Does any engine find the best move here...

[zullil]

Repeating on 2x1060 GPU's after 5 minutes. ID: JH.T8.610

Finally, after a longer think.

that’s a nice graphic

I find the evaluations quite difficult to read. Rather small, with colored backgrounds. Perhaps the software that made the graphics can be tweaked.

[AdminX]

I find the evaluations quite difficult to read. Rather small, with colored backgrounds. Perhaps the software that made the graphics can be tweaked.

I scaled the image down to 320x320, which might be why. I will try slightly bigger size next time. The default was 640x640.

[zullil]

I find the evaluations quite difficult to read. Rather small, with colored backgrounds. Perhaps the software that made the graphics can be tweaked.

I scaled the image down to 320x320, which might be why. I will try slightly bigger size next time. The default was 640x640.

That might be better, at least for some of us. The graphic does give a nice visual summary of the search results.

[MikeB]

How did you get these high scores?

Stockfish 10 at depth 27 shows only +1.xx:

He has modified the evaluation to give an approximate (percentage) probability of a win. So his engine's scores range from 0 to 100.

Correct - the traditional centipawn scoring is linear. LcO and other NN do not score a game based on centipawn evaluation but something that is very close to scoring %, but the developers provide an option to convert it a centipawn evaluation which many chess and chess computer enthusiasts understand. But for me, and I appreciate not everyone feels this way, the scoring % actually is more informative and make more sense. The olds of scoring a point in the game is non linear as compared to the centipawn evaluation, For example, a +300 centipawn evaluation, roughly equal to three pawns as we all know, etc. With % scoring, +300 centipawns is about 80%, +430 centiapawns is about 90%, to me scoring % is a more informative status of the expected outcome of the game. in McCain, I use a formula in uci.cpp to covert the centiapwn score to a scoring %. Lc0 does the opposite, it converts the scoring % to a centipwn evaluation prior to output. The Lc0 formula was reversed engineered and then tweaked for McCain. I tweaked it so that when the game starts with WTM, the evaluation will show , after a deep search, about a 55% chance of scoring - which is the generally accepted scoring % for white at the start of game based on historical data.

[Max]

Correct - the traditional centipawn scoring is linear. LcO and other NN do not score a game based on centipawn evaluation but something that is very close to scoring %, but the developers provide an option to convert it a centipawn evaluation which many chess and chess computer enthusiasts understand. But for me, and I appreciate not everyone feels this way, the scoring % actually is more informative and make more sense. The olds of scoring a point in the game is non linear as compared to the centipawn evaluation, For example, a +300 centipawn evaluation, roughly equal to three pawns as we all know, etc. With % scoring, +300 centipawns is about 80%, +430 centiapawns is about 90%, to me scoring % is a more informative status of the expected outcome of the game. in McCain, I use a formula in uci.cpp to covert the centiapwn score to a scoring %. Lc0 does the opposite, it converts the scoring % to a centipwn evaluation prior to output. The Lc0 formula was reversed engineered and then tweaked for McCain. I tweaked it so that when the game starts with WTM, the evaluation will show , after a deep search, about a 55% chance of scoring - which is the generally accepted scoring % for white at the start of game based on historical data.

Thank you for explaining this concept. Agreed, evaluation in form of winning chances are very informative. Especially when combined with multi pv search.

I missed McCain complete, but downloaded it instantly now.
Thanks!

[Master Om]

[Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[Analysis "Brainfish 240619 64 POPCNT"]
[AnalysisTime "1341"]
[FEN "r3r1k1/1b1nbp1p/1qn1p3/1p1pPp2/p2P4/P2BNN2/1P1Q1BPP/R4RK1 w - - 0 19"]
[SetUp "1"]

19. Ng4 Kh8 20. Nh6 Rf8 21. Rae1 b4 22. g4 bxa3 23. bxa3 f4 24. Bb1 Ba6 25.
Qc2 f5 26. exf6 Nxf6 27. Rxe6 Qb3 28. Rfe1 Bb5 29. Rxc6 Bxc6 30. Rxe7 Qxc2 31.
Bxc2 Rae8 32. Ra7 Ra8 33. Rxa8 Bxa8 34. Bh4 Bb7 35. Bxa4 Bc8 36. Bc6 Bxg4 37.
Nxg4 Nxg4 38. Bxd5 Nf6 39. Bc6 Ng4 40. a4 Rc8 41. d5 Ne3 42. a5 Rg8+ 43. Kh1
{(0:22:21) 8042921kN}{[%eval 379,39,Brainfish 240619 64 POPCNT][%meval
1341s|TB:1k]} *

[peter]

Hi Mike!

[d]r3r1k1/1b1nbp1p/1qn1p3/1p1pPp2/p2P4/P2BNN2/1P1Q1BPP/R4RK1 w - - 2 19

And older CorChess version at 4 threads of a 2x2GHz CPU, 2 GB hash:

r3r1k1/1b1nbp1p/1qn1p3/1p1pPp2/p2P4/P2BNN2/1P1Q1BPP/R4RK1 w - - 0 1

Analysis by CorChess 2.6 101118 64 POPCNT:

19.Lxf5 exf5 20.Sxd5 Dd8 21.Df4 Sf8
+/= (0.36) Tiefe: 7/8 00:00:00 45kN
19.Sxf5 exf5 20.Lxf5 Sf8 21.Dh6 Dd8 22.Ld3 b4
=/+ (-0.41) Tiefe: 8/9 00:00:00 49kN
...
19.Sxf5 exf5 20.Lxf5 Sf8 21.Dh6 Lc8 22.Ld3 Ld8 23.Sg5 Sxe5 24.Dxb6 Lxb6 25.dxe5 Ld8 26.Sf3 Ld7 27.Tad1 Sg6 28.Lg3 Lb6+ 29.Kh1 h6 30.Tc1 d4 31.Tfe1 La5 32.Ted1 Lb6 33.Lxg6 fxg6
+/= (0.68) Tiefe: 21/38 00:00:09 12016kN
19.Sg4
+/- (0.79 ++) Tiefe: 22/36 00:00:12 16819kN
...
19.Sg4
+- (2.07 ++) Tiefe: 22/36 00:00:13 18054kN
19.Sg4 Kh8 20.Sh6 Tf8 21.Sg5 Lxg5 22.Dxg5 f6 23.Dh5 La6 24.Tac1 b4 25.Lxa6 Dxa6 26.Sf7+ Txf7 27.Dxf7
+- (2.07) Tiefe: 22/36 00:00:13 18506kN
...
19.Sg4 Kh8 20.Sh6 Tf8 21.Tae1 La6 22.g4 b4 23.Lxa6 Txa6 24.gxf5 exf5 25.Sxf5 Dd8 26.Dh6 Tg8+ 27.Kh1 Df8 28.Dh5 bxa3 29.bxa3 Sd8 30.S3h4 Lg5 31.Lg3 Dxa3 32.Sd6 Sf6 33.Sg6+ fxg6 34.Dxg5 Sh5 35.Te3 Db2 36.Lh4 Dxd4 37.Th3 Sg7
+- (3.39) Tiefe: 28/54 00:03:05 284MN

[MikeB]

Hi Mike!

[d]r3r1k1/1b1nbp1p/1qn1p3/1p1pPp2/p2P4/P2BNN2/1P1Q1BPP/R4RK1 w - - 2 19

And older CorChess version at 4 threads of a 2x2GHz CPU, 2 GB hash:

r3r1k1/1b1nbp1p/1qn1p3/1p1pPp2/p2P4/P2BNN2/1P1Q1BPP/R4RK1 w - - 0 1

Analysis by CorChess 2.6 101118 64 POPCNT:

19.Lxf5 exf5 20.Sxd5 Dd8 21.Df4 Sf8
+/= (0.36) Tiefe: 7/8 00:00:00 45kN
19.Sxf5 exf5 20.Lxf5 Sf8 21.Dh6 Dd8 22.Ld3 b4
=/+ (-0.41) Tiefe: 8/9 00:00:00 49kN
...
19.Sxf5 exf5 20.Lxf5 Sf8 21.Dh6 Lc8 22.Ld3 Ld8 23.Sg5 Sxe5 24.Dxb6 Lxb6 25.dxe5 Ld8 26.Sf3 Ld7 27.Tad1 Sg6 28.Lg3 Lb6+ 29.Kh1 h6 30.Tc1 d4 31.Tfe1 La5 32.Ted1 Lb6 33.Lxg6 fxg6
+/= (0.68) Tiefe: 21/38 00:00:09 12016kN
19.Sg4
+/- (0.79 ++) Tiefe: 22/36 00:00:12 16819kN
...
19.Sg4
+- (2.07 ++) Tiefe: 22/36 00:00:13 18054kN
19.Sg4 Kh8 20.Sh6 Tf8 21.Sg5 Lxg5 22.Dxg5 f6 23.Dh5 La6 24.Tac1 b4 25.Lxa6 Dxa6 26.Sf7+ Txf7 27.Dxf7
+- (2.07) Tiefe: 22/36 00:00:13 18506kN
...
19.Sg4 Kh8 20.Sh6 Tf8 21.Tae1 La6 22.g4 b4 23.Lxa6 Txa6 24.gxf5 exf5 25.Sxf5 Dd8 26.Dh6 Tg8+ 27.Kh1 Df8 28.Dh5 bxa3 29.bxa3 Sd8 30.S3h4 Lg5 31.Lg3 Dxa3 32.Sd6 Sf6 33.Sg6+ fxg6 34.Dxg5 Sh5 35.Te3 Db2 36.Lh4 Dxd4 37.Th3 Sg7
+- (3.39) Tiefe: 28/54 00:03:05 284MN

That's pretty fast!

[peter]

That's pretty fast!

r3r1k1/1b1nbp1p/1qn1p3/1p1pPp2/p2P4/P2BNN2/1P1Q1BPP/R4RK1 w - - 0 1

Analysis by Xiphos 0.5 SSE:

Code: Select all

1.Rad1 Na5 2.Qc2 Nb3 3.Bxf5 exf5 4.Qxf5 
  =  (-0.13)   Depth: 7   00:00:00  45kN
...
1.Bxf5 exf5 2.Nxf5 Kh8 3.Qh6 Rg8 4.Qh5 Raf8 5.Nxe7 Nxe7 6.Ng5 
  =  (-0.08)   Depth: 11   00:00:00  474kN
...
1.Nd1 Nf8 2.Qh6 Ng6 3.Nc3 b4 4.axb4 Bf8 5.Qd2 Bxb4 6.Qh6 Bf8 7.Qd2 Bb4 
  =  (0.00)   Depth: 20   00:00:00  8864kN
1.Ng4 fxg4 2.Ng5 f5 3.exf6 Bxf6 4.Nxh7 Bg7 5.Qf4 Ne7 6.Qxg4 Rf8 7.Nxf8 Rxf8 8.Be3 Nf5 9.Bxf5 exf5 10.Rxf5 Qe6 11.Rxf8+ Nxf8 12.Qf3 Qe4 
  +/-  (0.82)   Depth: 21   00:00:02  46047kN
...
1.Ng4 fxg4 2.Ng5 f5 3.exf6 Bxf6 4.Nxh7 Bg7 5.Qf4 Ne7 6.Qxg4 Rf8 7.Nxf8 Rxf8 8.Be3 Nf5 9.Bxf5 exf5 10.Rxf5 Qe6 11.Rxf8+ Nxf8 12.Qf3 Qe4 13.Rf1 Qxf3 
  +/-  (0.82)   Depth: 26   00:00:03  63571kN
1.Ng4 Rf8 2.Qh6 Kh8 3.Qh5 Rg8 4.Nh6 Rg7 5.Nxf7+ Kg8 6.N3g5 Bxg5 7.Nxg5 Nf8 8.Qh6 Rg6 9.Qh4 Qc7 10.Rac1 Qg7 11.Be3 h6 12.Nxe6 Nxe6 13.Bxf5 Rxg2+ 14.Kh1 Nexd4 15.Bxd4 Nxd4 
  +/-  (1.31)   Depth: 27   00:00:11  261MN
1.Ng4 Rf8 2.Nh6+ Kh8 3.g4 fxg4 4.Ng5 f6 5.exf6 Bxf6 6.Rae1 Nxd4 7.Qf4 Bxg5 8.Bxd4+ e5 9.Qxf8+ Rxf8 10.Rxf8+ Nxf8 11.Bxb6 Bxh6 12.Rxe5 Bc6 13.Re7 Bg7 14.Bf5 b4 15.Rc7 Bb5 16.Bc5 Bxb2 17.Bxb4 Ng6 18.Bxg4 Bd4+ 19.Kg2 Be5 20.Rf7 Kg8 21.Rb7 
  +-  (1.71)   Depth: 28   00:00:17  402MN
...
1.Ng4 Rf8 2.Nh6+ Kh8 3.g4 fxg4 4.Ng5 f6 5.exf6 Bxf6 6.Nhf7+ Rxf7 7.Nxf7+ Kg7 8.Ng5 Nxd4 9.Qf4 h5 10.Rac1 Rc8 11.Rxc8 Bxc8 12.Nxe6+ Qxe6 13.Bxd4 Bxd4+ 14.Qxd4+ Qe5 15.Qf2 Nf6 16.Bxb5 d4 
  +-  (2.02)   Depth: 31   00:01:18  1795MN
1.Ng4 Rf8 2.Nh6+ Kh8 3.Rae1 Qc7 4.g4 fxg4 5.Ng5 Bxg5 6.Qxg5 f5 7.exf6 Nxf6 8.Bh4 Ng8 9.Nxg8 Qg7 10.Qh6 Nxd4 11.Bf6 Rxf6 12.Qxg7+ Kxg7 13.Nxf6 Nf3+ 14.Rxf3 gxf3 15.Rxe6 Ra6 16.Nh5+ Kf7 17.Rxa6 Bxa6 18.Kf2 Ke6 19.Kxf3 
  +-  (2.32)   Depth: 32   00:02:35  3533MN
1.Ng4 fxg4 2.Qh6 Nf8 3.Be3 Qc7 4.Bg5 Nxe5 5.Nxe5 f5 6.Rac1 Bxg5 7.Qxg5+ Qg7 8.Qxg7+ Kxg7 9.Rc7+ Kg8 10.Rxb7 Rab8 11.Ra7 Ng6 12.Nd7 Re7 13.Re1 Rc8 14.Bxb5 Rc2 15.Bxa4 Rxb2 16.Nf6+ Kf7 17.Rxe7+ Kxe7 
  +-  (2.61)   Depth: 33   00:03:29  4721MN
...
1.Ng4 fxg4 2.Qh6 Nf8 3.Be3 Qc7 4.Bg5 Nxe5 5.Nxe5 f5 6.Rac1 Bxg5 7.Qxg5+ Qg7 8.Qxg7+ Kxg7 9.Rc7+ Kg8 10.Rxb7 Reb8 11.Rc7 Rc8 12.Re7 Re8 13.Rxe8 Rxe8 14.Nc6 Ra8 15.Rc1 h5 16.Bxb5 h4 17.Ne5 Nh7 18.Nd7 Ng5 19.Rc7 h3 
  +-  (2.74)   Depth: 38   00:05:24  7257MN, tb=1

24 threads of 12x3GHz CPU, 32G hash.

[mwyoung]

[d]r3r1k1/1b1nbp1p/1qn1p3/1p1pPp2/p2P4/P2BNN2/1P1Q1BPP/R4RK1 w - - 2 19

Clearly white has the initiative, but there is one move the that stands far above the rest.

Might be an excellent position for Lc0 on fast hardware...

As I was typing this (McCain X4) - really not sure of this is a good search or no - found after a 6B node search :

Code: Select all

dep	score	nodes	time	(not shown:  tbhits	knps	seldep)
 37	+78.20 	6.93G	5:26.05	Ng4 Rf8 Nh6+ Kh8 Rae1 b4 g4 f4 Bb1 bxa3 bxa3 Ba6 Qc2 f5 exf6 Nxf6 Rxe6 Qb3 Rfe1 Rac8 Rxc6 Qxc2 Rxc2 Bxa3 Rxc8 Bxc8 g5 Ng8 Nxg8 Kxg8 Ba2 Rd8 Kg2 Kg7 Rb1 Bf5 Rb7+ Kf8 Ne5 Be4+ Kg1 Be7 Nd7+ Rxd7 Rxd7 Ke8 Rc7 Bxg5 
 37	+75.14!	6.55G	5:08.32	Ng4! 
 37	+71.70!	6.36G	4:59.34	Ng4! 
 37	+68.89!	6.28G	4:55.73	Ng4! 
 37	+66.66!	6.25G	4:54.35	Ng4! 
 37	+64.96!	6.19G	4:51.30	Ng4! 
 37	+63.71!	6.08G	4:46.07	Ng4! 
 36	+62.45 	3.98G	3:06.86	Kh1 Kh8 Nxf5 exf5 Bxf5 Nf8 Qh6 Ra6 Be3 Bc8 Bxc8 Rxc8 Ng5 Nxe5 Qxb6 Rxb6 dxe5 Bc5 Nxf7+ Kg7 Rac1 Rbc6 Bd2 Ng6 h4 h5 Rcd1 d4 Bg5 Re8 Nd8 Rc8 Nb7 Nxe5 Nxc5 Rxc5 Rxd4 Ng4 Bd2 Re6 Bc3 Kg6 Rd8 Nh6 Kh2 Ng4+ Kh3 Re3+ Rf3 Nf2+ Kh2 Ng4+  
 35	+62.95 	3.35G	2:37.11	Kh1 b4 g4 bxa3 bxa3 Ncxe5 Nxe5 Nxe5 dxe5 Bg5 Ng2 Bxd2 Bxb6 fxg4 Bd4 Rec8 Bxh7+ Kxh7 Rxf7+ Kg6 Rxb7 Rc1+ Rxc1 Bxc1 Rc7 Bg5 Kg1 Rb8 Ra7 Rb3 Rxa4 Kf5 Bf2 Kxe5 Rxg4 Kf5 Rg3 Rb1+ Ne1 e5 Rd3 d4 Kg2 Rb2 Rf3+ Ke4 Kh3 Ra2 Rg3 Kf5 Nf3 Rx 
 35	+64.62!	3.10G	2:25.66	Kh1! 
 35	+63.37!	2.97G	2:18.96	Kh1! 
 34	+62.10 	2.79G	2:10.59	Kh1 Kh8 Nxd5 exd5 Bxf5 Nf8 Qh6 Ra6 Be3 Bc8 Bxc8 Rxc8 Ng5 Nxe5 Qxb6 Rxb6 dxe5 Rg6 Nxf7+ Kg7 Rad1 Bc5 Rc1 Rgc6 Bd2 Ng6 h4 h5 Rcd1 d4 Bg5 Re8 Nd8 Rc8 Nb7 Nxe5 Nxc5 Rxc5 Rxd4 Ng4 Bd2 Re6 Bc3 Kg6 Rd8 Nf6 Kh2 Rc4 Kh3 Rg4 Rxf6+ Rxf6 B 
 34	+62.60?	2.60G	2:01.72	Kh1 Kh8? 
 34	+64.36!	2.54G	1:59.06	Kh1! 
 34	+63.56?	2.28G	1:46.54	Kh1 Kh8? 
 34	+64.17?	2.25G	1:45.18	Kh1 Kh8? 
 33	+64.77 	2.03G	1:34.95	Kh1 Kh8 Nxf5 exf5 Bxf5 Nf8 Qh6 Ra6 Be3 Bc8 Bxc8 Rxc8 Ng5 Nd8 e6 Kg8 exf7+ Nxf7 Nxf7 Qxh6 Nxh6+ Kh8 Rf5 Re6 Ng4 Nd7 Rxd5 Nf6 Re5 Rxe5 Nxe5 Kg7 Rc1 Rxc1+ Bxc1 Bd6 Bh6+ Kg8 Bf4 Bxe5 dxe5 Ne4 Be3 Kf7 g4 Ke6 Kg2 Kxe5 Kf3 Kd5 h4 Nd6  
 32	+64.20 	1.67G	1:18.36	Kh1 Kh8 Nxf5 exf5 Bxf5 Nf8 Qh6 Ra6 Be3 Bc8 Bxc8 Rxc8 Ng5 Nxe5 Qxb6 Rxb6 dxe5 Rg6 Nxf7+ Kg7 Rad1 Bc5 Bg1 Bxg1 Kxg1 Rc2 g3 Rc5 Nd6 Kg8 Rc1 Rxc1 Rxc1 d4 Rd1 Re6 Nxb5 Rxe5 Nxd4 Rd5 
 32	+63.71!	1.60G	1:15.13	Kh1! 
 31	+62.45 	1.43G	1:07.26	Kh1 Kh8 Nxf5 exf5 Bxf5 Nf8 Qh6 Ra6 Be3 Bc8 Bxc8 Rxc8 Ng5 Nxe5 Qxb6 Rxb6 dxe5 Bc5 Nxf7+ Kg7 Bd2 d4 Rac1 Rbc6 h4 Be7 Bh6+ Kg8 Rcd1 Rg6 h5 Rg4 Bd2 Rh4+ Kg1 Rxh5 Nh6+ Kh8 Nf5 Ng6 e6 Kg8 Nh6+ Kh8 Rf7 Rf8 Rdf1 Rxf7 
 31	+62.10!	1.37G	1:04.20	Kh1! 
 31	+60.46?	1.26G	0:59.41	Kh1 Kh8? 
 31	+62.02!	1.22G	0:57.42	Kh1! 
 31	+61.09?	1.15G	0:53.97	Kh1 Kh8? 
 30	+61.71 	1.11G	0:52.07	Kh1 Kh8 Nxf5 exf5 Bxf5 Nf8 Qh6 Ra6 Qh5 Bc8 Bd3 Nd8 Be3 Qb7 Nh4 b4 axb4 Bd7 Nf5 Bxf5 Rxf5 Bxb4 Bxa6 Qxa6 Qd1 Be7 Kg1 Qc6 Rxa4 Nfe6 Rh5 Ng7 Rh6 Qd7 Rh3 h5 Bh6 Nde6 Bxg7+ Kxg7 Ra6 Nf4 Rg3+ Kf8 
 30	+63.44!	1.09G	0:51.17	Kh1! 
 30	+60.62?	998.4M	0:46.95	Kh1 Kh8? 
 30	+62.91!	960.0M	0:45.10	Kh1! 
 30	+62.02?	921.5M	0:43.26	Kh1 Kh8? 
 30	+62.99?	904.1M	0:42.45	Kh1 Kh8? 
 30	+63.60?	897.8M	0:42.15	Kh1 Kh8? 
 29	+64.20 	779.5M	0:36.55	Kh1 b4 g4 Ncxe5 Nxe5 Nxe5 dxe5 bxa3 bxa3 Bg5 Ng2 Bxd2 Bxb6 Bc3 Rac1 Rac8 gxf5 Bxe5 fxe6 fxe6 Bc5 d4 Kg1 Rc7 Rfe1 Bg7 Bb4 Rf7 Rf1 Rxf1+ Rxf1 e5 Nh4 Bd5 Nf5 e4 Be2 Kh8 Nd6 Re5 Nf7+ Bxf7 Rxf7 
 28	+64.36 	573.1M	0:26.78	Kh1 b4 g4 Ncxe5 Nxe5 Nxe5 dxe5 bxa3 bxa3 Bg5 Ng2 Bxd2 Bxb6 Bc3 Rac1 Rac8 gxf5 Bxe5 fxe6 fxe6 Bc5 Rc7 Rce1 Bc3 Bd6 Rc6 Rb1 Ba6 Bxa6 Rxa6 Bg3 e5 Rb7 Rae6 Rf5 d4 Rff7 
 28	+63.25!	530.5M	0:24.80	Kh1! 
 28	+61.32!	514.8M	0:24.11	Kh1! 
 28	+59.91!	500.2M	0:23.42	Kh1! 
 28	+58.91!	489.2M	0:22.91	Kh1! 
 27	+57.91 	393.5M	0:18.30	Kh1 Qa5 Nxf5 Qxd2 Nxe7+ Nxe7 Nxd2 Ba6 Bh4 Ng6 Be1 Reb8 Rc1 Rc8 Rxc8+ Rxc8 h4 Ndxe5 dxe5 Nxe5 Bb1 b4 Rg1 Nd3 axb4 Nxb2 Nf3 a3 Bd2 Bc4 Kh2 a2 Bc2 Ra8 Ra1 Nd3 Bc3 Bb5 
 27	+58.16!	388.5M	0:18.07	Kh1! 
 27	+57.27!	371.1M	0:17.27	Kh1! 
 26	+56.38 	352.7M	0:16.42	Kh1 Qa5 Nxf5 Qxd2 Nxe7+ Nxe7 Nxd2 Ba6 b4 Nb6 Bh4 Nc4 Ra2 Rac8 Bg5 Nc6 Rf4 Nxd2 Rxd2 Nxd4 Rd1 Nf5 Bxf5 exf5 Rxf5 Rc3 Bf6 Rxa3 Rh5 Rc3 
 26	+61.87!	336.3M	0:15.64	Kh1! 
 26	+58.83!	297.9M	0:13.83	Kh1! 
 26	+56.50!	282.3M	0:13.08	Kh1! 
 26	+54.76!	194.8M	0:08.95	Bxf5! 
 26	+54.06?	168.7M	0:07.76	Bxf5 exf5? 
 26	+54.72?	163.0M	0:07.50	Bxf5 exf5? 
 25	+55.37 	142.5M	0:06.58	Bxf5 exf5 Nxf5 Kh8 Ng5 Bxg5 Qxg5 Rg8 Qh5 Rae8 Nd6 Nd8 Nxe8 Rxe8 Rac1 b4 axb4 Ba6 Rfd1 Rg8 Bh4 Bc4 Bxd8 Rxd8 Rc3 Rg8 Qf5 Qg6 Qxg6 Rxg6 Ra3 Bb5 Rf3 Kg7 Kf2 
 25	+53.16?	141.9M	0:06.56	Bxf5 exf5? 
 25	+55.49!	131.3M	0:06.07	Bxf5! 
 25	+54.64?	115.9M	0:05.38	Bxf5 exf5? 
 25	+55.65?	91.9M  	0:04.28	Bxf5 exf5? 
 25	+56.30?	82.4M  	0:03.85	Bxf5 exf5? 
 24	+56.95 	67.3M  	0:03.17	Bxf5 exf5 Nxf5 Kh8 Ng5 Bxg5 Qxg5 Rg8 Qh5 Raf8 Nd6 f5 Rae1 Na5 Nf7+ Rxf7 Qxf7 Qg6 Qxg6 hxg6 g4 Nc6 Bg3 Kg7 gxf5 Nxd4 f6+ Kf7 Rf4 Ne6 Rh4 Rd8 Rh7+ Kg8 Re7 Ndc5 
 24	+57.79!	64.5M  	0:03.04	Bxf5! 
 23	+56.02 	56.9M  	0:02.71	Bxf5 exf5 Nxf5 Kh8 Ng5 Bxg5 Qxg5 Rg8 Qh5 Raf8 Nd6 f5 Rae1 Na5 Nf7+ Rxf7 Qxf7 Qg6 Qxg6 hxg6 Bh4 g5 e6 Nb6 e7 Bc6 Bxg5 Rxg5 Re6 Rg8 h3 Kg7 Rxf5 Nbc4 g4 
 22	+54.68 	49.4M  	0:02.38	Bxf5 exf5 Nxf5 Kh8 Ng5 Bxg5 Qxg5 Rg8 Qh5 Rae8 Rae1 Nd8 Bh4 Qg6 Qxg6 Rxg6 Bxd8 Rxd8 Nd6 Rb8 Rxf7 Bc6 Rc1 Rb6 
 21	+54.80 	43.7M  	0:02.12	Bxf5 exf5 Nxf5 Kh8 Ng5 Bxg5 Qxg5 Rg8 Qh5 Rae8 Nd6 Nd8 Nxe8 Rxe8 Rae1 Rg8 Qh4 Qg6 Bg3 Bc6 Qe7 Qd3 Kh1 Qg6 Rf2 Qg4 
 20	+56.83 	38.4M  	0:01.88	Bxf5 exf5 Nxf5 Kh8 Ng5 Bxg5 Qxg5 Rg8 Qh5 Raf8 Nd6 Rg6 Rae1 f5 Nxf5 b4 Be3 Ba6 Nh4 Rxf1+ Rxf1 Ncxe5 Nxg6+ Qxg6 Qxg6 Nxg6 Rf7 
 19	+54.72 	28.6M  	0:01.40	Bxf5 exf5 Nxf5 Kh8 Ng5 Nd8 Bh4 Bxg5 Qxg5 Rg8 Qe3 Nc6 Nd6 Qxd4 Qxd4 Nxd4 Nxb7 Rg4 g3 Kg8 Rad1 Ra6 h3 Re4 Nd6 Rxe5 Nxf7 
 18	+54.02 	17.1M  	0:00.84	Bxf5 exf5 Nxf5 Kh8 Ng5 Bxg5 Qxg5 Rg8 Qh5 Nd8 Bg3 Rg6 Rf2 Ne6 Raf1 Rf8 Nd6 Qxd4 Nxb7 Rg5 Qd1 Nxe5 Qxd4 Nxd4 
 17	+56.63 	9.15M  	0:00.44	Bxf5 exf5 Nxf5 Kh8 Qh6 Rg8 Be3 Rg6 Qh5 Rag8 Kh1 Bd8 N3h4 Bxh4 Nxh4 Nd8 Rf5 Qe6 Raf1 R6g7 R1f3 
 16	+55.37 	6.81M  	0:00.32	Bxf5 exf5 Nxf5 Kh8 Qh6 Rg8 Qh5 Rg6 Nxe7 Nxe7 Ng5 Rxg5 Qxg5 Re8 Qf4 f6 Bh4 f5 Bxe7 Rxe7 
 15	+52.67 	4.27M  	0:00.20	Nxf5 exf5 Bxf5 Nf8 Qh6 Bc8 Bc2 Ng6 Be3 Be6 Rf2 Rec8 Ng5 Bxg5 Qxg5 Na5 Raf1 
 14	+52.47 	3.59M  	0:00.17	Nxd5 exd5 Bxf5 Nf8 Qh6 Bc8 Bc2 Ng6 Be3 b4 axb4 Ba6 Rf2 Nxb4 Ng5 Bxg5 Qxg5 Nxc2 Rxc2 
 13	+51.97 	2.71M  	0:00.13	Nxf5 exf5 Bxf5 Bc8 Qh6 Nf8 Bc2 Ng6 Be3 Bd7 Rf2 b4 Ng5 Bxg5 Qxg5 Nxd4 
 12	+50.74 	1.17M  	0:00.06	Nc2 Nf8 Qh6 b4 Nxb4 Nxb4 axb4 Qxb4 Rfb1 Ng6 Be2 Ba6 Bxa6 Rxa6 Ng5 Bxg5 Qxg5 Qb6 
 11	+53.20 	372066	0:00.02	Nc2 Nf8 Qh6 b4 axb4 Nxb4 Nxb4 Qxb4 Qh5 Qb3 Rad1 Ba6 Bxa6 Rxa6 Ng5 Bxg5 Qxg5+ Ng6 
 10	+53.16 	310368	0:00.02	Nc2 Nf8 Qh6 b4 axb4 Nxb4 Nxb4 Qxb4 Qh5 Ba6 Bxa6 Rxa6 Ng5 Bxg5 
  9	+53.49 	158482	0:00.01	Nc2 Na5 Qh6 Nc4 Rab1 Nf8 b3 axb3 Rxb3 
  8	+56.67 	59491  	0:00.00	Bxf5 exf5 Nxf5 Kh8 Qf4 Rf8 Rfc1 b4 Ng5 
  7	+66.04 	24648  	0:00.00	Bxf5 Nxd4 Nxd4 exf5 Nexf5 Nxe5 Bg3 
  6	+71.97 	9507    	0:00.00	Bxf5 
  5	+56.63 	5250    	0:00.00	Bxf5 b4 Bb1 bxa3 
  4	+54.15 	3542    	0:00.00	Bxf5 b4 Bb1 b3 
  3	+53.33 	1940    	0:00.00	Bxf5 exf5 Nxf5 
  2	+49.75 	1059    	0:00.00	Nc2 b4 
  1	+49.01 	458      	0:00.00	Nc2 
  1	Found 1511 tablebases 

You need a upgrade. Even my POS using 36 threads was 2x faster.

New game Line
r3r1k1/1b1nbp1p/1qn1p3/1p1pPp2/p2P4/P2BNN2/1P1Q1BPP/R4RK1 w - - 0 1

Analysis by Stockfish 270619 64 POPCNT:

1.Ng4
White is clearly better: +- (2.39 ++) Depth: 36/77 00:02:26 5058MN, tb=18629
(, 27.06.2019)