What engine see the Queen Sacrifice of Caruana game vs Naka?

[royb]

Looking forward to Aristides! Is it due within the week or is it further out than that?

[MikeB]

Looking forward to Aristides! Is it due within the week or is it further out than that?

< 24 hours

[beram]

The real question is, what engine can see the sacrifice in less than 3 minutes using a regular PC, not a super pc with more than 2 processors etc......
http://en.chessbase.com/post/caruana-br ... -challenge

I think I have a regular PC at least not super with Fritzmark 23,1
Houdini 5 finds it within 3minutes: 1.Nf5
= (0.23 !) Depth: 29/62 00:02:40 1607mN
SF 18-12-17 within 6minutes: 1.Nf5
= (0.20 !) Depth: 38/59 00:05:51 2699mN

Code: Select all

New game
r1bqk2r/5p2/p2p1b1p/3Np3/1P1NP1BP/8/1PP2B2/2KR3R w kq - 0 1

Analysis by Stockfish 181216 64 POPCNT:

1.Nc6 Bxg4 2.Nxd8 Bxd8 3.Rd2 Rc8 4.b3 Bf3 5.Re1 Rg8 6.c4 h5 7.Rd3 Bg2 8.Nc3 Be7 9.Rg1 Rg4 10.Kd2 Rb8 11.Rg3 Rxg3 12.Bxg3 Bf3 13.Ke3 Bg4 14.Nd5 f5 15.exf5 Bxf5 16.Ra1 Bc8 17.c5 Bd8 18.c6 Rb5 19.Rd1 Be6 20.c7 Rxd5 21.Rxd5 Bxc7 22.Rd2 Bxb3 
  =  (0.09)   Depth: 37/59   00:04:15  1950mN
1.Nc6 Bxg4 
  =  (0.02 --)   Depth: 38/59   00:04:30  2070mN
1.Nc6 
  =  (0.09 !)   Depth: 38/59   00:04:58  2281mN
1.Nf5 
  =  (0.20 !)   Depth: 38/59   00:05:51  2699mN
1.Nf5 
  +/=  (0.35 !)   Depth: 38/59   00:06:17  2900mN
1.Nf5 
  +/=  (0.56 !)   Depth: 38/59   00:07:04  3269mN

(Bram Mourik, AMD X6 1090T 3200 Mhz (Fritzmark 23,1 6x) 19.12.2016)

90: New game
r1bqk2r/5p2/p2p1b1p/3Np3/1P1NP1BP/8/1PP2B2/2KR3R w kq - 0 1

Analysis by Houdini 5 x64-popc:

1.Nc6 Bxg4 2.Nxd8 Bxd8 3.Rd2 Rc8 4.b3 Rc6 5.c4 Bf3 6.Re1 Rf8 7.Re3 Bg4 8.Rg3 h5 9.Rg1 Bf3 10.Nc3 f5 11.exf5 Rxf5 12.Kb2 Be7 13.Rg8+ Rf8 14.Rg7 Rf7 15.Rg8+ 
  =  (0.00 --)   Depth: 29/62   00:02:38  1583mN
1.Nf5 
  =  (0.23 !)   Depth: 29/62   00:02:40  1607mN
1.Nf5 
  +/=  (0.36 !)   Depth: 29/62   00:02:46  1666mN
1.Nf5 
  +/=  (0.55 !)   Depth: 29/62   00:03:02  1829mN
1.Nf5 
  +/-  (0.77 !)   Depth: 29/62   00:03:29  2102mN

(Bram Mourik, AMD X6 1090T 3200 Mhz (Fritzmark 23,1 6x) 19.12.2016)

[/u]

[chessmobile]

What i want to know is this move Nf5! impressive as it may be, is it winning? The consensus is that Nakamura blundered with Rb8? and that Bxf5 was needed, is the + 1 evaluation enough to win?

[MikeB]

What i want to know is this move Nf5! impressive as it may be, is it winning? The consensus is that Nakamura blundered with Rb8? and that Bxf5 was needed, is the + 1 evaluation enough to win?

I know I said Rb8 was a blunder at first glance , but I believe now it may be best - sure white is winning, but a lot moves to be played. Nf5 is a very good move - best move for White for sure. Nc6 is a little more drawish than Nf5.

[Eelco de Groot]

What i want to know is this move Nf5! impressive as it may be, is it winning? The consensus is that Nakamura blundered with Rb8? and that Bxf5 was needed, is the + 1 evaluation enough to win?

The pgn gives 23. c4 but Stockfish plays Kb1 first, then c4. But it may just be a transposition. Some deeper analysis, I just left the computer running for a while. I don't think Nc6 is a serious alternative for 21. Nf5 but I did not investigate it further (the game comments suggest a bit that Caruana rejected it, but it is a bit unclear English )

After 21. Nf5 (best) Bxf5 (best) 22. Bxf5 Rb8

[d]1r1qk2r/5p2/p2p1b1p/3NpB2/1P2P2P/8/1PP2B2/2KR3R w k -

.
.
.
44/83 411:54 +3.18 23.Kb1 Be7 24.c4 O-O 25.c5 dxc5
26.Nxe7+ Qxe7 27.Bxc5 Qf6 28.Rd6 Qg7
29.Rg1 Qxg1+ 30.Bxg1 Rxb4 31.Rd2 Rfb8
32.Be3 h5 33.Bc5 R4b7 34.Rg2+ Kh8
35.Bd6 Re8 36.Rg5 (250.260.049.045) 10126

45/83 501:03 +3.27++ 23.Kb1 (304.888.686.312) 10141

45/83 581:34 +3.18-- 23.Kb1 Be7 (354.329.670.871) 10154

45/83 595:24 +3.06-- 23.Kb1 Be7 (362.644.112.713) 10151

45/83 647:33 +3.16++ 23.Kb1 (394.539.330.348) 10154

45/83 828:41 +3.39++ 23.Kb1 (504.913.606.465) 10154

45/83 856:17 +3.21-- 23.Kb1 Be7 (521.363.453.049) 10147

45/91 949:05 +2.81-- 23.Kb1 Be7 (577.464.263.733) 10140

45/112 1035:59+2.99 23.Kb1 Be7 24.c4 a5 25.b5 a4 26.b6 Kf8
27.Nxe7 Qxe7 28.c5 Qe8 29.cxd6 Qb5
30.d7 a3 31.d8Q+ Rxd8 32.Rxd8+ Ke7
33.Rd7+ Kf6 34.Rd6+ Ke7 35.Rd2 Rd8
36.Rhd1 (630.912.245.174) 10149

46/119 1036:25+3.12 23.Kb1 Be7 (631.196.193.577) 10150

best move: Kc1-b1 time: 1036:25.922 min n/s: 10.150.146 nodes: 631.196.193.577

The 46 plies deep short PV is bogus (from ending the analysis), the last real PV is 45 plies deep. To judge the advantage you have to keep in mind that White is a queen for knight and bishop down, so already +1 is some serious amount of compensation when about one light piece down (but immaterial, King Safety points). If this is fully winning, I don't know.

[Dann Corbit]

Not sure if it is fully winning, but I am sure that it is very beautiful.

[Jouni]

Chessbase has now long article

http://en.chessbase.com/post/workshop-t ... -sacrifice

[Dicaste]

What i want to know is this move Nf5! impressive as it may be, is it winning? The consensus is that Nakamura blundered with Rb8? and that Bxf5 was needed, is the + 1 evaluation enough to win?

The pgn gives 23. c4 but Stockfish plays Kb1 first, then c4. But it may just be a transposition. Some deeper analysis, I just left the computer running for a while. I don't think Nc6 is a serious alternative for 21. Nf5 but I did not investigate it further (the game comments suggest a bit that Caruana rejected it, but it is a bit unclear English )

After 21. Nf5 (best) Bxf5 (best) 22. Bxf5 Rb8

[d]1r1qk2r/5p2/p2p1b1p/3NpB2/1P2P2P/8/1PP2B2/2KR3R w k -

.
.
.
44/83 411:54 +3.18 23.Kb1 Be7 24.c4 O-O 25.c5 dxc5
26.Nxe7+ Qxe7 27.Bxc5 Qf6 28.Rd6 Qg7
29.Rg1 Qxg1+ 30.Bxg1 Rxb4 31.Rd2 Rfb8
32.Be3 h5 33.Bc5 R4b7 34.Rg2+ Kh8
35.Bd6 Re8 36.Rg5 (250.260.049.045) 10126

45/83 501:03 +3.27++ 23.Kb1 (304.888.686.312) 10141

45/83 581:34 +3.18-- 23.Kb1 Be7 (354.329.670.871) 10154

45/83 595:24 +3.06-- 23.Kb1 Be7 (362.644.112.713) 10151

45/83 647:33 +3.16++ 23.Kb1 (394.539.330.348) 10154

45/83 828:41 +3.39++ 23.Kb1 (504.913.606.465) 10154

45/83 856:17 +3.21-- 23.Kb1 Be7 (521.363.453.049) 10147

45/91 949:05 +2.81-- 23.Kb1 Be7 (577.464.263.733) 10140

45/112 1035:59+2.99 23.Kb1 Be7 24.c4 a5 25.b5 a4 26.b6 Kf8
27.Nxe7 Qxe7 28.c5 Qe8 29.cxd6 Qb5
30.d7 a3 31.d8Q+ Rxd8 32.Rxd8+ Ke7
33.Rd7+ Kf6 34.Rd6+ Ke7 35.Rd2 Rd8
36.Rhd1 (630.912.245.174) 10149

46/119 1036:25+3.12 23.Kb1 Be7 (631.196.193.577) 10150

best move: Kc1-b1 time: 1036:25.922 min n/s: 10.150.146 nodes: 631.196.193.577

The 46 plies deep short PV is bogus (from ending the analysis), the last real PV is 45 plies deep. To judge the advantage you have to keep in mind that White is a queen for knight and bishop down, so already +1 is some serious amount of compensation when about one light piece down (but immaterial, King Safety points). If this is fully winning, I don't know.

Code: Select all

FEN: 1r1qk2r/5p2/p2p1b1p/3NpB2/1P2P2P/8/1PP2B2/2KR3R w k - 0 1

M8R x64:
 8+	00:00	 16k	1.310k	-0,24	Be3 Kf8 Rhg1
 8	00:00	 17k	1.314k	-0,19	Be3 Bxh4 Rdg1 h5 c3 Rf8 Kc2 Rb5
 9-	00:00	 24k	1.378k	-0,20	Be3 h5
 9	00:00	 28k	1.409k	-0,22	Be3 Bxh4 Rdg1 h5 c3 Kf8 Kb1 Bf6 Kc2 Rb5
 10+	00:00	 31k	1.442k	-0,15	Be3 Bg7 Rdg1
 10	00:00	 55k	1.422k	-0,16	Be3 Be7 Rdg1 Bxh4 b3 h5 Kb2 Rf8 c3 Rb5
 11-	00:00	 58k	1.409k	-0,17	Be3 Bxh4
 11+	00:00	 66k	1.383k	-0,09	Rdg1 Kf8 Be3
 11-	00:00	 79k	1.406k	-0,13	Rdg1 Rb5
 11-	00:00	 153k	1.169k	-0,60	Rdg1 Rb5
 11	00:00	 161k	1.181k	-0,49	Be3 Bxh4 Rdg1 h5 Kb1 Rb5 c3 Rxd5 exd5 Bf6 Bd3 a5 b5
 11+	00:00	 187k	1.165k	-0,26	c3 Rg8 Be3
 11	00:00	 210k	1.233k	-0,14	c3 Rg8 Be3 Bxh4 Bxh6 Bf2 Kc2 Rg2 Bh3 Rg6 Bf5 Rg8
 11	00:00	 212k	1.239k	-0,04	Kb1 Kf8 c3 Rb5 Be3 Bxh4 Rdg1 Rxd5 exd5 Bf6 Bd3 Ke7 Bxa6
 12-	00:00	 222k	1.221k	-0,05	Kb1 Kf8
 12+	00:00	 280k	1.241k	-0,01	Kb1 Kf8 Rd3
 12-	00:00	 298k	1.232k	-0,06	Kb1 0-0
 12	00:00	 328k	1.255k	-0,01	Kb1 Kf8 Rd3 Rg8 Rh2 Kg7 Ra3 Kh8 c3 Rg7 Rxa6 Be7
 13-	00:00	 341k	1.264k	-0,02	Kb1 Kf8
 13-	00:00	 346k	1.264k	-0,09	Kb1 Kf8
 13-	00:00	 420k	1.324k	-0,19	Kb1 h5
 13+	00:00	 452k	1.340k	-0,11	Kb1 h5 c4
 13	00:00	 486k	1.353k	 0,00	Kb1 h5 c4 Rg8 Rhg1 Rxg1 Rxg1 Kf8 Be3 Bg7 Bg5 f6 Be3 Rb7
 14-	00:00	 528k	1.383k	-0,01	Kb1 Rg8
 14-	00:00	 562k	1.395k	-0,08	Kb1 Kf8
 14-	00:00	 570k	1.388k	-0,18	Kb1 Kf8
 14+	00:00	 597k	1.400k	-0,10	Kb1 Kf8 c4
 14	00:00	 670k	1.436k	 0,00	Kb1 Kf8 c4 Be7 Rdg1 Qe8 b3 Bd8 Be3 h5 Kb2 Qc6 Rg2 Qb7 Rgg1 Qc6
 15-	00:00	 729k	1.453k	-0,01	Kb1 Kf8
 15	00:00	 754k	1.429k	 0,00	Kb1 Rg8 Be3 Bxh4 Bxh6 Bf2 Bg7 Bh4 Bh6
 16	00:00	 800k	1.425k	 0,00	Kb1 Rg8 Be3 Bxh4 Bxh6 Bf2 Bg7 Bh4 Bh6
 17	00:00	 907k	1.440k	 0,00	Kb1 Rg8 Be3 Bxh4 Bxh6 Bf2 Bg7 Bh4 Bh6
 18	00:00	 1.156k	1.453k	 0,00	Kb1 Rg8 Be3 Bxh4 Bxh6 Bf2 Bg7 Bh4 Bh6
 19+	00:01	 1.820k	1.605k	+0,06	Kb1 Rg8 Be3
 19+	00:01	 2.654k	1.662k	+0,12	Kb1 Rg8 Be3
 19+	00:02	 3.673k	1.714k	+0,21	Kb1 Rg8 Be3
 19+	00:03	 6.390k	1.723k	+0,34	Kb1 Rg8 c4
 19	00:04	 7.632k	1.719k	+0,36	Kb1 0-0 Rd3 Kh8 Ra3 Rg8 Rxa6 Rb7 Be3 Bg7 c4 Bf8 Rh2 Rg3 b5 h5 Bb6 Qb8 Bf2 Rd3 Kc2 Rf3 Be3
 20-	00:04	 7.735k	1.718k	+0,35	Kb1 0-0
 20-	00:04	 8.123k	1.703k	+0,29	Kb1 0-0
 20+	00:04	 8.342k	1.711k	+0,33	Kb1 0-0 Rd3
 20-	00:04	 8.453k	1.710k	+0,26	Kb1 0-0
 20+	00:05	 8.643k	1.710k	+0,36	Kb1 0-0 Rd3
 20	00:05	 8.998k	1.711k	+0,23	Kb1 0-0 Rd3 Rb7 Ra3 Kh8 Rxa6 Rg8 Be3 h5 Rb6 Rxb6 Bxb6 Qb8 Bc7 Qb7 Nxf6 Qxc7 Nxg8 Kxg8 c3 Kf8 Rd1 Ke7 Rd5 Qb6 Kc1
 21+	00:05	 9.464k	1.723k	+0,29	Kb1 0-0 Rd3
 21-	00:05	 9.793k	1.724k	+0,25	Kb1 0-0
 21	00:05	 10.247k	1.730k	+0,25	Kb1 0-0 Rd3 Rb7 Ra3 Kh8 Rxa6 Rg8 Rb6 Rxb6 Bxb6 Qb8 Bc7 Qb7 Nxf6 Qxc7 Nxg8 Kxg8 c3 Kf8 Rd1 Ke7 Kc2 Qa7 Kb3 Qf2
 22-	00:06	 10.637k	1.734k	+0,24	Kb1 0-0
 22+	00:07	 12.651k	1.776k	+0,27	Kb1 0-0 Rd3
 22	00:07	 12.723k	1.780k	+0,26	Kb1 0-0 Rd3 Rb7 Ra3 Kh8 Rxa6 Rg8 Rb6 Rxb6 Bxb6 Qb8 Bc7 Qa7 Nxf6 Qxc7 Nxg8 Kxg8 c3 Kf8 Rd1 Ke7 Kc2 Qa7 Kb3 Qf2 Ra1 h5
 23-	00:07	 13.353k	1.765k	+0,25	Kb1 0-0
 23+	00:08	 15.785k	1.773k	+0,28	Kb1 0-0 Rd3
 23-	00:09	 16.750k	1.772k	+0,23	Kb1 0-0
 23+	00:09	 17.648k	1.771k	+0,30	Kb1 0-0 Rd3
 23-	00:10	 18.817k	1.764k	+0,20	Kb1 0-0
 23+	00:11	 20.406k	1.760k	+0,34	Kb1 0-0 Rd3
 23-	00:13	 22.838k	1.735k	+0,13	Kb1 0-0
 23	00:14	 24.671k	1.722k	+0,11	Kb1 0-0 Rd3 Kh8 Ra3 Rb5 Rxa6 Rxd5 exd5 Bg7 Rf1 e4 Bxe4 Re8 Bd3 Qf6 c3 Qf3 Bc2 Qxd5 Rb6 Ra8 Re1 Qf3 Be3 Qd5 Bf4 Be5 Bxh6 Qa2+ Kc1
 24-	00:14	 25.739k	1.718k	+0,10	Kb1 0-0
 24+	00:16	 28.667k	1.723k	+0,17	c4 Rg8 Kb1
 24+	00:24	 42.179k	1.731k	+0,23	c4 Rg8 Kb1
 24+	00:29	 51.694k	1.749k	+0,32	c4 a5 c5
 24-	00:29	 52.004k	1.749k	+0,25	c4 a5
 24+	00:37	 64.464k	1.739k	+0,35	c4 a5 b5

[Eelco de Groot]

Hello Eren, I don't know M8R but thanks for the analysis! Here c4 seems edged out just a little by Kb1 at depth 25 and 26 but very close. Program has a lot of hash knowledge at this point, but mainly after Kb1 Be7 (I think the rest has been wiped out because I went to the very end of the Kb1 Be7 line and left program running for several hours then back) so after Kb1 O-O eval is lower.


[d]1r1qk2r/5p2/p2p1b1p/3NpB2/1P2P2P/8/1PP2B2/2KR3R w k -

Engine: Rainbow Serpent 20160903_016 HT (7 threads, 512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

26 2:57 +1.67 23.Kb1 O-O 24.Rdg1+ Kh8 25.Be3 Bg7
26.Rh3 Rg8 27.Rhg3 Qe8 28.c3 Qd8
29.c4 a5 30.b5 Qf8 31.Kc2 Rd8
32.Bxh6 Bxh6 33.Rxg8+ Qxg8 34.Rxg8+ Kxg8
35.b6 Kf8 36.Kb3 (1.701.389.783) 9587

26 2:57 +1.65 23.c4 a5 24.b5 Bg7 25.Rdg1 O-O 26.h5 a4
27.Bh4 f6 28.Rg6 Rb7 29.Bxf6 Rxf6
30.Nxf6+ Kh8 31.Nd5 a3 32.bxa3 Qa5
33.Rg3 Qa7 34.Rc3 Qf2 35.Kb1 Bf8
36.Rhc1 (1.701.389.783) 9587

26 2:57 +0.39 23.Rd3 Rb5 24.c4 Rxd5 25.Rxd5 Qb8
26.c5 Be7 27.Kb1 Qxb4 28.cxd6 Bxd6
29.Rhd1 Be7 30.Rxe5 f6 31.Bc5 Qxc5
32.Bg6+ Kf8 33.Rxc5 Bxc5 34.Rd7 h5
35.Rf7+ Kg8 36.Rxf6 (1.701.389.783) 9587

26 2:57 0.00 23.Rh2 Rb5 24.Rh3 O-O 25.Be3 Rxd5
26.Rxd5 Bxh4 27.Kb1 Kg7 28.Bc5 Be7
29.Rg3+ Kh8 30.Rh3 Kg7 (1.701.389.783) 9587

26 2:57 0.00 23.Rh3 Rg8 24.Rc3 Bxh4 25.Ba7 Bg5+
26.Kb1 Rb7 27.Bf2 Rb8 28.Ba7 (1.701.389.783) 9587