Stockfish entombs a bishop

[modolief]

Here's a position from a 15 minute game played by a IM Chessexplained on ICC. He's analyzing after the game with Stockfish. At 32:45 he points out that Stockfish's choice of move for the position, 17..h5 for black, is obviously bad.

http://www.youtube.com/watch?v=uIkhZInjkbE

[/img]

[kgburcham]

[D] r3n1k1/1p2pp1p/1p2bbp1/6N1/3P3P/P1N1P3/1P3PP1/4KB1R b K -

anybody get any program to play ...h5?
I tried Stockfish and Houdini, both show -1 point for ...h5.
Not sure how this guy got his program to play this move.
kgburcham

[syzygy]

Maybe if you add a black pawn on d5.

[kgburcham]

Maybe if you add a black pawn on d5.

whoops
thanks Ronald

[Isaac]

Stockfish 250214 with 5 pieces syzygy would play Ne6 at depth 28 but h5 at prior depths.

Code: Select all

info depth 27 seldepth 40 score cp -98 nodes 99978982 nps 1822073 tbhits 2 time 54871 multipv 1 pv h7h5 g5e6 f7e6 f1d3 g8f7 g2g3 e8d6 e1e2 a8c8 h1a1 c8a8 f2f3 a8d8 a1c1 d8a8 f3f4
info depth 28 currmove h7h5 currmovenumber 1
info depth 28 currmove e8d6 currmovenumber 2
info depth 28 currmove a8d8 currmovenumber 6
info depth 28 currmove a8a4 currmovenumber 8
info depth 28 currmove e8c7 currmovenumber 9
info depth 28 currmove a8c8 currmovenumber 11
info depth 28 currmove f6e5 currmovenumber 14
info depth 28 currmove a8a6 currmovenumber 15
info depth 28 currmove g8g7 currmovenumber 16
info depth 28 currmove a8a7 currmovenumber 17
info depth 28 currmove h7h6 currmovenumber 18
info depth 28 currmove g8f8 currmovenumber 19
info depth 28 currmove e6c8 currmovenumber 20
info depth 28 currmove f6h8 currmovenumber 21
info depth 28 currmove a8b8 currmovenumber 22
info depth 28 currmove e6h3 currmovenumber 23
info depth 28 currmove g8h8 currmovenumber 24
info depth 28 currmove f6g5 currmovenumber 25
info depth 28 seldepth 40 score cp -95 nodes 144491853 nps 1834840 tbhits 2 time 78749 multipv 1 pv h7h5 g5e6 f7e6 f1d3 g8f7 g2g3 e8d6 e1e2 a8c8 h1a1 c8a8 f2f3 a8d8 a1c1 d8a8 e2d2 a8c8 c1b1 c8a8 b1h1 f6g7 h1c1 a8c8
info depth 29 currmove h7h5 currmovenumber 1
info depth 29 currmove e8d6 currmovenumber 2
info depth 29 seldepth 41 score cp -87 lowerbound nodes 240291333 nps 1824951 tbhits 21 time 131670 multipv 1 pv e8d6 g5e6 f7e6
info depth 29 currmove e8d6 currmovenumber 1
info depth 29 currmove a8a5 currmovenumber 2
info depth 29 currmove e6d7 currmovenumber 4
info depth 29 currmove g8g7 currmovenumber 5
info depth 29 currmove e6f5 currmovenumber 6
info depth 29 currmove h7h6 currmovenumber 9
info depth 29 currmove h7h5 currmovenumber 14
info depth 29 currmove e6h3 currmovenumber 15
info depth 29 currmove a8a6 currmovenumber 16
info depth 29 currmove a8a4 currmovenumber 17
info depth 29 currmove e6c8 currmovenumber 18
info depth 29 currmove a8a7 currmovenumber 19
info depth 29 currmove a8b8 currmovenumber 20
info depth 29 currmove g8f8 currmovenumber 21
info depth 29 currmove g8h8 currmovenumber 22
info depth 29 currmove f6h8 currmovenumber 23
info depth 29 currmove f6e5 currmovenumber 24
info depth 29 currmove f6g5 currmovenumber 25
info depth 29 currmove f6d4 currmovenumber 26
info depth 29 currmove a8a3 currmovenumber 27
info depth 29 seldepth 41 score cp -88 nodes 308813150 nps 1850820 tbhits 21 time 166852 multipv 1 pv e8d6 g5e6 f7e6 f1d3 a8a5 f2f4 d6f5 e1f2 f6h4 g2g3 h4f6 g3g4 f5d6 g4g5 f6g7 d3e2 d6c4 e2c4 d5c4 f2f3 a5a8 c3e4 a8a4 h1h2 a4a8 f3g4 a8a4 g4g3
info depth 30 currmove e8d6 currmovenumber 1

Stockfish thinks that white has an advantage of about 98 cp at depth 30, the latest evaluation score returned before I closed the program.

P.S.:Here's the FEN I used: r3n1k1/1p2pp1p/1p2bbp1/3p2N1/3P3P/P1N1P3/1P3PP1/4KB1R.

Edit: And here is the non-syzygy Stockfish 250214 analysis with similar given time:

Code: Select all

info depth 30 seldepth 41 score cp -98 nodes 249278943 nps 1882700 tbhits 0 time 132405 multipv 1 pv h7h5 g5e6 f7e6 g2g3 g8f7 f1d3 e8d6 e1d2 a8c8 f2f3 f6g7 h1b1 c8a8 b1g1 a8h8 g1f1 h8a8 g3g4 b6b5 d3b5 e6e5 d4e5
info depth 31 currmove h7h5 currmovenumber 1
info depth 31 currmove e8d6 currmovenumber 2
info depth 31 seldepth 42 score cp -98 nodes 401875779 nps 1925651 tbhits 0 time 208696 multipv 1 pv h7h5 g5e6 f7e6 g2g3 g8f7 f1d3 e8d6 e1d2 a8c8 f2f3 f6g7 h1b1 c8a8 b1g1 a8h8
info depth 32 currmove h7h5 currmovenumber 1
stop
info depth 32 seldepth 45 score cp -98 nodes 428127713 nps 1927956 tbhits 0 time 222063 multipv 1 pv h7h5 g5e6 f7e6 g2g3 g8f7 f1d3 e8d6 e1d2 a8c8 f2f3 f6g7 h1b1 c8a8 b1g1 a8h8
info nodes 428127713 tbhits 0 time 222063
bestmove h7h5 ponder g5e6

[modolief]

I don't understand that time notation; how long does it take Stockfish to prefer 17..Ne6?

[Isaac]

I don't understand that time notation; how long does it take Stockfish to prefer 17..Ne6?

The time is in ms.
On my "old" hardware (Intel duo core E6300 at 1.86 Ghz) it took 131 seconds which is slightly more than 2 minutes. This means less than 35 seconds for modern pc's I guess.
Note that the non-syzygy version of SF would not play Ne6 even after 3 minutes on my slow hardware.

[kgburcham]

I don't understand that time notation; how long does it take Stockfish to prefer 17..Ne6?

[D] r3n1k1/1p2pp1p/1p2bbp1/3p2N1/3P3P/P1N1P3/1P3PP1/4KB1R b K -

Engine: Stockfish 260214 64 SSE4.2 (8192 MB)
by Tord Romstad, Marco Costalba and Joona
12/16 0:00 +0.99 1...Bxg5 2.hxg5 Nd6 3.Be2 Kg7 4.f4 Rd8 5.Kf2 Nc4 6.Bxc4 dxc4 7.Kf3 Bf5 (271.835) 5663
13/18 0:00 +0.97 1...Bxg5 2.hxg5 Nd6 3.Be2 Kg7 4.Kd2 Rc8 5.Bf3 Nc4+ 6.Kc2 Bf5+ 7.Kb3 e6 8.Be2 Na5+ 9.Kb4 Nc6+ 10.Kb5 (312.444) 6509
14/19 0:00 +0.91 1...Rd8 2.Nxe6 fxe6 3.f4 h5 4.Bd3 Kf7 5.Kf2 Nd6 6.g3 Bg7 7.Rc1 Ra8 8.Kf3 Rc8 9.g4 hxg4+ 10.Kxg4 (690.845) 7272
15/21 0:00 +0.84 1...Rd8 2.Nxe6 fxe6 3.f4 h5 4.Bd3 Kf7 5.Kf2 Nd6 6.g3 Bg7 7.Kf3 Rc8 8.Re1 Ra8 9.Rc1 Rc8 10.Rh1 Ra8 11.Nb5 Rc8 12.Nxd6+ exd6 13.g4 e5 14.Kg3 exd4 (1.091.611) 7687
16/22 0:00 +0.81 1...Nd6 2.Nxe6 fxe6 3.g3 Ra5 4.f4 b5 5.Kf2 Nc4 6.Bxc4 bxc4 7.Kf3 Ra8 8.h5 g5 9.Kg4 gxf4 10.Kxf4 (2.205.280) 9384
17/23 0:00 +0.79 1...Nd6 2.Nxe6 fxe6 3.Bd3 Ra5 4.f4 Nf5 5.Bxf5 exf5 6.Kf2 e6 7.Kf3 Kf7 8.h5 g5 9.g4 gxf4 10.Kxf4 fxg4 11.Kxg4 (2.919.168) 9795
18/23 0:00 +0.86 1...Nd6 2.Nxe6 fxe6 3.Bd3 Ra5 4.f4 Nf5 5.Bxf5 exf5 6.Kf2 Kf7 7.h5 g5 8.Ne2 gxf4 9.Nxf4 e5 10.dxe5 Bxe5 (3.651.847) 9712
19/29 0:00 +0.85 1...Nd6 2.Nxe6 fxe6 3.g3 e5 4.Nxd5 exd4 5.Nxf6+ exf6 6.exd4 Re8+ 7.Be2 Kg7 8.Kf1 Re4 9.Bf3 Rxd4 10.Ke2 b5 11.Rd1 Rxd1 12.Kxd1 f5 13.Bd5 Ne4 14.Ke2 (4.678.378) 9338
20/29 0:00 +0.85 1...Nd6 2.Nxe6 fxe6 3.g3 e5 4.Nxd5 exd4 5.Nxf6+ exf6 6.exd4 Re8+ 7.Be2 Kg7 8.Kf1 Re4 9.Bf3 Rxd4 10.Ke2 b5 11.Rd1 Rxd1 12.Kxd1 f5 13.Bd5 Ne4 14.Ke2 (5.586.840) 9389
21/34 0:01 +0.88 1...Nd6 2.Nxe6 fxe6 3.g3 Ra5 4.f4 h5 5.Bd3 Kf7 6.Kd2 Ra8 7.Rc1 Rc8 8.Na4 Rxc1 9.Kxc1 b5 10.Nc5 Bg7 11.Kd2 b6 (10.792.797) 10298
22/34 0:01 +0.84 1...Nd6 2.Nxe6 fxe6 3.g3 Ra5 4.f4 b5 5.g4 b4 6.axb4 Ra1+ 7.Kf2 Rc1 8.g5 Rc2+ 9.Kf3 Bg7 10.h5 Kf7 11.h6 Bf8 12.Bd3 Rxb2 13.Rb1 Rxb1 14.Bxb1 (14.775.380) 10737
23/34 0:02 +0.82 1...Nd6 2.Nxe6 fxe6 3.g3 Ra5 4.f4 b5 5.Kf2 Nc4 6.Bh3 Kf7 7.Rb1 Nd2 8.Rc1 Nc4 9.Rc2 Bg7 10.Kf3 h5 11.g4 b4 12.axb4 hxg4+ 13.Bxg4 Ra1 14.Bh3 (22.068.031) 10775
24/34 0:02 +0.86 1...Nd6 2.Nxe6 fxe6 3.g3 Ra5 4.f4 b5 5.Kf2 Nc4 6.Bh3 Kf7 7.Rb1 Nd2 8.Rc1 Nc4 9.Rc2 Nd6 10.Kf3 Bg7 11.Bf1 b4 12.axb4 Ra1 13.Bd3 Re1 (27.804.423) 10781
25/34 0:03 +0.94-- 1...Nd6 2.Nxe6 fxe6 3.g3 Ra5 4.f4 b5 5.Kf2 Nc4 6.Bh3 Kf7 7.Rb1 Nd2 8.Rc1 Nc4 9.Rc2 Nd6 10.Kf3 Bg7 11.Bf1 b4 12.axb4 Ra1 13.Bd3 Re1 14.h5 (36.001.736) 11128
25/34 0:04 +0.83 1...Nd6 2.Nxe6 fxe6 3.Bd3 Ra5 4.f4 Nf5 5.Bxf5 exf5 6.h5 Kf7 7.hxg6+ Kxg6 8.Kf2 e6 9.Rc1 Bh4+ 10.Kf3 Bd8 11.Rh1 h5 12.Ne2 Ra4 13.Nc3 Ra5 (46.923.175) 11287
26/35 0:05 +0.91-- 1...Nd6 2.Nxe6 fxe6 3.g3 Kf7 4.Bd3 Nc4 5.Bxc4 dxc4 6.f4 Bg7 7.Kd2 h5 8.Ne4 (64.499.596) 11625
26/37 0:06 +0.89 1...Nd6 2.Nxe6 fxe6 3.g3 Ra5 4.f4 b5 5.Kf2 Nc4 6.Bh3 Kf7 7.Rb1 Bg7 8.Bg4 h5 9.Bf3 Kf6 10.Ke2 Nd6 11.Kd3 Kf7 12.g4 hxg4 13.Bxg4 Ra8 14.Rg1 (80.949.111) 11745
27/37 0:10 +0.97-- 1...Nd6 2.Kd2 Bxg5 3.hxg5 Ne4+ 4.Nxe4 dxe4 5.f4 exf3 6.gxf3 Ra5 7.f4 Bd5 8.Rh4 (123.753.251) 12146
27/37 0:11 +0.81++ 1...Nd6 2.Kd2 Ra5 3.Nxe6 fxe6 4.f4 Nf5 5.Be2 Nxh4 6.g4 h6 7.Bd3 Kf7 8.Rf1 Bg7 9.g5 hxg5 10.fxg5+ Nf5 (141.442.025) 12198
27/37 0:11 +0.90 1...Nd6 2.Kd2 Ra5 3.Nxe6 fxe6 4.f4 Nf5 5.Be2 Nxh4 6.g3 Nf5 7.g4 Nd6 8.g5 Bg7 9.Bg4 Nc4+ 10.Ke2 Kf7 11.Nd1 Rb5 12.Rxh7 Kg8 13.Rh2 Kf7 14.Kf3 Rb3 (145.090.895) 12185
28/42 0:16 +0.94 1...h5 2.Nxe6 fxe6 3.Be2 Rd8 4.f4 Kf7 5.Bd3 Nd6 6.g3 Bg7 7.Kf2 Rc8 8.Rd1 Nf5 9.Na4 Rc6 10.Kf3 Nd6 11.Rh1 Bf6 12.Nc3 Rc8 13.Re1 Ra8 14.Rc1 Rc8 (199.313.407) 12395
29/45 0:24 +1.03-- 1...h5 2.Nxe6 fxe6 3.Be2 Rd8 4.f4 Kf7 5.g4 hxg4 6.Bxg4 Ng7 7.h5 gxh5 8.Bxh5+ Nxh5 9.Rxh5 Kg6 10.Rh1 Kf5 11.Rh7 Rg8 12.Kf2 Kg6 13.Rh1 (309.075.155) 12736
29/48 0:33 +0.97 1...Nd6 2.Kd2 h5 3.Nxe6 fxe6 4.Bd3 Kf7 5.g3 Rc8 6.Rc1 Rc6 7.Rb1 Bg7 8.Ke2 Rc8 9.Rd1 Ra8 10.Kd2 Rc8 11.Rg1 Bf6 12.Ke2 b5 13.Rb1 Bg7 14.Bxb5 Nxb5 (440.712.680) 13021
30/48 0:46 +1.02 1...Bxg5 2.hxg5 Kg7 3.Bd3 Nd6 4.Ke2 Ra5 5.g3 Bd7 6.f3 e6 7.f4 Ra8 8.Kf3 Bc6 9.Rh2 Bd7 10.Be2 Ba4 11.Nxa4 Rxa4 12.Bd3 Ra8 13.Rh1 Ra4 14.g4 b5 (605.314.358) 12930
31/48 1:02 +1.04 1...Bxg5 2.hxg5 Ra5 3.Kd2 Nd6 4.Bd3 Kg7 5.Ra1 Ra8 6.f3 h6 7.gxh6+ Kxh6 8.Nb5 Nxb5 9.Bxb5 Kg7 10.Rh1 Ra5 11.Bd3 Ra8 12.Rc1 Rc8 13.Rxc8 Bxc8 14.e4 dxe4 (807.795.438) 12992
32/48 1:08 +1.04 1...Bxg5 2.hxg5 Ra5 3.Kd2 Nd6 4.Bd3 Kg7 5.Ra1 Ra8 6.f3 h6 7.gxh6+ Kxh6 8.Nb5 Nxb5 9.Bxb5 Kg7 10.Rh1 Ra5 11.Bd3 Ra8 12.Rc1 Rc8 13.Rxc8 Bxc8 14.e4 dxe4 (886.378.967) 13008
33/48 1:18 +1.03 1...Bxg5 2.hxg5 Ra5 3.Kd2 Nd6 4.Bd3 Kg7 5.Ra1 Ra8 6.f3 h6 7.gxh6+ Kxh6 8.Nb5 Nxb5 9.Bxb5 Kg7 10.Rh1 Ra5 11.Be2 Ra8 12.Rc1 Rc8 13.Rxc8 Bxc8 14.e4 dxe4 (1.024.832.169) 13008
34/50 1:54 +0.94++ 1...Bxg5 2.hxg5 Kg7 3.Bd3 Nd6 4.Ke2 Ra5 5.Rc1 Ra8 6.Rh1 (1.502.598.646) 13135
34/50 2:24 +0.94 1...Bxg5 2.hxg5 Kg7 3.Bd3 Nd6 4.Ke2 Ra5 5.Ra1 Ra8 6.f3 h5 7.gxh6+ Kxh6 8.Rh1+ Kg7 9.Nb5 Nxb5 10.Bxb5 Rc8 11.Bd3 Bd7 12.Kd2 f5 13.g3 Rh8 14.Rxh8 Kxh8 (1.908.594.191) 13196
35/53 3:46 +0.95 1...Nd6 2.Kd2 Bxg5 3.hxg5 f6 4.gxf6 exf6 5.Be2 Rc8 6.g4 Kg7 7.f3 Rh8 8.e4 dxe4 9.fxe4 Bc4 10.Rh2 Rd8 11.Ke3 h6 12.Bd3 b5 13.Rh1 Re8 14.Rg1 Kf7 (3.026.034.502) 13357
36/55 6:04 +1.04-- 1...Nd6 2.Nxe6 fxe6 3.Bd3 Ra5 4.f4 Nf5 5.Kd2 h5 6.g3 Nxg3 7.Rg1 Nf5 8.Rxg6+ Kh7 9.Rg1 Kh6 10.Nb5 Bxh4 11.Ke2 Ng3+ 12.Kf3 Ne4 13.Rg2 (4.866.305.079) 13335
36/55 7:18 +1.05 1...Bxg5 2.hxg5 Nd6 3.Bd3 Ra5 4.Kd2 Bf5 5.f3 Bd7 6.Re1 e6 7.a4 Kg7 8.b3 Ra8 9.Rh1 Rh8 10.e4 dxe4 11.Nxe4 Nxe4+ 12.fxe4 Ra8 13.Ke3 Ra5 14.Kf4 e5+ (5.862.332.372) 13365
37/58 8:44 +1.04 1...Bxg5 2.hxg5 Kg7 3.Bd3 f6 4.gxf6+ exf6 5.f3 h5 6.Bc2 Nd6 7.Bb3 Rc8 8.Kd2 Nc4+ 9.Bxc4 Rxc4 10.g3 Rc8 11.e4 Rd8 12.Ke3 Kf7 13.Na4 b5 14.Nc5 b6 (6.999.746.994) 13353
38/58 9:03 +0.95++ 1...Bxg5 2.hxg5 Kg7 3.Bd3 Nd6 4.f3 Ra5 5.Kd2 Bd7 6.Rb1 e6 7.Rh1 b5 8.Kc2 Nc4 9.e4 Ne3+ 10.Kb3 Nxg2 11.exd5 Nf4 12.dxe6 Nxe6 13.d5 Nxg5 14.Kb4 Ra6 (7.251.603.817) 13349
38/58 9:35 +1.04 1...Bxg5 2.hxg5 Kg7 3.Bd3 f6 4.gxf6+ exf6 5.f3 h5 6.Bc2 Nd6 7.Bb3 Rc8 8.Kd2 Nc4+ 9.Bxc4 Rxc4 10.g3 Rc8 11.e4 dxe4 12.fxe4 Kf7 13.Ke3 Ke7 14.Rc1 Kd6 (7.679.847.663) 13347
39/58 11:11 +0.95++ 1...Bxg5 2.hxg5 Kg7 3.Bd3 f6 4.gxf6+ exf6 5.f3 h5 6.Bc2 Nd6 7.Bb3 Rc8 8.Kd2 Nc4+ 9.Bxc4 Rxc4 10.Kd3 Rc8 11.b3 Ra8 12.a4 Kf7 13.Nb5 Rd8 (8.991.075.825) 13386
39/58 12:00 +0.95 1...Bxg5 2.hxg5 Kg7 3.Bd3 f6 4.gxf6+ exf6 5.f3 h5 6.Bc2 Nd6 7.Bb3 Rc8 8.Kd2 Nc4+ 9.Bxc4 Rxc4 10.Kd3 f5 11.g3 Rc8 12.Rc1 Kf7 13.Rc2 Kg7 14.Rh2 Kf6 (9.643.349.257) 13383

[modolief]

So between the times of 16 seconds and 33 seconds, the most cutting edge Stockfish is considering a move that an IM would discard out of hand? How is this an effective analysis engine unless you're prepared to really sit there for a long time ... on every move?

[Isaac]

So between the times of 16 seconds and 33 seconds, the most cutting edge Stockfish is considering a move that an IM would discard out of hand? How is this an effective analysis engine unless you're prepared to really sit there for a long time ... on every move?

I'd rather trust a +3000 elo engine than an IM, in such a position at least.
Stockfish thinks that white has an around +100 cp advantage with either h5, Nd6 or Bxg5. All possible moves for black are "bad" according to SF. It is not really clear to me as to why h5 would not be the best for black and if it is not the best move for black, it is not much worse than the best move; at least according to Stockfish.