27.f6 g6 28.Ng7 Qd7 29.h5 g5 30.Nf5 Kh7
31.Re2 Rd8 32.Kg2 Bf8 33.exd5 Qxd5
34.Qxd5 Rxd5 35.Rfe1 Bd6 36.Kf3 Rd7
37.Ke4 Bc7 38.b4 Rd8 39.Ne7 R5d6
40.Kxe5 (215.341.479) 2564

31/45 2:10 +1.25++ 27.g5 dxe4 28.Qxe4 Rg8 29.Rg2 Qd6
30.f6 g6 31.Ng7 Qc6 32.Qe2 Kh7
33.Qxe5 (345.143.745) 2648

31/45 2:31 +1.43++ 27.g5 Rd7 28.gxh6 gxh6 29.exd5 Rg8+
30.Rg2 Rxg2+ 31.Qxg2 Bf8 32.f6 Qa8
33.Qg3 Rd6 34.Rf2 Rxd5 (403.280.414) 2668

31/52 2:58 +1.41 27.g5 dxe4 28.Qxe4 Bd6 29.f6 g6
30.Ng3 Rg8 31.h5 Bf8 32.Rff2 Qc8
33.Qxe5 gxh5 34.Ne4 Rg6 35.Kf1 Rc6
36.Rhg2 h4 37.Qh2 h3 38.Rg3 Qe6
39.b3 Qc8 40.Kg1 (477.305.030) 2675

32/52 3:16 +1.50++ 27.g5 dxe4 28.Qxe4 Bd6 29.f6 g6
30.Ng3 Rg8 31.h5 Bf8 32.Rff2 Qc8
33.Qxe5 gxh5 34.Ne4 Rg6 35.Kf1 Rc6
36.Rhg2 h4 37.Qh2 h3 38.Rg3 Qe6
39.b3 Qc8 40.Kg1 (528.226.202) 2682

32/52 3:34 +1.58++ 27.g5 dxe4 28.Qxe4 Bd6 29.f6 g6
30.Ng3 Rg8 31.h5 b4 32.Rff2 gxh5
33.Rxh5 (574.696.453) 2684

32/52 3:49 +1.50 27.g5 dxe4 28.Qxe4 Bd6 29.f6 g6
30.Ng3 Rg8 31.h5 Bf8 32.Rff2 Qc8
33.Qxe5 gxh5 34.Ne4 Rg6 35.Kf1 Rc6
36.Rhg2 h4 37.Qh2 h3 38.Rg3 Qe6
39.b3 Bd6 40.Nxd6 (616.929.178) 2688

33/52 3:55 +1.50 27.g5 dxe4 28.Qxe4 Bd6 29.f6 g6
30.Ng3 Rg8 31.h5 Bf8 32.Rff2 Qc8
33.Qxe5 gxh5 34.Ne4 Rg6 35.Kf1 Rc6
36.Rhg2 h4 37.Qh2 h3 38.Rg3 Qe6
39.b3 Bd6 40.Nxd6 (633.729.632) 2686

best move: g4-g5 time: 3:57.262 min n/s: 2.686.763 nodes: 633.729.632

[zullil]

Another try was somewhat faster:

Code: Select all

info depth 33 seldepth 56 score cp 130 lowerbound nodes 967174949 nps 16252309 time 59510 multipv 1 pv g4g5 d5e4 f3e4 c5d6 f5f6 g7g6 h5g3 f8g8 h4h5 g6h5

[Eelco de Groot]

In case this position is evaluated for a testsuite, Scid Serpent 20140216_009 finds that f6 is possible in the second position as well

27. g5 is still best, but 27. f6 may be good enough for a sizable advantage, if you search the position deep enough.


[D]3q1r1k/2r2pp1/p6p/1pbppP1N/3pP1PP/3P1Q2/PPP4R/5RK1 w - -

Engine: Scid Serpent 20140216_009 MP (Q6700, 4 threads, 512 MB)
by Tord Romstad, Marco Costalba and Joona Kiiski

.
.
.
.

24/41 0:17 +0.91 27.exd5 e4 28.Qxe4 Bd6 29.Nf4 Re8
30.Qf3 Re3 31.Qf2 Rc8 32.b3 Qf6
33.a4 bxa4 34.bxa4 Rc3 35.Rg2 Be5
36.Nh5 Qd8 37.f6 g6 38.Nf4 Qxf6
39.g5 Qg7 (33.596.666) 1961

25/41 0:18 +0.99++ 27.exd5 e4 28.Qxe4 Bd6 29.Nf4 Re8
30.Qf3 Kg8 31.g5 hxg5 32.hxg5 Qxg5+
33.Rg2 Qh4 34.Qh5 Qxh5 35.Nxh5 Rec8
36.Rff2 Be5 37.Nxg7 Bxg7 38.f6 (38.115.925) 2018

25/41 0:21 +1.07++ 27.f6 g6 28.Ng7 Qc8 29.h5 dxe4
30.Qxe4 g5 31.Nf5 Re8 32.Nxh6 Bf8
33.Nf5 Rxc2 34.Ne7 Bxe7 35.Rxc2 Qxc2
36.fxe7 Rxe7 37.Qf5 (43.588.093) 2030

.
.
.
.

33/51 4:36 +1.08 27.f6 g6 28.Ng7 Qd7 29.h5 g5 30.Nf5 Kh7
31.exd5 Rd8 32.d6 Bxd6 33.Qe4 Bf8
34.Ne7+ Kh8 35.Qxe5 Re8 36.Rg2 Bxe7
37.fxe7+ Kg8 38.Qf6 Qxe7 39.Qxd4 Rd7
40.Qb6 (679.498.004) 2455

34/51 7:17 +1.16++ 27.g5 dxe4 28.Qxe4 Rg8 29.Rg2 f6
30.gxf6 gxf6 31.Rg6 Rxg6+ 32.fxg6 Be7
33.g7+ Kg8 34.Qg6 (1.110.988.107) 2538

.
.
.
.

37/58 39:12 +2.04 27.g5 dxe4 28.Qxe4 Bd6 29.f6 g6
30.Ng7 Qa8 31.h5 Qxe4 32.dxe4 Kg8
33.hxg6 fxg6 34.Ne6 h5 35.Nxc7 Bxc7
36.Kf2 Kf7 37.Ke2 Rc8 38.Kd3 Bb6
39.a3 a5 40.a4 (6.148.188.141) 2612

38/60 51:09 +2.12++ 27.g5 dxe4 28.Qxe4 Bd6 29.f6 g6
30.Ng7 Qa8 (8.076.657.112) 2631

38/64 60:41 +2.20++ 27.g5 dxe4 28.Qxe4 (9.595.441.822) 2635

38/70 75:17 +2.32++ 27.g5 dxe4 28.Qxe4 Bd6 29.f6 g6 (11.948.411.624) 2644

38/70 96:00 +2.49 27.g5 dxe4 28.Qxe4 Bd6 29.f6 g6
30.Ng7 Qa8 31.h5 Qxe4 32.dxe4 Kg8
33.hxg6 fxg6 34.Ne6 h5 35.Nxc7 Bxc7
36.Kf2 Ba5 37.Kf3 Rf7 38.Rd1 Rc7
39.Rd3 Bb6 40.Rb3 (15.300.299.956) 2656

39/70 96:50 +2.49 27.g5 dxe4 28.Qxe4 Bd6 29.f6 g6
30.Ng7 Qa8 31.h5 Qxe4 32.dxe4 Kg8
33.hxg6 fxg6 34.Ne6 h5 35.Nxc7 Bxc7
36.Kf2 Ba5 37.Kf3 Rf7 38.Rd1 Rc7
39.Rd3 Bb6 40.Rb3 (15.432.297.846) 2655

best move: g4-g5 time: 96:51.225 min n/s: 2.655.939 nodes: 15.432.297.846

Now to two best moves:

3q1r1k/2r2pp1/p6p/1pbppP1N/3pP1PP/3P1Q2/PPP4R/5RK1 w - -

Engine: Scid Serpent 20140216_009 MP (512 MB)
by Tord Romstad, Marco Costalba and Joona Kiiski

38 19:23 +2.49 27.g5 dxe4 28.Qxe4 Bd6 29.f6 g6
30.Ng7 Qa8 31.h5 Qxe4 32.dxe4 Kg8
33.hxg6 fxg6 34.Ne6 h5 35.Nxc7 Bxc7
36.Kf2 Ba5 37.Kf3 Rf7 38.Rd1 Rc7
39.Rd3 Bb6 40.Rb3 (3.003.227.869) 2580

38 19:23 +1.33 27.f6 g6 28.Ng7 dxe4 29.Qxe4 Qa8
30.Qxe5 Qc6 31.Qf4 Kh7 32.h5 g5
33.Qf5+ Kh8 34.Re2 Rfc8 35.Ne6 Rd7
36.Qxc5 Qxc5 37.Nxc5 Rxc5 38.Rff2 Rd8
39.Kf1 Kg8 40.a3 (3.003.227.869) 2580
______________________________________________

39 65:20 +2.49 27.g5 dxe4 28.Qxe4 Bd6 29.f6 g6
30.Ng7 Qa8 31.h5 Qxe4 32.dxe4 Kg8
33.hxg6 fxg6 34.Ne6 h5 (10.484.465.747) 2674

39 65:20 +1.33 27.f6 g6 28.Ng7 dxe4 29.Qxe4 Qa8
30.Qxe5 Qc6 31.Qf4 Kh7 32.h5 g5
33.Qf5+ Kh8 34.Re2 Rfc8 35.Ne6 Rd7
36.Qxc5 Qxc5 37.Nxc5 Rxc5 38.Rff2 Rd8
39.Kf1 Kg8 40.a3 (10.484.465.747) 2674
_______________________________________________

44 932:22 +2.54 27.g5 dxe4 28.Qxe4 Bd6 29.f6 g6
30.Ng7 Qa8 31.h5 Qxe4 32.dxe4 Kg8
33.hxg6 fxg6 34.Ne6 h5 35.Nxc7 Bxc7
36.Kg2 Rc8 37.Kf3 b4 38.Rd1 Kf7
39.c3 bxc3 40.bxc3 (155.861.214.449) 2786

44 932:22 +1.65 27.f6 g6 28.Ng7 dxe4 29.Qxe4 Qa8
30.Qxe5 Qc6 31.Qf4 Kh7 32.h5 g5
33.Qf5+ Kh8 34.Qf3 Rfc8 35.Re2 Qxf3
36.Rxf3 Rc6 37.Kg2 Kh7 38.Rff2 b4
39.Re5 Bd6 40.Rd5 (155.861.214.449) 2786

Eelco

[zullil]

In case this position is evaluated for a testsuite, Scid Serpent 20140216_009 finds that f6 is possible in the second position as well

27. g5 is still best, but 27. f6 may be good enough for a sizable advantage, if you search the position deep enough.

Here are the top three, at depth 45 using the latest Stockfish:

Code: Select all

info depth 45 seldepth 74 score cp 269 nodes 248619170380 nps 20478410 time 12140550 multipv 1 pv g4g5 d5e4 f3e4 f7f6 g5h6 g7h6 h2g2 f8g8 f1f2 g8g2 f2g2 c7f7 g2g6 d8f8 e4h1 c5d6 h1d5 b5b4 d5c6 h8h7 c6a6 h7h8 g1f1 h8h7 a6c6 d6e7 c6g2 h7h8 g2d5 e7d6 d5c6 d6e7 c6e6 e7d8 e6d5 d8e7 d5g2 e7d6

info depth 45 seldepth 74 score cp 222 nodes 248619170380 nps 20478410 time 12140550 multipv 2 pv h5g3 f7f6 e4d5 c7f7 h2g2 c5e7 g3e4 a6a5 d5d6 e7d6 f3h3 f7c7 g4g5 d6e7 f1f3 f8f7 g5h6 g7h6 f3g3 d8f8 g3g6 a5a4 c2c3 c7c8 c3d4 e5d4 h3g4 f7h7 h4h5 b5b4 g4f4 c8d8 g1h2 b4b3 a2b3 a4b3 g2g1 d8e8 h2h1 e8c8 f4g4 f8f7 g4g3 f7f8 g1g2 c8a8 g2g1 a8c8

info depth 45 seldepth 74 score cp 173 nodes 248619170380 nps 20478410 time 12140550 multipv 3 pv f5f6 g7g6 h5g7 d5e4 f3e4 d8a8 e4e5 a8c6 e5f4 h8h7 h4h5 g6g5 f4f5 h7g8 h2e2 f8d8 f5f2 d8c8 f2g2 c6g2 g1g2 c7c6 a2a3 g8f8 e2e4 c5a7 f1f2 c8d8 g7f5 c6f6 f2e2 f6c6 g2f2 a7c5 f2e1 a6a5 e4e5 b5b4 a3b4 a5b4 e1f1 d8c8 f1f2 c6f6 f2g2 f6a6 g2f3 a6c6 e5d5 c6f6 d5d7 f8g8 f3g3 g8f8 g3g2 f6e6 g2f3 b4b3

[zullil]

In case this position is evaluated for a testsuite, Scid Serpent 20140216_009 finds that f6 is possible in the second position as well

27. g5 is still best, but 27. f6 may be good enough for a sizable advantage, if you search the position deep enough.

Here are the top three, at depth 45 using the latest Stockfish:

Code: Select all

info depth 45 seldepth 74 score cp 269 nodes 248619170380 nps 20478410 time 12140550 multipv 1 pv g4g5 d5e4 f3e4 f7f6 g5h6 g7h6 h2g2 f8g8 f1f2 g8g2 f2g2 c7f7 g2g6 d8f8 e4h1 c5d6 h1d5 b5b4 d5c6 h8h7 c6a6 h7h8 g1f1 h8h7 a6c6 d6e7 c6g2 h7h8 g2d5 e7d6 d5c6 d6e7 c6e6 e7d8 e6d5 d8e7 d5g2 e7d6

info depth 45 seldepth 74 score cp 222 nodes 248619170380 nps 20478410 time 12140550 multipv 2 pv h5g3 f7f6 e4d5 c7f7 h2g2 c5e7 g3e4 a6a5 d5d6 e7d6 f3h3 f7c7 g4g5 d6e7 f1f3 f8f7 g5h6 g7h6 f3g3 d8f8 g3g6 a5a4 c2c3 c7c8 c3d4 e5d4 h3g4 f7h7 h4h5 b5b4 g4f4 c8d8 g1h2 b4b3 a2b3 a4b3 g2g1 d8e8 h2h1 e8c8 f4g4 f8f7 g4g3 f7f8 g1g2 c8a8 g2g1 a8c8

info depth 45 seldepth 74 score cp 173 nodes 248619170380 nps 20478410 time 12140550 multipv 3 pv f5f6 g7g6 h5g7 d5e4 f3e4 d8a8 e4e5 a8c6 e5f4 h8h7 h4h5 g6g5 f4f5 h7g8 h2e2 f8d8 f5f2 d8c8 f2g2 c6g2 g1g2 c7c6 a2a3 g8f8 e2e4 c5a7 f1f2 c8d8 g7f5 c6f6 f2e2 f6c6 g2f2 a7c5 f2e1 a6a5 e4e5 b5b4 a3b4 a5b4 e1f1 d8c8 f1f2 c6f6 f2g2 f6a6 g2f3 a6c6 e5d5 c6f6 d5d7 f8g8 f3g3 g8f8 g3g2 f6e6 g2f3 b4b3

One more step:

Code: Select all

info depth 46 seldepth 75 score cp 316 nodes 329875982035 nps 20615774 time 16001144 multipv 1 pv g4g5 d5e4 f3e4 f7f6 g5h6 g7h6 h2g2 f8g8 f1f2 g8g2 f2g2 d8f8 g2g6 c7f7 e4g2 c5d6 g2c6 b5b4 c6a6 d6e7 g1g2 e7d6 a6a4 d6e7 a4c6 e7d8 c6d5 h8h7 g2g1 h7h8 d5e6 d8e7 g1g2 h8h7

info depth 46 seldepth 75 score cp 222 nodes 329875982035 nps 20615774 time 16001144 multipv 2 pv h5g3 f7f6 e4d5 c7f7 h2g2 c5e7 g3e4 a6a5 d5d6 e7d6 f3h3 f7c7 g4g5 d6e7 f1f3 f8f7 g5h6 g7h6 f3g3 d8f8 g3g6 a5a4 c2c3 c7c8 c3d4 e5d4 h3g4 f7h7 h4h5 b5b4 g4f4 c8d8 g1h2 b4b3 a2b3 a4b3 g2g1 d8e8 h2h1 e8c8 f4g4 f8f7 g4g3 f7f8 g1g2 c8a8 g2g1

info depth 46 seldepth 75 score cp 173 nodes 329875982035 nps 20615774 time 16001144 multipv 3 pv f5f6 g7g6 h5g7 d5e4 f3e4 d8a8 e4e5 a8c6 e5f4 h8h7 h4h5 g6g5 f4f5 h7g8 h2e2 f8d8 f5f2 d8c8 f2g2 c6g2 g1g2 c7c6 a2a3 g8f8 e2e4 c5a7 f1f2 c8d8 g7f5 c6f6 f2e2 f6c6 g2f2 a7c5 f2e1 a6a5 e4e5 b5b4 a3b4 a5b4 e1f1 d8c8 f1f2 c6f6 f2g2 f6a6 g2f3 a6c6 e5d5 c6f6 d5d7 f8g8 f3g3 g8f8 g3g2 f6e6 g2f3

Seems that SF feels g5 is clearly best here.