Test position 30...Nxb3

[Vinvin]

From the SF dev forum : https://groups.google.com/forum/?fromgr ... 715W1Ruh3g
30...Nxb3! SF has trouble here !
[d]r5rk/1n1b2qp/3p1b2/p1nP1p2/1pPNpP2/1P2B1P1/1N2B1RP/1R1Q3K b - - 19 30

[Eelco de Groot]

For a working link you can click on the small square to the far right, after the date of the first message: https://groups.google.com/d/msg/fishcoo ... WhwAwUCgAJ

[acase]

From the SF dev forum : https://groups.google.com/forum/?fromgr ... 715W1Ruh3g
30...Nxb3! SF has trouble here !
[d]r5rk/1n1b2qp/3p1b2/p1nP1p2/1pPNpP2/1P2B1P1/1N2B1RP/1R1Q3K b - - 19 30

You are correct, Stockfish does indeed take a long time to find 30...Nxb3!.


on my 20 core xeon with 16gb hash table it takes a little more than 48 minutes to find 30...Nxb3!



FEN: r5rk/1n1b2qp/3p1b2/p1nP1p2/1pPNpP2/1P2B1P1/1N2B1RP/1R1Q3K b - - 19 30

Stockfish_17121816_x64_bmi2:

49/7 14:23 23,961,715k 27,763k 0.00 30. ... Ra8-c8 31.Be2-h5 Rc8-a8 32.Nd4-c6 Ra8-c8 33.Nc6-d4
50/48 37:12 60,946,642k 27,298k 0.00 30. ... Ra8-f8 31.Be2-h5 Qg7-h6 32.Kh1-g1 Qh6-g7 33.Kg1-h1
51/53+ 48:14 79,388,941k 27,429k +0.07 30. ... Nc5xb3
51/57+ 51:27 84,857,453k 27,484k +0.15 30. ... Nc5xb3
51/96+ 1:10:06 116,723,529k 27,751k +0.92 30. ... Nc5xb3
51/97 1:26:42 145,535,327k 27,976k +0.95 30. ... Nc5xb3 31.Nd4xb3 a5-a4 32.Nb2xa4 Bd7xa4 33.Qd1-g1 Ba4xb3 34.Rb1xb3 Ra8-a1 35.Be2-d1 Rg8-a8 36.g3-g4 Nb7-c5 37.Rb3xb4 Nc5-d3 38.Rb4-b3 Qg7-c7 39.g4-g5 Bf6-g7 40.Rg2-c2 Qc7-a5 41.Rb3-b7 Qa5-e1 42.Qg1xe1 Nd3xe1 43.Rc2-c1 Ra1xc1 44.Be3xc1 Bg7-d4 45.c4-c5 Ne1-d3 46.c5-c6 Ra8-a1 47.c6-c7 Ra1xc1 48.Kh1-g2 h7-h6 49.Rb7-b8+ Kh8-g7 50.c7-c8Q Nd3xf4+ 51.Kg2-g3 Rc1xc8 52.Rb8xc8 h6xg5 53.h2-h4 Bd4-e5 54.h4xg5 Nf4xd5+ 55.Kg3-f2 Nd5-f4 56.Bd1-b3 Kg7-g6 57.Rc8-g8+ Kg6-h5 58.Kf2-f1 Kh5-g4 59.Kf1-e1 Nf4-h5 60.Bb3-d1+ Kg4-h4 61.Bd1-b3 Nh5-g3 62.Bb3-d5 Kh4-g4 63.g5-g6 Kg4-g5 64.Ke1-d2 f5-f4 65.Kd2-d1 Kg5-h6 66.Kd1-e1 Be5-d4 67.Rg8-e8 Kh6xg6 68.Re8xe4 Ng3xe4 69.Bd5xe4+ Kg6-f6 70.Be4-g2 Kf6-f5 71.Ke1-e2 Bd4-e3 72.Bg2-f3 Kf5-g5 73.Ke2-d3
52/94 1:36:04 161,902,522k 28,086k +0.93 30. ... Nc5xb3 31.Nd4xb3 a5-a4 32.Nb2xa4 Bd7xa4 33.Qd1-g1 Ba4xb3 34.Rb1xb3 Ra8-a1 35.Be2-d1 Nb7-c5 36.Rb3xb4 Rg8-a8 37.g3-g4 Nc5-d3 38.Rb4-b3 Qg7-c7 39.g4-g5 Bf6-g7 40.Rg2-c2 Qc7-a5 41.Rb3-b7 Qa5-e1 42.Qg1xe1 Nd3xe1 43.Rc2-c1 Ra1xc1 44.Be3xc1 Bg7-d4 45.c4-c5 Ne1-d3 46.c5-c6 Ra8-a1 47.c6-c7 Ra1xc1 48.Kh1-g2 h7-h6 49.Rb7-b8+ Kh8-g7 50.c7-c8Q Rc1xc8 51.Rb8xc8 Nd3xf4+ 52.Kg2-g3 h6xg5 53.h2-h4 Bd4-e5 54.h4xg5 Nf4xd5+ 55.Kg3-f2 Nd5-f4 56.Bd1-b3 Kg7-g6 57.Rc8-g8+ Kg6-h5 58.Bb3-d1+ Kh5-h4 59.Bd1-b3 Nf4-h5 60.Bb3-d5 Kh4-g4 61.Bd5-c6 Nh5-f4 62.Kf2-e1 d6-d5 63.g5-g6 Kg4-g5 64.g6-g7 e4-e3 65.Bc6-b5 Be5-d4 66.Rg8-e8 Bd4xg7 67.Re8xe3 Nf4-g2+ 68.Ke1-d2 Ng2xe3 69.Kd2xe3 Kg5-g4 70.Bb5-d7 d5-d4+ 71.Ke3-d2 Kg4-g5 72.Bd7-b5 Kg5-f4 73.Bb5-e2 Bg7-e5
53/85+ 1:45:54 178,869,748k 28,146k +1.00 30. ... Nc5xb3
53/85- 1:46:06 179,186,864k 28,147k +0.85 30. ... Nc5xb3 31.Nd4xb3
53/85 1:46:49 180,416,414k 28,146k +0.92 30. ... Nc5xb3 31.Nd4xb3 a5-a4 32.Nb2xa4 Bd7xa4 33.Qd1-g1 Ba4xb3 34.Rb1xb3 Ra8-a1 35.Be2-d1 Rg8-a8 36.Rb3xb4 Nb7-c5 37.g3-g4 Nc5-d3 38.Rb4-b3 Qg7-c7 39.g4-g5 Bf6-g7 40.Rg2-c2 Qc7-a5 41.Rb3-b7 Qa5-e1 42.Qg1xe1 Nd3xe1 43.Rc2-c1 Ra1xc1 44.Be3xc1 Bg7-d4 45.c4-c5 Ne1-d3 46.c5-c6 Ra8-a1 47.c6-c7 Ra1xc1 48.Kh1-g2 h7-h6 49.Rb7-b8+ Kh8-g7 50.c7-c8Q Rc1xc8 51.Rb8xc8 Nd3xf4+ 52.Kg2-g3 Bd4-e5 53.Kg3-h4 e4-e3 54.Rc8-c2 Nf4xd5 55.g5xh6+ Kg7xh6 56.Rc2-e2 Nd5-f4 57.Re2-a2 d6-d5 58.Ra2-a6+ Nf4-g6+ 59.Kh4-h3 d5-d4 60.Bd1-e2 Kh6-g5 61.Be2-b5 Be5-f6 62.Ra6-a1 Ng6-e5 63.Ra1-d1 d4-d3 64.Rd1xd3 Ne5xd3 65.Bb5xd3 Bf6-e5 66.Kh3-g2

[shrapnel]

[You are correct, Stockfish does indeed take a long time to find 30...Nxb3!.

Yep. Time for AlphaZero or some similar Engine to take over from the doddering Fishstock.

[AndrewGrant]

Yep. Time for AlphaZero or some similar Engine to take over from the doddering Fishstock.

EtherealDev

Code: Select all

info depth 21 score cp 23 time 2043 nodes 14737087 nps 7209000 hashfull 11 pv c5b3 d4b3 a5a4
b2a4 d7a4 d1g1 a4b3 b1b3 a8a1 e2f1 b7a5 b3b4 f6c3 b4b6 a5c4 f1c4 a1g1 e3g1 c3d4 b6d6 d4g1 h1g1

EtherealDev with more threads and hash

Code: Select all

info depth 10 score cp 38 time 92 nodes 833680 nps 8964000 hashfull 0 pv c5b3

I've grown tired of your comments. It's not yet time for doomsaying.

[acase]

[You are correct, Stockfish does indeed take a long time to find 30...Nxb3!.

Yep. Time for AlphaZero or some similar Engine to take over from the doddering Fishstock.

Well Anil, I wouldn't quite call Stockfish "doddering" (I get the feeling that you think that Stockfish, Houdini, and Komodo all suck now because you believe there is something better now, but that's like saying that Kevin Durant sucks at basketball because Lebron James is better), and AlphaZero (or something similar) obviously isn't something that we could get, especially at an affordable price point (hardware + software), but not only all of that I mentioned, I don't know that AlphaZero or something similar would be any good at test positions.


Also It should be noted that Stockfish did eventually find the correct move.


p.s. if you are the type who thinks that only the best of everything is good enough for you, then how come your hardware isn't the best, and if you say it was the best when you bought it, then why haven't you upgraded since then to the latest and greatest?

[royb]

[You are correct, Stockfish does indeed take a long time to find 30...Nxb3!.

Yep. Time for AlphaZero or some similar Engine to take over from the doddering Fishstock.

Well Anil, I wouldn't quite call Stockfish "doddering" (I get the feeling that you think that Stockfish, Houdini, and Komodo all suck now because you believe there is something better now, but that's like saying that Kevin Durant sucks at basketball because Lebron James is better), and AlphaZero (or something similar) obviously isn't something that we could get, especially at an affordable price point (hardware + software), but not only all of that I mentioned, I don't know that AlphaZero or something similar would be any good at test positions.


Also It should be noted that Stockfish did eventually find the correct move.


p.s. if you are the type who thinks that only the best of everything is good enough for you, then how come your hardware isn't the best, and if you say it was the best when you bought it, then why haven't you upgraded since then to the latest and greatest?

Komodo 11.2 finds ...Nxb3 very quickly (4 seconds if memory serves on my single core test).

[peter]

Hi!

[Komodo 11.2 finds ...Nxb3 very quickly

Time for shared hash for SF (settings) and the big commercial ones to offer shared hash with SF.

And of course high time for revival of the learning files of Houdini and SF (PA). Komodo should get one by all means of course too.

When all that works fine, maybe Google will be ready for the rematch too then.

[peter]

Hi!
Junior 13.3 sticks to ...Nxb3 on single core almost from the very start

[MikeGL]

From the SF dev forum : https://groups.google.com/forum/?fromgr ... 715W1Ruh3g
30...Nxb3! SF has trouble here !
[d]r5rk/1n1b2qp/3p1b2/p1nP1p2/1pPNpP2/1P2B1P1/1N2B1RP/1R1Q3K b - - 19 30

Looks like this 30...Nxb3 is a very deep line which just ends in an equal endgame.
I followed all the PV lines and variations posted above, but can't see a direct winning line for
black. Just an equal endgame with opposite-colored-bishops.