Are we sure that Stockfish NNUE is better than the Normal Stockfish ?

[Chessqueen]

Because the Neural Net in Lc0 is bigger than the one in SF NNUE, do we have any examples of where Lc0 evaluates a position better?
I'm thinking of positions where a deep AB search won't help much, such as blocked positions.

Openings. Positionally in the openings, it seems Lc0 > SF NNUE > SF AB > other AB.

Let see How Stockfish NNUE handle a 1 move Opening that are balanced like 1.h3 or 1.g3 which does NOT give an edge to White since the first move only occupy the 3rd rank. Let see what happens NEXT game? https://tcec-chess.com/live.html

https://www.chess.com/forum/view/chess- ... o-play-1h3

Please do NOT think that 1.h3 automatically give Black the edge ==>

This game was played between Smallfish = Stockfish for iphone and Komodo 13.3 using My Dell inspiron 22 All in one. I fed the first move 1.h3 the Clemenz Opening knowing that none of the two had the best variations programmed to their opening book, I just wanted to see how both engines and iphone Xs Max versus my Dell handle this not too well known opening.

[pgn][Event "Lucas Chess"]
[Site "?"]
[Date "2020.08.18"]
[Round "1"]
[White "iphone Xs Max Smallfish"]
[Black "Dell inspiron 22 Komodo 13.3 64-bit"]
[Result "0.5"]
[BlackElo "3365"]
[ECO "A00"]
[Opening "Clemenz (Mead-Basman-de Klerk-Welling) Opening"]
[WhiteElo "3400"]
[TimeControl "485+6"]
[Termination "normal"]


1. h3 e5 2. e4 Nf6 3. Nc3 Bb4 4. Nge2 O-O 5. a3 Bc5 6. b4 Bb6 7. g3 d5 8.
exd5 Nxd5 9. Bg2 Nxc3 10. Nxc3 Nc6 {(Nb8-c6 O-O Bc8-e6 d2-d3 f7-f5 Nc3-a4
f5-f4 Na4xb6 c7xb6 Bc1-b2 Be6-d5 Bg2xd5+ Qd8xd5 Qd1-e2 Nc6-d4 Bb2xd4 Qd5xd4
g3-g4 f4-f3 Qe2-e3 Ra8-e8 Qe3xd4 e5xd4 Rf1-e1 Re8-c8 Re1-e4 Rc8xc2 Re4xd4
Rc2-c3) +1.00/25 44} 11. O-O Be6 {(Bc8-e6 Nc3-a4 Qd8-d7 Kg1-h2 Bb6-d4
Bc1-b2 Bd4xb2 Na4xb2 b7-b6 d2-d3 Ra8-d8 Qd1-e2 Be6-d5 Bg2xd5 Qd7xd5 c2-c3
f7-f5 Ra1-e1 b6-b5 a3-a4 b5xa4 Nb2xa4 Qd5xd3 b4-b5 Nc6-e7 Qe2xe5 Ne7-g6
Qe5xc7 f5-f4 Na4-c5 f4xg3+ Qc7xg3 Qd3xb5) +0.77/22 2} 12. d3 f5 {(f7-f5
Nc3-a4 f5-f4 Bc1-b2 Be6-d5 Na4xb6 c7xb6 Bg2xd5+ Qd8xd5 g3-g4 Nc6-d4 c2-c4
Qd5-d6 Bb2xd4 Qd6xd4 Qd1-f3 Ra8-d8 Rf1-d1 Rd8-d7 Ra1-c1 Qd4-b2 Rc1-b1
Qb2xa3 Rb1-a1 Qa3xb4 Ra1xa7 Qb4-b2 Ra7xb7 Rd7xb7 Qf3xb7) +0.55/25 30} 13.
Na4 f4 {(f5-f4 Bc1-b2 Be6-d5 Na4xb6 a7xb6 Bg2-e4 Bd5xe4 d3xe4 Qd8-g5 Qd1-f3
Nc6-d4 Bb2xd4 e5xd4 g3-g4 Qg5-e5 a3-a4 Kg8-h8 Rf1-d1 Rf8-e8 Qf3-d3 c7-c5
a4-a5 b6xa5 b4xc5 Qe5xc5 Qd3xd4 Qc5xc2 e4-e5 f4-f3) +0.60/26 0} 14. Bb2 Bd5
{(Be6-d5 Na4xb6 a7xb6 Bg2-e4 Bd5xe4 d3xe4 Nc6-d4 Bb2xd4 e5xd4 Qd1-f3 Qd8-g5
g3-g4 Ra8-d8 Ra1-c1 Qg5-e5 Rf1-d1 g7-g5 a3-a4 Kg8-g7 c2-c3 c7-c6 c3xd4
Rd8xd4 Rd1xd4 Qe5xd4 Qf3-c3 Rf8-d8 Kg1-g2 Qd4xc3) +0.64/24 0} 15. Bxd5+
Qxd5 {(Qd8xd5 Na4xb6 a7xb6 Qd1-e2 Nc6-d4 Bb2xd4 e5xd4 Ra1-e1 Qd5-d7 Qe2-e6+
Qd7xe6 Re1xe6 Ra8xa3 Re6-e7 Ra3-c3 Rf1-c1 Rc3-c6 Re7-d7 f4xg3 f2xg3)
+0.90/22 5} 16. g4 Nd4 {(Nc6-d4 Bb2xd4 e5xd4 Rf1-e1 Qd5-c6 Na4xb6 a7xb6
Re1-e4 Ra8-e8 Re4xd4 h7-h5 Ra1-c1 f4-f3 Qd1-d2 Qc6-f6 Rd4-e4 Re8xe4 d3xe4
h5xg4 Qd2-d5+ Rf8-f7 h3xg4 c7-c6 Qd5-h5 Qf6-f4 Rc1-e1 Qf4-d2 Re1-f1 Qd2xc2
e4-e5 Rf7-f8 Rf1-e1 Qc2-c3) +0.50/26 28} 17. Bxd4 exd4 {(e5xd4 Rf1-e1
Qd5-c6 Na4xb6 a7xb6 Re1-e4 Ra8-e8 Re4xd4 h7-h5 Rd4-e4 Re8xe4 d3xe4 Qc6xe4
g4xh5 Rf8-f5 Kg1-h2 Qe4-e5 Qd1-e1 Rf5xh5 Qe1xe5 Rh5xe5 Ra1-d1 Re5-e2 Kh2-g2
Re2xc2 Rd1-d4 b6-b5 Rd4xf4 Rc2-c3 Rf4-e4 Kg8-f7 Re4-f4+ Kf7-e6 Rf4-g4
Rc3xa3 Rg4xg7) +0.18/27 3} 18. Re1 Qc6 {(Qd5-c6 Na4xb6 a7xb6 Re1-e5 Ra8-e8
Qd1-e2 f4-f3 Qe2-e4 Qc6xe4 Re5xe4 Re8xe4 d3xe4 Rf8-e8 Kg1-h2 Re8xe4 Kh2-g3
Re4-e2 c2-c3 d4xc3 Ra1-c1 b6-b5 Rc1xc3 Re2-e7 Rc3xf3 Re7-d7 Rf3-c3 Kg8-f7
Rc3-e3 Kf7-f6 h3-h4 c7-c6 f2-f4 g7-g6 g4-g5+ Kf6-f7) +0.11/28 6} 19. Nxb6
axb6 {(a7xb6 Re1-e4 Ra8-e8 Re4xd4 h7-h5 Rd4-e4 Re8xe4 d3xe4 Qc6xe4 g4xh5
Rf8-f5 Kg1-h2 Qe4-e5 Qd1-e1 Rf5xh5 Ra1-d1 Qe5xe1 Rd1xe1 Kg8-f7 Re1-d1
Kf7-e6 Kh2-g2 Rh5-h8 Rd1-e1+ Ke6-d7 Re1-d1+ Kd7-e7 Rd1-e1+ Ke7-d7) 0.00/30
3} 20. Re4 Rae8 {(Ra8-e8 Re4xd4 h7-h5 Rd4-e4 Re8xe4 d3xe4 Qc6xe4 g4xh5
Rf8-f5 Kg1-h2 Qe4-e5 Qd1-e1 Rf5xh5 Ra1-d1 Qe5-f5 Rd1-d8+ Kg8-h7 Qe1-c3
f4-f3 Qc3-d3 Rh5xh3+ Kh2-g1 Qf5xd3 Rd8xd3 Rh3-h4 Rd3-c3 c7-c6 Rc3xf3 b6-b5
Rf3-d3 Rh4-c4 Rd3-d7 Kh7-g6 Rd7xb7 Rc4xc2 Rb7-d7 Rc2-e2) +0.13/29 5} 21.
Rxd4 h5 {(h7-h5 g4xh5) +0.03/30 0} 22. gxh5 Re5 {(Re8-e5) +0.10/28 0} 23.
Re4 Rg5+ {(Re5-g5+ Kg1-h2 f4-f3 Re4-g4 Qc6-d6+ Kh2-h1 Rg5xh5 Rg4-g3 Qd6-e6
Kh1-h2 Rh5-e5 Rg3xf3 Re5-f5 Rf3xf5 Rf8xf5 Qd1-e1 Qe6-f7 Kh2-g1 Rf5-g5+
Kg1-f1 Qf7-g6 Qe1-c3 Rg5-g1+ Kf1-e2 Qg6-e6+ Ke2-d2 Qe6-h6+ Kd2-e2) 0.00/30
10} 24. Kh2 f3 {(f4-f3 Re4-g4 Qc6-d6+ Kh2-h1 Rg5xh5 Rg4-g3 Qd6-e6 Kh1-h2
Rh5-e5 Rg3xf3 Re5-f5 Rf3xf5 Rf8xf5 Qd1-e1 Qe6-f7 Kh2-g1 Rf5-g5+ Kg1-f1
Qf7-g6 Qe1-c3 Rg5-g1+ Kf1-e2 Qg6-e6+ Ke2-d2 Qe6-h6+ Kd2-e2 Qh6-h5+) 0.00/33
3} 25. Rg4 Qd6+ {(Qc6-d6+ Kh2-h1 Rg5xh5 Rg4-g3 Qd6-e6 Kh1-h2 Rh5-e5 Rg3xf3
Re5-f5 Rf3xf5 Rf8xf5 Qd1-e1 Qe6-f7 Kh2-g1 Qf7-g6+ Kg1-h2) 0.00/33 6} 26.
Kh1 Rxh5 {(Rg5xh5 Rg4-g3 Qd6-e6 Kh1-h2 Rh5-e5 Rg3xf3 Re5-f5 Rf3xf5 Rf8xf5
Qd1-e1 Qe6-f7 Kh2-g1 Rf5-g5+ Kg1-f1 Qf7-g6 Qe1-c3 Rg5-g1+ Kf1-e2 Qg6-h5+
Ke2-e3 Qh5xh3+ Ke3-d2 Qh3-h6+ Kd2-e2 Qh6-e6+ Ke2-d2) 0.00/35 17} 27. Rg3
Rf6 {(Rf8-f6 d3-d4 Qd6-c6 Kh1-h2 Qc6-d6) 0.00/33 0} 28. d4 Qd5 {(Qd6-d5
Qd1-d3 Rf6-h6 Kh1-g1 Rh5xh3 Rg3xh3 Rh6xh3 Qd3-b3 Qd5xb3 c2xb3 Rh3-h4 Ra1-d1
Rh4-g4+ Kg1-f1 Rg4-h4) 0.00/31 2} 29. Qd3 Rfh6 {(Rf6-h6 Kh1-g1 Rh5xh3
Rg3xh3 Rh6xh3 Qd3-b3 Qd5xb3 c2xb3 Rh3-h4 Ra1-d1 Rh4-g4+ Kg1-f1 Rg4-h4)
0.00/33 3} 30. Kg1 Rxh3 {(Rh5xh3 Rg3xh3 Rh6xh3 Qd3-b3 Qd5xb3 c2xb3 Rh3-h4
b4-b5 Rh4-g4+ Kg1-h2 Rg4-g2+ Kh2-h3 Rg2xf2 Kh3-g3 Rf2-c2 Kg3xf3 Kg8-f7
a3-a4 Kf7-e6 Kf3-e4 g7-g5 Ra1-g1 Rc2-b2 Rg1xg5 Rb2xb3 Rg5-g7 Rb3-c3 d4-d5+
Ke6-d6 Ke4-d4 Rc3-c1 Rg7-g6+ Kd6-d7 Rg6-g7+ Kd7-d8 d5-d6 Rc1-d1+ Kd4-c4
Rd1xd6 Rg7-g8+ Kd8-d7 Rg8-g7+) 0.00/31 6} 31. Rxh3 Rxh3 {(Rh6xh3 Qd3-b3
Qd5xb3 c2xb3 Rh3-h4 b4-b5 Rh4-g4+ Kg1-h2 Rg4-g2+ Kh2-h3 Rg2xf2 Kh3-g3
Rf2-b2 Ra1-c1 f3-f2 Kg3-g2 Rb2xb3 Kg2xf2 c7-c6 b5xc6 b7xc6 Rc1xc6 Rb3xa3
Rc6xb6 Kg8-f7 Rb6-b5 Kf7-f6 Rb5-b6+ Kf6-f5 d4-d5 Ra3-d3 d5-d6 g7-g5 Rb6-b7
Rd3-d2+ Kf2-g3 Rd2xd6 Rb7-b5+ Kf5-e6 Rb5xg5 Rd6-d3+ Kg3-g2 Rd3-a3 Rg5-b5
Ke6-f6 Rb5-b6+ Kf6-e5 Rb6-b5+ Ke5-e4) +0.13/34 2} 32. Qb3 Qxb3 {(Qd5xb3
c2xb3 Rh3-h4 Ra1-c1 Rh4-g4+ Kg1-h1 Rg4-g2 Rc1xc7 Rg2xf2 Kh1-g1 Rf2-b2 d4-d5
g7-g5 d5-d6 Rb2-d2 Rc7-c3 Kg8-f7 Rc3xf3+ Kf7-e6 Rf3-e3+ Ke6xd6 Re3-g3
Rd2-b2 Kg1-f1 Rb2-a2 Rg3xg5 Ra2xa3 Rg5-g3 Ra3-a8 Kf1-e2 b6-b5 Rg3-g7 Kd6-c6
Ke2-d3 Ra8-d8+ Kd3-c3) +0.13/34 14} 33. cxb3 Rh4 {(Rh3-h4 Ra1-c1 Rh4-g4+
Kg1-h1 Rg4-g2 Rc1xc7 g7-g5 Rc7-c1 Rg2xf2 Rc1-g1 Rf2-g2 Rg1xg2 f3xg2+ Kh1xg2
Kg8-f7 b4-b5 Kf7-e6 a3-a4 Ke6-d5 b3-b4 Kd5-c4 d4-d5 Kc4xd5 a4-a5 b6xa5
b4xa5 Kd5-c5 a5-a6 Kc5-b6 a6xb7 Kb6xb7 Kg2-f3 Kb7-b6 Kf3-g4 Kb6xb5 Kg4-g3
Kb5-a6 Kg3-f3 Ka6-b6 Kf3-g4 Kb6-c6 Kg4-f5 g5-g4 Kf5-f4) +0.13/33 13} 34.
Rc1 Rg4+ {(Rh4-g4+ Kg1-h1 Rg4-g2 Rc1xc7 g7-g5 Rc7-c1 Rg2xf2 Rc1-g1 Rf2-g2
Rg1xg2 f3xg2+ Kh1xg2 Kg8-f7 Kg2-f3 Kf7-e6 Kf3-e4 b6-b5 a3-a4 Ke6-d6 a4xb5
b7-b6 Ke4-f5 Kd6-d5 Kf5xg5 Kd5xd4 Kg5-f4 Kd4-c3 Kf4-e3 Kc3xb4 Ke3-d2 Kb4xb3
Kd2-d3 Kb3-b4 Kd3-c2 Kb4xb5 Kc2-b3 Kb5-c5 Kb3-c3 b6-b5 Kc3-b3 b5-b4 Kb3-a2
Kc5-c4 Ka2-b2 b4-b3 Kb2-b1 b3-b2) +0.13/35 13} 35. Kh1 Rg2 {(Rg4-g2 Rc1xc7
g7-g5 Rc7-c1 Rg2xf2 Rc1-g1 Rf2-g2 Rg1xg2 f3xg2+ Kh1xg2 Kg8-f7 Kg2-f3 Kf7-e6
Kf3-e4 b6-b5 a3-a4 Ke6-d6 d4-d5 g5-g4 a4xb5 g4-g3 Ke4-f3 Kd6xd5 Kf3xg3
b7-b6 Kg3-g4 Kd5-e4 Kg4-g5 Ke4-d3 Kg5-f4 Kd3-d4 Kf4-f3 Kd4-c3 Kf3-e4 Kc3xb3
Ke4-d3 Kb3xb4 Kd3-c2 Kb4xb5 Kc2-b3 Kb5-c5 Kb3-c3 b6-b5 Kc3-b3 b5-b4 Kb3-a2)
+0.13/34 6} 36. Rxc7 g5 {(g7-g5 Rc7-c1 Rg2xf2 Rc1-g1 Rf2-g2 Rg1xg2 f3xg2+
Kh1xg2 Kg8-f7 Kg2-f3 Kf7-e6 Kf3-e4 b6-b5 a3-a4 Ke6-d6 a4xb5 b7-b6 Ke4-f5
Kd6-d5 Kf5xg5 Kd5xd4 Kg5-f4 Kd4-c3 Kf4-e4 Kc3xb3 Ke4-d3 Kb3xb4 Kd3-c2
Kb4xb5 Kc2-b3 Kb5-c5 Kb3-c3 b6-b5 Kc3-b3 b5-b4 Kb3-a2 Kc5-c4 Ka2-b2 b4-b3
Kb2-b1 Kc4-c3 Kb1-a1 Kc3-d2 Ka1-b1 Kd2-e2 Kb1-a1 b3-b2+ Ka1xb2 Ke2-f2
Kb2-a3 Kf2-g2 Ka3-b4 Kg2-h1 Kb4-b5) +0.13/35 6} 37. Rc1 Rxf2 {(Rg2xf2
Rc1-g1 Rf2-g2 Rg1xg2 f3xg2+ Kh1xg2 Kg8-f7 Kg2-f3 Kf7-e6 Kf3-e4 b6-b5 a3-a4
g5-g4 a4xb5 g4-g3 Ke4-f3 Ke6-d5 Kf3xg3 Kd5xd4 b5-b6 Kd4-c3 Kg3-f4 Kc3xb4
Kf4-e4 Kb4-c5 Ke4-d3 Kc5xb6 Kd3-c4 Kb6-a5 Kc4-c5 b7-b5 Kc5-c6 Ka5-b4 Kc6-b6
Kb4xb3 Kb6xb5 Kb3-c3 Kb5-a4 Kc3-c4 Ka4-a5 Kc4-c5 Ka5-a4) 0.00/37 2} 38. Rg1
Rg2 {(Rf2-g2 Rg1xg2 f3xg2+ Kh1xg2 Kg8-f7 Kg2-f3 Kf7-e6 b4-b5 Ke6-d5 Kf3-e3
g5-g4 a3-a4 g4-g3 b3-b4 Kd5-c4 d4-d5 Kc4xd5 a4-a5 b6xa5 b4xa5 Kd5-c5 a5-a6
b7xa6 b5xa6 g3-g2 Ke3-f2 Kc5-b6 Kf2xg2 Kb6xa6 Kg2-f2 Ka6-a5 Kf2-e3 Ka5-b6
Ke3-e4 Kb6-a5) 0.00/38 4} 39. b5 Kf7 {(Kg8-f7 Rg1xg2 f3xg2+ Kh1xg2 Kf7-e6
a3-a4 Ke6-d5 b3-b4 Kd5xd4 a4-a5 b6xa5 b4xa5 Kd4-c5 a5-a6 b7xa6 b5xa6 Kc5-c6
Kg2-f2 Kc6-b6 Kf2-f3 Kb6xa6 Kf3-g4 Ka6-b6 Kg4xg5 Kb6-a7 Kg5-f4 Ka7-b7
Kf4-f5 Kb7-c7 Kf5-e6 Kc7-c6 Ke6-f7 Kc6-d5 Kf7-e7 Kd5-c5 Ke7-f8 Kc5-d4
Kf8-e8 Kd4-c3 Ke8-e7 Kc3-d3 Ke7-d7 Kd3-e4 Kd7-c6 Ke4-d3) 0.00/38 15} 40.
Rxg2 fxg2+ {(f3xg2+ Kh1xg2 Kf7-e6 a3-a4 Ke6-d5 b3-b4 Kd5xd4 a4-a5 b6xa5
b4xa5 Kd4-c5 a5-a6 b7xa6 b5xa6 Kc5-c6 Kg2-f2 Kc6-b6 Kf2-f3 Kb6xa6 Kf3-g4
Ka6-a7 Kg4-f5 g5-g4 Kf5xg4 Ka7-b7 Kg4-g5 Kb7-a8 Kg5-f6 Ka8-a7 Kf6-f5 Ka7-b8
Kf5-e4 Kb8-a8 Ke4-d5 Ka8-b7 Kd5-d6 Kb7-a7 Kd6-d7 Ka7-b6 Kd7-e7 Kb6-c5
Ke7-f8 Kc5-c4 Kf8-e8 Kc4-b4 Ke8-f7 Kb4-a5 Kf7-f6 Ka5-b5 Kf6-e5 Kb5-c4
Ke5-f4 Kc4-d5 Kf4-g5 Kd5-e5 Kg5-g6 Ke5-d4) 0.00/48 0} 41. Kxg2 Ke6 {(Kf7-e6
a3-a4 Ke6-d5 b3-b4 Kd5xd4 a4-a5 b6xa5 b4xa5 Kd4-c5 a5-a6 b7xa6 b5xa6 Kc5-c6
Kg2-f2 Kc6-b6 Kf2-f3 Kb6xa6 Kf3-g4 Ka6-a7 Kg4-f5 Ka7-a8 Kf5-e5 Ka8-b7
Ke5-e4 Kb7-c7 Ke4-f3 Kc7-b6 Kf3-g3 Kb6-a7 Kg3-h2 Ka7-a6 Kh2-h3 Ka6-b7
Kh3-g2 Kb7-c8 Kg2-f3 Kc8-b8 Kf3-e3 g5-g4 Ke3-d4 Kb8-b7 Kd4-d3 Kb7-a6 Kd3-e3
Ka6-b5 Ke3-e4 Kb5-c5 Ke4-f4 g4-g3 Kf4-f3 Kc5-b6 Kf3xg3 Kb6-b7 Kg3-g4 Kb7-b8
Kg4-g5 Kb8-c7 Kg5-f6 Kc7-b7) 0.00/54 13} 42. Kf3 Kd5 {(Ke6-d5 Kf3-e3 g5-g4
a3-a4 g4-g3 b3-b4 Kd5-c4 d4-d5 Kc4xd5 a4-a5 b6xa5 b4xa5 Kd5-c5 a5-a6 b7xa6
b5xa6 g3-g2 Ke3-f2 Kc5-b6 a6-a7 Kb6xa7 Kf2xg2 Ka7-b7 Kg2-f3 Kb7-c6 Kf3-g2
Kc6-d7 Kg2-g3 Kd7-c8 Kg3-f4 Kc8-c7 Kf4-e5 Kc7-b6 Ke5-e6 Kb6-a5 Ke6-f5
Ka5-b4 Kf5-e4 Kb4-b3 Ke4-e5 Kb3-b2 Ke5-d6 Kb2-c3 Kd6-e7 Kc3-b3 Ke7-f7
Kb3-c2 Kf7-f6 Kc2-d3 Kf6-e5 Kd3-e2 Ke5-f4 Ke2-f2 Kf4-g5 Kf2-e3 Kg5-g6
Ke3-d3 Kg6-f6 Kd3-d4) 0.00/52 13} 43. Ke3 g4 {(g5-g4 a3-a4 g4-g3 b3-b4
Kd5-c4 d4-d5 Kc4xd5 a4-a5 b6xa5 b4xa5 Kd5-c5 a5-a6 b7xa6 b5xa6 g3-g2 Ke3-f2
Kc5-b6 a6-a7 Kb6xa7 Kf2xg2 Ka7-b8 Kg2-g3 Kb8-c7 Kg3-f3 Kc7-b7 Kf3-g4 Kb7-b8
Kg4-g5 Kb8-c8 Kg5-f4 Kc8-d8 Kf4-e5 Kd8-c7 Ke5-e6 Kc7-b8 Ke6-d7 Kb8-a8
Kd7-c6 Ka8-a7 Kc6-d6 Ka7-b7 Kd6-e7 Kb7-c6 Ke7-f8 Kc6-b5 Kf8-g7 Kb5-b6
Kg7-f6 Kb6-a6 Kf6-g6 Ka6-b7 Kg6-f6 Kb7-b6 Kf6-g7 Kb6-a5 Kg7-g6 Ka5-a6
Kg6-h6 Ka6-a7 Kh6-g7) 0.00/52 5} 44. a4 g3 {(g4-g3 b3-b4 Kd5-c4 d4-d5
Kc4xd5 a4-a5 b6xa5 b4xa5 Kd5-c5 a5-a6 b7xa6 b5xa6 g3-g2 Ke3-f2 Kc5-b6 a6-a7
Kb6xa7 Kf2xg2 Ka7-a8 Kg2-h2 Ka8-b7 Kh2-h1 Kb7-b8 Kh1-g1 Kb8-a7 Kg1-f2
Ka7-b7 Kf2-e3 Kb7-c7 Ke3-e4 Kc7-b8 Ke4-f3 Kb8-c8 Kf3-f4 Kc8-d7 Kf4-e5
Kd7-d8 Ke5-d6 Kd8-c8 Kd6-e7 Kc8-c7 Ke7-f6 Kc7-d6 Kf6-g6 Kd6-d5 Kg6-f5
Kd5-c6 Kf5-e4 Kc6-c5 Ke4-d3 Kc5-b6 Kd3-d2 Kb6-b5 Kd2-c3 Kb5-c6 Kc3-b2
Kc6-b6 Kb2-a3 Kb6-c7 Ka3-a4 Kc7-d7) 0.00/54 5} 45. b4 Kc4 {(Kd5-c4 d4-d5
Kc4xd5 a4-a5 b6xa5 b4xa5 Kd5-c5 a5-a6 b7xa6 b5xa6 g3-g2 Ke3-f2 Kc5-b6 a6-a7
Kb6xa7 Kf2xg2 Ka7-a8 Kg2-h2 Ka8-b8 Kh2-g3 Kb8-a7 Kg3-f3 Ka7-b7 Kf3-f2
Kb7-b8 Kf2-f3 Kb8-c8 Kf3-g4 Kc8-c7 Kg4-h3 Kc7-c6 Kh3-g2 Kc6-b5 Kg2-f1
Kb5-c6 Kf1-e2 Kc6-d7 Ke2-d1 Kd7-c6 Kd1-c1 Kc6-c5 Kc1-b2 Kc5-b5 Kb2-a1
Kb5-a4 Ka1-b2 Ka4-a5 Kb2-b3 Ka5-b5 Kb3-c3 Kb5-b6 Kc3-d4 Kb6-c6 Kd4-e4
Kc6-d7 Ke4-f4 Kd7-e6 Kf4-g3 Ke6-f6 Kg3-g2 Kf6-g5 Kg2-f1) 0.00/56 6} 46. d5
Kxd5 {(Kc4xd5 a4-a5 b6xa5 b4xa5 Kd5-c5 a5-a6 b7xa6 b5xa6 g3-g2 Ke3-f2
Kc5-b6 a6-a7 Kb6xa7 Kf2xg2 Ka7-b7 Kg2-f2 Kb7-b8 Kf2-g1 Kb8-c7 Kg1-f2 Kc7-c8
Kf2-e1 Kc8-d7 Ke1-d2 Kd7-e6 Kd2-c1 Ke6-d6 Kc1-c2 Kd6-c5 Kc2-d2 Kc5-b4
Kd2-d1 Kb4-c3 Kd1-e2 Kc3-b3 Ke2-f3 Kb3-c4 Kf3-g2 Kc4-d4 Kg2-f1 Kd4-c5
Kf1-f2 Kc5-c4 Kf2-f3 Kc4-b3 Kf3-g2 Kb3-c3 Kg2-g1 Kc3-d3 Kg1-f1 Kd3-d4
Kf1-g2 Kd4-e4 Kg2-f1 Ke4-d3 Kf1-g1 Kd3-c3 Kg1-f2 Kc3-b2 Kf2-g3 Kb2-c1
Kg3-g4) 0.00/57 6} 47. a5 g2 {(g3-g2 Ke3-f2 b6xa5 b4xa5 Kd5-d6 Kf2xg2
Kd6-c5 a5-a6 b7xa6 b5xa6 Kc5-b6 Kg2-f2 Kb6-a7 Kf2-f3 Ka7xa6 Kf3-e3 Ka6-a5
Ke3-d2 Ka5-b4 Kd2-e3 Kb4-c5 Ke3-d3 Kc5-b6 Kd3-c4 Kb6-c7 Kc4-c3 Kc7-d7
Kc3-d3 Kd7-e7 Kd3-d2 Ke7-d6 Kd2-c3 Kd6-e5 Kc3-d3 Ke5-f6 Kd3-c2 Kf6-g5
Kc2-b2 Kg5-g6 Kb2-c3 Kg6-f7 Kc3-d4 Kf7-f6 Kd4-c5 Kf6-e6 Kc5-c4 Ke6-f7
Kc4-c3 Kf7-g6 Kc3-d2 Kg6-g7 Kd2-c2 Kg7-f8 Kc2-b1 Kf8-e7 Kb1-a2 Ke7-d7
Ka2-a3 Kd7-c7 Ka3-b2 Kc7-d6) 0.00/57 11} 48. Kf2 bxa5 {(b6xa5 b4xa5 g2-g1Q+
Kf2xg1 Kd5-c5 a5-a6 b7xa6 b5xa6 Kc5-c6 a6-a7 Kc6-b7 a7-a8Q+ Kb7xa8 Kg1-f1
Ka8-b7 Kf1-f2 Kb7-a6 Kf2-e3 Ka6-b6 Ke3-f3 Kb6-c7 Kf3-g3 Kc7-c8 Kg3-f4
Kc8-c7 Kf4-f5 Kc7-c6 Kf5-e6 Kc6-b7 Ke6-d7 Kb7-b6 Kd7-e8 Kb6-c5 Ke8-f8
Kc5-c4 Kf8-g8 Kc4-b3 Kg8-h7 Kb3-b4 Kh7-h6 Kb4-c5 Kh6-g7 Kc5-b6 Kg7-f6
Kb6-c7 Kf6-f7 Kc7-b8 Kf7-e8 Kb8-a8 Ke8-d7 Ka8-b7 Kd7-e6 Kb7-c6 Ke6-e7
Kc6-b5 Ke7-d8 Kb5-a6 Kd8-e8 Ka6-b6 Ke8-e7 Kb6-c6 Ke7-f7 Kc6-b5 Kf7-g6
Kb5-a4 Kg6-f6) 0.00/61 11} 49. bxa5 g1=Q+ {(g2-g1Q+ Kf2xg1 Kd5-c5 a5-a6
b7xa6 b5xa6 Kc5-b6 a6-a7 Kb6xa7 Kg1-f2 Ka7-b6 Kf2-f3 Kb6-b7 Kf3-g4 Kb7-a7
Kg4-g3 Ka7-a6 Kg3-h3 Ka6-b6 Kh3-h2 Kb6-c6 Kh2-h3 Kc6-c7 Kh3-g3 Kc7-d7
Kg3-h4 Kd7-c6 Kh4-g5 Kc6-b6 Kg5-f5 Kb6-a7 Kf5-g6 Ka7-b8 Kg6-f7 Kb8-c8
Kf7-g7 Kc8-c7 Kg7-h7 Kc7-b8 Kh7-g8 Kb8-a7 Kg8-f8 Ka7-a6 Kf8-g7 Ka6-b5
Kg7-g6 Kb5-a4 Kg6-f6 Ka4-b3 Kf6-g7 Kb3-c3 Kg7-f7 Kc3-c2 Kf7-f8 Kc2-b2
Kf8-g8 Kb2-c1 Kg8-h7 Kc1-d1 Kh7-h6 Kd1-d2 Kh6-g6 Kd2-c2 Kg6-g5 Kc2-b3)
0.00/62 10} 50. Kxg1 Kc5 {(Kd5-c5 a5-a6 Kc5-b6 a6xb7 Kb6xb7 Kg1-g2 Kb7-c7
Kg2-f3 Kc7-b6 Kf3-f4 Kb6-c5 b5-b6 Kc5-c6 Kf4-g3 Kc6-b7 Kg3-h4 Kb7xb6 Kh4-g5
Kb6-b5 Kg5-g4 Kb5-c6 Kg4-h3 Kc6-c5 Kh3-g2 Kc5-b6 Kg2-h2 Kb6-a6 Kh2-h1
Ka6-b5 Kh1-g1 Kb5-c4 Kg1-f2 Kc4-d5 Kf2-g3 Kd5-d6 Kg3-g2 Kd6-d7 Kg2-f2
Kd7-c7 Kf2-f3 Kc7-b7 Kf3-f4 Kb7-a8 Kf4-g3 Ka8-b8 Kg3-g4 Kb8-a7 Kg4-g5
Ka7-a6 Kg5-g6 Ka6-b5 Kg6-f7 Kb5-c4 Kf7-e7 Kc4-c5 Ke7-e6 Kc5-c6 Ke6-f7
Kc6-d6 Kf7-g7) 0.00/66 4} 51. a6 bxa6 {(b7xa6 b5xa6 Kc5-b6 a6-a7 Kb6xa7
Kg1-f2 Ka7-a8 Kf2-e3 Ka8-b7 Ke3-f3 Kb7-b6 Kf3-f2 Kb6-a7 Kf2-e2 Ka7-a8
Ke2-d3 Ka8-b7 Kd3-d2 Kb7-b6 Kd2-c2 Kb6-a5 Kc2-c3 Ka5-a6 Kc3-d4 Ka6-b6
Kd4-c4 Kb6-c7 Kc4-d4 Kc7-b7 Kd4-d3 Kb7-c6 Kd3-e4 Kc6-b6 Ke4-f4 Kb6-c5
Kf4-e3 Kc5-d6 Ke3-e2 Kd6-c6 Ke2-d2 Kc6-b5 Kd2-d1 Kb5-a6 Kd1-e2 Ka6-a5
Ke2-d3 Ka5-b5 Kd3-c3 Kb5-c5 Kc3-c2 Kc5-b6 Kc2-b3 Kb6-a7 Kb3-a2 Ka7-b7
Ka2-b2 Kb7-c8 Kb2-c1 Kc8-c7 Kc1-d1 Kc7-d7 Kd1-d2 Kd7-d8) 0.00/64 9} 52.
bxa6 Kb6 {(Kc5-b6 a6-a7 Kb6xa7 Kg1-f2 Ka7-b6 Kf2-f3 Kb6-a5 Kf3-g4 Ka5-b4
Kg4-f3 Kb4-c4 Kf3-g3 Kc4-d3 Kg3-g4 Kd3-c2 Kg4-f4 Kc2-c3 Kf4-g5 Kc3-c2
Kg5-f6 Kc2-b1 Kf6-f5 Kb1-b2 Kf5-e5 Kb2-a3 Ke5-d4 Ka3-a2 Kd4-c3 Ka2-a3
Kc3-c2 Ka3-b4 Kc2-d3 Kb4-c5 Kd3-c3 Kc5-b6 Kc3-d3 Kb6-a5 Kd3-c2 Ka5-a6
Kc2-d2 Ka6-a7 Kd2-e1 Ka7-b6 Ke1-d1 Kb6-c7 Kd1-c2 Kc7-d6 Kc2-c3 Kd6-e7
Kc3-b3 Ke7-d7 Kb3-b2 Kd7-d8 Kb2-c3 Kd8-c8 Kc3-d2 Kc8-d7 Kd2-d3 Kd7-e6
Kd3-d4 Ke6-f6 Kd4-e4) 0.00/64 9} 53. Kf2 Ka7 {(Kb6-a7 Kf2-e1 Ka7-b8 a6-a7+
Kb8xa7 Ke1-f2 Ka7-b6 Kf2-f3 Kb6-a5 Kf3-e2 Ka5-b5 Ke2-e3 Kb5-a6 Ke3-f3
Ka6-b6 Kf3-g4 Kb6-a5 Kg4-g3 Ka5-b5 Kg3-g4 Kb5-a4 Kg4-g5 Ka4-b4 Kg5-h4
Kb4-b3 Kh4-g3 Kb3-a3 Kg3-f2 Ka3-b2 Kf2-g2 Kb2-c2 Kg2-g1 Kc2-b3 Kg1-f1
Kb3-b4 Kf1-f2 Kb4-c3 Kf2-e3 Kc3-c2 Ke3-f4 Kc2-b1 Kf4-g3 Kb1-c1 Kg3-g4
Kc1-d1 Kg4-f4 Kd1-c2 Kf4-e5 Kc2-d3 Ke5-f6 Kd3-d2 Kf6-f5 Kd2-e1 Kf5-f4
Ke1-f2 Kf4-e5 Kf2-g2 Ke5-f6 Kg2-f3 Kf6-f7 Kf3-f2 Kf7-f6) 0.00/63 13} Draw
[/quote][/pgn]

[Werewolf]

Lc0 is clearly improving faster then Stockfish at this point in time. Even at 3m+2s time controls vs past matches at the same time controls.

Is this definitely true? The T60 graph has been flat for ages. I know there are other improvements outside of the nets, but this is quite a big thing

[Milos]

Stockfish nnue has a secret weapon. The Kamehameha blast. Of course, he has to go to level 5 before he can use it. You don't just go Kamehameha blasting stuff willy-nilly.

At the very end it will be LCZero Vs Stockfish NNUE, but I predict a very close encounter of the 3rd kind, LCZero from Planet 1140b Vs StockFish NNUE from Planet Earth, Now I am more convinced than ever
https://tcec-chess.com/live.html

I agree. I just played 200 games with Stockfish 12 Vs Lc0 26.2. Stockfish 12 won by only 24 Elo in 200 games at 3m+2s. And in testing. We can see how badly Stockfish NNUE has scaled in past testing. At longer time controls.

Both are the best chess engines, and the winner may only be decided by hardware and time controls.

The sprinter Stockfish 12 vs. the marathon runner Lc0. Who wins the race. May depend on the distance of the race!


Lc0 is clearly improving faster then Stockfish at this point in time. Even at 3m+2s time controls vs past matches at the same time controls.

Code: Select all

Result:
--------------------------------------------------------------------------
  #  name          games    wins   draws  losses   score    los%  elo+/-
  1. Stockfish 12    200      16     182       2   107.0   100.0    24.4
  2. Lc0 v0.26.2     200       2     182      16    93.0     0.0   -24.4

Cross table:
--------------------------------------------------------------------------
  #  name             score   games                                                                                                                                                                                                        1                                                                                                                                                                                                        2
  1. Stockfish 12     107.0     200                                                                                                                                                                                                        x =====1==1===1====================1======1========11========================1==========================1============================================1====1===1=========1==0================1===1=1==0====
  2. Lc0 v0.26.2       93.0     200 =====0==0===0====================0======0========00========================0==========================0============================================0====0===0=========0==1================0===0=0==1====                                                                                                                                                                                                        x

Tech:
--------------------------------------------------------------------------

Tech (average nodes, depths, time/m per move, others per game), counted for computing moves only, ignored moves with zero nodes:
  #  name            nodes/m         NPS  depth/m   time/m    moves     time
  1. Stockfish 12    125173K    26565996     42.5      4.7     54.1    255.1
  2. Lc0 v0.26.2        101K       20342     10.0      4.9     54.1    267.2
     all ---          61216K    12984844     26.3      4.8     54.1    261.2

For years ppl come up with the BS theory that A/B engines tuned in micro-bullet would be weak in LTC and for years they are so bluntly proven wrong. Impact of eval on horizon effects is minimal and it doesn't change whether you search to depth 20 or depth 100. SF-NN search is SF and SF is proven to scale better than Lc0 (and as a matter of fact any MCTS engine) in LTC. Ergo SF-NN scales better than Lc0 in LTC.
Your claims are simply BS reflecting your cluelessness in the matter. You effectively draw conclusions from STC (just because it's not micro-bullet but blitz instead) with a sample size that is a joke.
The result in the superfinal will be much worse sweep than last year. And then ppl like you would be astonished and would come up with all kind of ridiculous excuses to justify what is basically their cluelessness.

[mwyoung]

Stockfish nnue has a secret weapon. The Kamehameha blast. Of course, he has to go to level 5 before he can use it. You don't just go Kamehameha blasting stuff willy-nilly.

At the very end it will be LCZero Vs Stockfish NNUE, but I predict a very close encounter of the 3rd kind, LCZero from Planet 1140b Vs StockFish NNUE from Planet Earth, Now I am more convinced than ever
https://tcec-chess.com/live.html

I agree. I just played 200 games with Stockfish 12 Vs Lc0 26.2. Stockfish 12 won by only 24 Elo in 200 games at 3m+2s. And in testing. We can see how badly Stockfish NNUE has scaled in past testing. At longer time controls.

Both are the best chess engines, and the winner may only be decided by hardware and time controls.

The sprinter Stockfish 12 vs. the marathon runner Lc0. Who wins the race. May depend on the distance of the race!


Lc0 is clearly improving faster then Stockfish at this point in time. Even at 3m+2s time controls vs past matches at the same time controls.

Code: Select all

Result:
--------------------------------------------------------------------------
  #  name          games    wins   draws  losses   score    los%  elo+/-
  1. Stockfish 12    200      16     182       2   107.0   100.0    24.4
  2. Lc0 v0.26.2     200       2     182      16    93.0     0.0   -24.4

Cross table:
--------------------------------------------------------------------------
  #  name             score   games                                                                                                                                                                                                        1                                                                                                                                                                                                        2
  1. Stockfish 12     107.0     200                                                                                                                                                                                                        x =====1==1===1====================1======1========11========================1==========================1============================================1====1===1=========1==0================1===1=1==0====
  2. Lc0 v0.26.2       93.0     200 =====0==0===0====================0======0========00========================0==========================0============================================0====0===0=========0==1================0===0=0==1====                                                                                                                                                                                                        x

Tech:
--------------------------------------------------------------------------

Tech (average nodes, depths, time/m per move, others per game), counted for computing moves only, ignored moves with zero nodes:
  #  name            nodes/m         NPS  depth/m   time/m    moves     time
  1. Stockfish 12    125173K    26565996     42.5      4.7     54.1    255.1
  2. Lc0 v0.26.2        101K       20342     10.0      4.9     54.1    267.2
     all ---          61216K    12984844     26.3      4.8     54.1    261.2

For years ppl come up with the BS theory that A/B engines tuned in micro-bullet would be weak in LTC and for years they are so bluntly proven wrong. Impact of eval on horizon effects is minimal and it doesn't change whether you search to depth 20 or depth 100. SF-NN search is SF and SF is proven to scale better than Lc0 (and as a matter of fact any MCTS engine) in LTC. Ergo SF-NN scales better than Lc0 in LTC.
Your claims are simply BS reflecting your cluelessness in the matter. You effectively draw conclusions from STC (just because it's not micro-bullet but blitz instead) with a sample size that is a joke.
The result in the superfinal will be much worse sweep than last year. And then ppl like you would be astonished and would come up with all kind of ridiculous excuses to justify what is basically their cluelessness.

The only one that is clueless here is you. As I test at the longer time controls, as well as short time controls. Along with 1 core testing, and up to 32 threads.

And I am not talking about A/B engine only testing at micro-bullet. And I never have. I am talking about NNUE! And my sample size is huge. This is not my only test. I test non stop.

My conclusion is what the data is showing us, and if it changes all will see that also. I test openly, and to video.


"SF-NN search is SF and SF is proven to scale better than Lc0"

[mwyoung]

Stockfish nnue has a secret weapon. The Kamehameha blast. Of course, he has to go to level 5 before he can use it. You don't just go Kamehameha blasting stuff willy-nilly.

At the very end it will be LCZero Vs Stockfish NNUE, but I predict a very close encounter of the 3rd kind, LCZero from Planet 1140b Vs StockFish NNUE from Planet Earth, Now I am more convinced than ever
https://tcec-chess.com/live.html

I agree. I just played 200 games with Stockfish 12 Vs Lc0 26.2. Stockfish 12 won by only 24 Elo in 200 games at 3m+2s. And in testing. We can see how badly Stockfish NNUE has scaled in past testing. At longer time controls.

Both are the best chess engines, and the winner may only be decided by hardware and time controls.



The sprinter Stockfish 12 vs. the marathon runner Lc0. Who wins the race. May depend on the distance of the race!


Lc0 is clearly improving faster then Stockfish at this point in time. Even at 3m+2s time controls vs past matches at the same time controls.

Code: Select all

Result:
--------------------------------------------------------------------------
  #  name          games    wins   draws  losses   score    los%  elo+/-
  1. Stockfish 12    200      16     182       2   107.0   100.0    24.4
  2. Lc0 v0.26.2     200       2     182      16    93.0     0.0   -24.4

Cross table:
--------------------------------------------------------------------------
  #  name             score   games                                                                                                                                                                                                        1                                                                                                                                                                                                        2
  1. Stockfish 12     107.0     200                                                                                                                                                                                                        x =====1==1===1====================1======1========11========================1==========================1============================================1====1===1=========1==0================1===1=1==0====
  2. Lc0 v0.26.2       93.0     200 =====0==0===0====================0======0========00========================0==========================0============================================0====0===0=========0==1================0===0=0==1====                                                                                                                                                                                                        x

Tech:
--------------------------------------------------------------------------

Tech (average nodes, depths, time/m per move, others per game), counted for computing moves only, ignored moves with zero nodes:
  #  name            nodes/m         NPS  depth/m   time/m    moves     time
  1. Stockfish 12    125173K    26565996     42.5      4.7     54.1    255.1
  2. Lc0 v0.26.2        101K       20342     10.0      4.9     54.1    267.2
     all ---          61216K    12984844     26.3      4.8     54.1    261.2

For years ppl come up with the BS theory that A/B engines tuned in micro-bullet would be weak in LTC and for years they are so bluntly proven wrong. Impact of eval on horizon effects is minimal and it doesn't change whether you search to depth 20 or depth 100. SF-NN search is SF and SF is proven to scale better than Lc0 (and as a matter of fact any MCTS engine) in LTC. Ergo SF-NN scales better than Lc0 in LTC.
Your claims are simply BS reflecting your cluelessness in the matter. You effectively draw conclusions from STC (just because it's not micro-bullet but blitz instead) with a sample size that is a joke.
The result in the superfinal will be much worse sweep than last year. And then ppl like you would be astonished and would come up with all kind of ridiculous excuses to justify what is basically their cluelessness.

The only one that is clueless here is you. As I test at the longer time controls, as well as short time controls. Along with 1 core testing, and up to 32 threads.

And I am not talking about A/B engine only testing at micro-bullet. And I never have. I am talking about NNUE! And my sample size is huge. This is not my only test. I test non stop.

My conclusion is what the data is showing us, and if it changes all will see that also. I test openly, and to video.


"SF-NN search is SF and SF is proven to scale better than Lc0"

"The result in the superfinal will be much worse sweep than last year. And then ppl like you would be astonished and would come up with all kind of ridiculous excuses to justify what is basically their cluelessness"

For reference here are the results of last season's superfinal...

TCEC Season 184 May 2020 – Jul 2020 Stockfish 202006170741 LCZero v0.25.1-svjio-t60-3972-mlh + 23 = 61 - 16

Code: Select all

Wins	Losses	Draws
23
16
61
Points		Games
53.5
/	
100
Winning percentage
53.5

Elo difference : 
+24

[Jouni]

In CCC current score is same for Classic and NNUE !

[Cornfed]

Quick TCEC question as you guys would know:
Are all 8 remaining engines NN's with the exception of Komodo? Or is Komodo getting lapped by A/B engines as well?

[Leo]

In my own testing SF NNUE is doing great.

[mwyoung]

In my own testing SF NNUE is doing great.

SF NNUE is the best program under most conditions. You are just not going to see crazy Elo performance with more threads, and/or a longer time control. Like you do when testing at bullet or fast blitz. On my system for example it crushes Lc0 under those conditions, and all other engines. But at longer time controls, it can perform under 20 Elo or lower at 15m+15s. For a huge drop in performance.

Some feel maybe this is due to reaching the end of gains for chess engine development. As the other best engines can reach this level with more time, and faster computers. Time will tell in short order.

[rcmaddox]

Stockfish nnue has a secret weapon. The Kamehameha blast. Of course, he has to go to level 5 before he can use it. You don't just go Kamehameha blasting stuff willy-nilly.

This discussion has left me in the dust.