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

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

User avatar
Nordlandia
Posts: 2821
Joined: Fri Sep 25, 2015 9:38 pm
Location: Sortland, Norway

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

Post by Nordlandia »

SF Classicial in combination with probing partial 7-piece syzygy managed to hold this as black. Whereas SF NNUE lost. Komodo consulted 5-piece syzygy.

My theory is that SF Classical is better when it comes to late endgames when EGTB consulting matters. SF NNUE is able to probe EGTB like a magnitude slower because of the lesser speed.

[d]8/R7/2b2kn1/8/6P1/7P/5K2/8 w - - 1 21

[pgn][Event "?"]
[Site "i7-5960X 4.5GHz / 2x 2080 Super OC"]
[Date "2020.10.21"]
[Round "?"]
[White "komodo-14-64bit-bmi2"]
[Black "stockfish_20101813_x64_bmi2"]
[Result "1/2-1/2"]
[FEN "5nk1/pb6/3bp3/2pn3N/P7/8/5PPP/2R1R1K1 w - - 0 1"]
[GameDuration "00:51:28"]
[GameEndTime "2020-10-21T08:43:48.542 Vest-Europa (sommertid)"]
[GameStartTime "2020-10-21T07:52:20.251 Vest-Europa (sommertid)"]
[PlyCount "40"]
[SetUp "1"]
[Termination "adjudication"]
[TimeControl "1800+30"]

1. Ng3 {+0.56/39 56.68s} Nf6 {-0.83/37 50.19s} 2. Red1 {+0.83/41 69.11s}
Be7 {-0.81/43 66.96s} 3. Rb1 {+1.13/40 52.90s} Bc6 {-1.04/48 275.7s}
4. Rb8 {+1.14/37 45.01s} a5 {-0.94/44 27.84s} 5. Rc1 {+1.30/40 102.8s}
Kf7 {-0.96/36 32.20s} 6. Rc8 {+1.24/39 58.81s} Bxa4 {-0.90/41 66.50s}
7. R8xc5 {+1.37/39 65.19s} Bxc5 {-0.83/40 29.21s} 8. Rxc5 {+1.41/38 71.97s}
Ng6 {-0.92/45 217.4s} 9. Rxa5 {+1.71/39 134.3s} Bd7 {-1.17/42 33.75s}
10. Ne2 {+1.73/35 53.60s} e5 {-1.17/44 148.4s} 11. h3 {+1.73/37 52.17s}
Bc6 {-1.17/46 21.94s} 12. Nc3 {+1.78/37 53.06s} Nf4 {-1.23/48 34.56s}
13. f3 {+1.93/39 92.89s} N4d5 {-1.09/48 30.24s} 14. Ra3 {+1.85/36 54.62s}
Ne7 {-1.25/42 42.08s} 15. Kf2 {+2.25/37 52.30s} e4 {0.00/56 35.94s}
16. fxe4 {+2.43/36 39.28s} Nxe4+ {0.00/82 65.34s} 17. Nxe4 {+1.73/36 41.81s}
Bxe4 {0.00/150 38.89s} 18. Ra4 {+1.98/43 120.3s} Bc6 {0.00/46 262.0s}
19. Ra7 {+1.94/41 54.25s} Kf6 {0.00/53 110.5s} 20. g4 {+1.63/43 74.69s}
Ng6 {0.00/60 32.65s, Draw by adjudication: user decision} 1/2-1/2[/pgn]
Top
Chessqueen
Posts: 5582
Joined: Wed Sep 05, 2018 2:16 am
Location: Moving
Full name: Jorge Picado

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

Post by Chessqueen »

Nordlandia wrote: ↑Wed Oct 21, 2020 8:47 am SF Classicial in combination with probing partial 7-piece syzygy managed to hold this as black. Whereas SF NNUE lost. Komodo consulted 5-piece syzygy.

My theory is that SF Classical is better when it comes to late endgames when EGTB consulting matters. SF NNUE is able to probe EGTB like a magnitude slower because of the lesser speed.

[d]8/R7/2b2kn1/8/6P1/7P/5K2/8 w - - 1 21

[pgn][Event "?"]
[Site "i7-5960X 4.5GHz / 2x 2080 Super OC"]
[Date "2020.10.21"]
[Round "?"]
[White "komodo-14-64bit-bmi2"]
[Black "stockfish_20101813_x64_bmi2"]
[Result "1/2-1/2"]
[FEN "5nk1/pb6/3bp3/2pn3N/P7/8/5PPP/2R1R1K1 w - - 0 1"]
[GameDuration "00:51:28"]
[GameEndTime "2020-10-21T08:43:48.542 Vest-Europa (sommertid)"]
[GameStartTime "2020-10-21T07:52:20.251 Vest-Europa (sommertid)"]
[PlyCount "40"]
[SetUp "1"]
[Termination "adjudication"]
[TimeControl "1800+30"]

1. Ng3 {+0.56/39 56.68s} Nf6 {-0.83/37 50.19s} 2. Red1 {+0.83/41 69.11s}
Be7 {-0.81/43 66.96s} 3. Rb1 {+1.13/40 52.90s} Bc6 {-1.04/48 275.7s}
4. Rb8 {+1.14/37 45.01s} a5 {-0.94/44 27.84s} 5. Rc1 {+1.30/40 102.8s}
Kf7 {-0.96/36 32.20s} 6. Rc8 {+1.24/39 58.81s} Bxa4 {-0.90/41 66.50s}
7. R8xc5 {+1.37/39 65.19s} Bxc5 {-0.83/40 29.21s} 8. Rxc5 {+1.41/38 71.97s}
Ng6 {-0.92/45 217.4s} 9. Rxa5 {+1.71/39 134.3s} Bd7 {-1.17/42 33.75s}
10. Ne2 {+1.73/35 53.60s} e5 {-1.17/44 148.4s} 11. h3 {+1.73/37 52.17s}
Bc6 {-1.17/46 21.94s} 12. Nc3 {+1.78/37 53.06s} Nf4 {-1.23/48 34.56s}
13. f3 {+1.93/39 92.89s} N4d5 {-1.09/48 30.24s} 14. Ra3 {+1.85/36 54.62s}
Ne7 {-1.25/42 42.08s} 15. Kf2 {+2.25/37 52.30s} e4 {0.00/56 35.94s}
16. fxe4 {+2.43/36 39.28s} Nxe4+ {0.00/82 65.34s} 17. Nxe4 {+1.73/36 41.81s}
Bxe4 {0.00/150 38.89s} 18. Ra4 {+1.98/43 120.3s} Bc6 {0.00/46 262.0s}
19. Ra7 {+1.94/41 54.25s} Kf6 {0.00/53 110.5s} 20. g4 {+1.63/43 74.69s}
Ng6 {0.00/60 32.65s, Draw by adjudication: user decision} 1/2-1/2[/pgn]
Very soon we will find out if LCZero can beat Stockfish Classical by at least 6 points, but I do NOT think that LCZero will get more than 53 to 47 in favor of LCZero.

PS: I believe that in order to make the 100 games fair they should have 30 games from the start position with no Opening, another 30 games with balanced positions and the rest of 40 games with White having a small advantage, but NOT a HUGE Advantage.
https://tcec-chess.com/live.html
Do NOT worry and be happy, we all live a short life :roll:
Top
Chessqueen
Posts: 5582
Joined: Wed Sep 05, 2018 2:16 am
Location: Moving
Full name: Jorge Picado

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

Post by Chessqueen »

Chessqueen wrote: ↑Thu Oct 22, 2020 5:46 pm
Nordlandia wrote: ↑Wed Oct 21, 2020 8:47 am SF Classicial in combination with probing partial 7-piece syzygy managed to hold this as black. Whereas SF NNUE lost. Komodo consulted 5-piece syzygy.

My theory is that SF Classical is better when it comes to late endgames when EGTB consulting matters. SF NNUE is able to probe EGTB like a magnitude slower because of the lesser speed.

[d]8/R7/2b2kn1/8/6P1/7P/5K2/8 w - - 1 21

[pgn][Event "?"]
[Site "i7-5960X 4.5GHz / 2x 2080 Super OC"]
[Date "2020.10.21"]
[Round "?"]
[White "komodo-14-64bit-bmi2"]
[Black "stockfish_20101813_x64_bmi2"]
[Result "1/2-1/2"]
[FEN "5nk1/pb6/3bp3/2pn3N/P7/8/5PPP/2R1R1K1 w - - 0 1"]
[GameDuration "00:51:28"]
[GameEndTime "2020-10-21T08:43:48.542 Vest-Europa (sommertid)"]
[GameStartTime "2020-10-21T07:52:20.251 Vest-Europa (sommertid)"]
[PlyCount "40"]
[SetUp "1"]
[Termination "adjudication"]
[TimeControl "1800+30"]

1. Ng3 {+0.56/39 56.68s} Nf6 {-0.83/37 50.19s} 2. Red1 {+0.83/41 69.11s}
Be7 {-0.81/43 66.96s} 3. Rb1 {+1.13/40 52.90s} Bc6 {-1.04/48 275.7s}
4. Rb8 {+1.14/37 45.01s} a5 {-0.94/44 27.84s} 5. Rc1 {+1.30/40 102.8s}
Kf7 {-0.96/36 32.20s} 6. Rc8 {+1.24/39 58.81s} Bxa4 {-0.90/41 66.50s}
7. R8xc5 {+1.37/39 65.19s} Bxc5 {-0.83/40 29.21s} 8. Rxc5 {+1.41/38 71.97s}
Ng6 {-0.92/45 217.4s} 9. Rxa5 {+1.71/39 134.3s} Bd7 {-1.17/42 33.75s}
10. Ne2 {+1.73/35 53.60s} e5 {-1.17/44 148.4s} 11. h3 {+1.73/37 52.17s}
Bc6 {-1.17/46 21.94s} 12. Nc3 {+1.78/37 53.06s} Nf4 {-1.23/48 34.56s}
13. f3 {+1.93/39 92.89s} N4d5 {-1.09/48 30.24s} 14. Ra3 {+1.85/36 54.62s}
Ne7 {-1.25/42 42.08s} 15. Kf2 {+2.25/37 52.30s} e4 {0.00/56 35.94s}
16. fxe4 {+2.43/36 39.28s} Nxe4+ {0.00/82 65.34s} 17. Nxe4 {+1.73/36 41.81s}
Bxe4 {0.00/150 38.89s} 18. Ra4 {+1.98/43 120.3s} Bc6 {0.00/46 262.0s}
19. Ra7 {+1.94/41 54.25s} Kf6 {0.00/53 110.5s} 20. g4 {+1.63/43 74.69s}
Ng6 {0.00/60 32.65s, Draw by adjudication: user decision} 1/2-1/2[/pgn]
Very soon we will find out if LCZero can beat Stockfish Classical by at least 6 points, but I do NOT think that LCZero will get more than 53 to 47 in favor of LCZero.

PS: I believe that in order to make the 100 games fair they should have 30 games from the start position with no Opening, another 30 games with balanced positions and the rest of 40 games with White having a small advantage, but NOT a HUGE Advantage.
https://tcec-chess.com/live.html
I really wonder how the score would have ended if 50 games were played from the starting position with No Opening for SF NNUE Vs LCZero.?
Do NOT worry and be happy, we all live a short life :roll:
Top
Chessqueen
Posts: 5582
Joined: Wed Sep 05, 2018 2:16 am
Location: Moving
Full name: Jorge Picado

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

Post by Chessqueen »

Chessqueen wrote: ↑Sat Oct 24, 2020 7:16 am
Chessqueen wrote: ↑Thu Oct 22, 2020 5:46 pm
Nordlandia wrote: ↑Wed Oct 21, 2020 8:47 am SF Classicial in combination with probing partial 7-piece syzygy managed to hold this as black. Whereas SF NNUE lost. Komodo consulted 5-piece syzygy.

My theory is that SF Classical is better when it comes to late endgames when EGTB consulting matters. SF NNUE is able to probe EGTB like a magnitude slower because of the lesser speed.

[d]8/R7/2b2kn1/8/6P1/7P/5K2/8 w - - 1 21

[pgn][Event "?"]
[Site "i7-5960X 4.5GHz / 2x 2080 Super OC"]
[Date "2020.10.21"]
[Round "?"]
[White "komodo-14-64bit-bmi2"]
[Black "stockfish_20101813_x64_bmi2"]
[Result "1/2-1/2"]
[FEN "5nk1/pb6/3bp3/2pn3N/P7/8/5PPP/2R1R1K1 w - - 0 1"]
[GameDuration "00:51:28"]
[GameEndTime "2020-10-21T08:43:48.542 Vest-Europa (sommertid)"]
[GameStartTime "2020-10-21T07:52:20.251 Vest-Europa (sommertid)"]
[PlyCount "40"]
[SetUp "1"]
[Termination "adjudication"]
[TimeControl "1800+30"]

1. Ng3 {+0.56/39 56.68s} Nf6 {-0.83/37 50.19s} 2. Red1 {+0.83/41 69.11s}
Be7 {-0.81/43 66.96s} 3. Rb1 {+1.13/40 52.90s} Bc6 {-1.04/48 275.7s}
4. Rb8 {+1.14/37 45.01s} a5 {-0.94/44 27.84s} 5. Rc1 {+1.30/40 102.8s}
Kf7 {-0.96/36 32.20s} 6. Rc8 {+1.24/39 58.81s} Bxa4 {-0.90/41 66.50s}
7. R8xc5 {+1.37/39 65.19s} Bxc5 {-0.83/40 29.21s} 8. Rxc5 {+1.41/38 71.97s}
Ng6 {-0.92/45 217.4s} 9. Rxa5 {+1.71/39 134.3s} Bd7 {-1.17/42 33.75s}
10. Ne2 {+1.73/35 53.60s} e5 {-1.17/44 148.4s} 11. h3 {+1.73/37 52.17s}
Bc6 {-1.17/46 21.94s} 12. Nc3 {+1.78/37 53.06s} Nf4 {-1.23/48 34.56s}
13. f3 {+1.93/39 92.89s} N4d5 {-1.09/48 30.24s} 14. Ra3 {+1.85/36 54.62s}
Ne7 {-1.25/42 42.08s} 15. Kf2 {+2.25/37 52.30s} e4 {0.00/56 35.94s}
16. fxe4 {+2.43/36 39.28s} Nxe4+ {0.00/82 65.34s} 17. Nxe4 {+1.73/36 41.81s}
Bxe4 {0.00/150 38.89s} 18. Ra4 {+1.98/43 120.3s} Bc6 {0.00/46 262.0s}
19. Ra7 {+1.94/41 54.25s} Kf6 {0.00/53 110.5s} 20. g4 {+1.63/43 74.69s}
Ng6 {0.00/60 32.65s, Draw by adjudication: user decision} 1/2-1/2[/pgn]
Very soon we will find out if LCZero can beat Stockfish Classical by at least 6 points, but I do NOT think that LCZero will get more than 53 to 47 in favor of LCZero.

PS: I believe that in order to make the 100 games fair they should have 30 games from the start position with no Opening, another 30 games with balanced positions and the rest of 40 games with White having a small advantage, but NOT a HUGE Advantage.
https://tcec-chess.com/live.html
I really wonder how the score would have ended if 50 games were played from the starting position with No Opening for SF NNUE Vs LCZero.?
I see that stockfish Classical is NOT as strong as LCZero ==> https://tcec-chess.com/live.html
Do NOT worry and be happy, we all live a short life :roll:
Top
User avatar
Nordlandia
Posts: 2821
Joined: Fri Sep 25, 2015 9:38 pm
Location: Sortland, Norway

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

Post by Nordlandia »

I have said that it is better to use a contempt minus value, since the classic is weaker. But no one listens to me.

For example minus 12 or more.

Will be able to squeeze out more draws than otherwise lost.
Top
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

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

Post by MikeB »

Nordlandia wrote: ↑Fri Nov 06, 2020 7:34 pm I have said that it is better to use a contempt minus value, since the classic is weaker. But no one listens to me.

For example minus 12 or more.

Will be able to squeeze out more draws than otherwise lost.
You are absolutely correct when SF is playing a stronger engine.
Image
Top
Milos
Posts: 4190
Joined: Wed Nov 25, 2009 1:47 am

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

Post by Milos »

Chessqueen wrote: ↑Sat Oct 24, 2020 7:16 am
Chessqueen wrote: ↑Thu Oct 22, 2020 5:46 pm
Nordlandia wrote: ↑Wed Oct 21, 2020 8:47 am SF Classicial in combination with probing partial 7-piece syzygy managed to hold this as black. Whereas SF NNUE lost. Komodo consulted 5-piece syzygy.

My theory is that SF Classical is better when it comes to late endgames when EGTB consulting matters. SF NNUE is able to probe EGTB like a magnitude slower because of the lesser speed.

[d]8/R7/2b2kn1/8/6P1/7P/5K2/8 w - - 1 21

[pgn][Event "?"]
[Site "i7-5960X 4.5GHz / 2x 2080 Super OC"]
[Date "2020.10.21"]
[Round "?"]
[White "komodo-14-64bit-bmi2"]
[Black "stockfish_20101813_x64_bmi2"]
[Result "1/2-1/2"]
[FEN "5nk1/pb6/3bp3/2pn3N/P7/8/5PPP/2R1R1K1 w - - 0 1"]
[GameDuration "00:51:28"]
[GameEndTime "2020-10-21T08:43:48.542 Vest-Europa (sommertid)"]
[GameStartTime "2020-10-21T07:52:20.251 Vest-Europa (sommertid)"]
[PlyCount "40"]
[SetUp "1"]
[Termination "adjudication"]
[TimeControl "1800+30"]

1. Ng3 {+0.56/39 56.68s} Nf6 {-0.83/37 50.19s} 2. Red1 {+0.83/41 69.11s}
Be7 {-0.81/43 66.96s} 3. Rb1 {+1.13/40 52.90s} Bc6 {-1.04/48 275.7s}
4. Rb8 {+1.14/37 45.01s} a5 {-0.94/44 27.84s} 5. Rc1 {+1.30/40 102.8s}
Kf7 {-0.96/36 32.20s} 6. Rc8 {+1.24/39 58.81s} Bxa4 {-0.90/41 66.50s}
7. R8xc5 {+1.37/39 65.19s} Bxc5 {-0.83/40 29.21s} 8. Rxc5 {+1.41/38 71.97s}
Ng6 {-0.92/45 217.4s} 9. Rxa5 {+1.71/39 134.3s} Bd7 {-1.17/42 33.75s}
10. Ne2 {+1.73/35 53.60s} e5 {-1.17/44 148.4s} 11. h3 {+1.73/37 52.17s}
Bc6 {-1.17/46 21.94s} 12. Nc3 {+1.78/37 53.06s} Nf4 {-1.23/48 34.56s}
13. f3 {+1.93/39 92.89s} N4d5 {-1.09/48 30.24s} 14. Ra3 {+1.85/36 54.62s}
Ne7 {-1.25/42 42.08s} 15. Kf2 {+2.25/37 52.30s} e4 {0.00/56 35.94s}
16. fxe4 {+2.43/36 39.28s} Nxe4+ {0.00/82 65.34s} 17. Nxe4 {+1.73/36 41.81s}
Bxe4 {0.00/150 38.89s} 18. Ra4 {+1.98/43 120.3s} Bc6 {0.00/46 262.0s}
19. Ra7 {+1.94/41 54.25s} Kf6 {0.00/53 110.5s} 20. g4 {+1.63/43 74.69s}
Ng6 {0.00/60 32.65s, Draw by adjudication: user decision} 1/2-1/2[/pgn]
Very soon we will find out if LCZero can beat Stockfish Classical by at least 6 points, but I do NOT think that LCZero will get more than 53 to 47 in favor of LCZero.

PS: I believe that in order to make the 100 games fair they should have 30 games from the start position with no Opening, another 30 games with balanced positions and the rest of 40 games with White having a small advantage, but NOT a HUGE Advantage.
https://tcec-chess.com/live.html
I really wonder how the score would have ended if 50 games were played from the starting position with No Opening for SF NNUE Vs LCZero.?
That is very simple to answer - it would be 50 draws, so 25:25.
Top
MMarco
Posts: 195
Joined: Sun Apr 12, 2020 1:09 am
Full name: Marc-O Moisan-Plante

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

Post by MMarco »

MMarco wrote: ↑Tue Oct 20, 2020 8:07 pm
Chessqueen wrote: ↑Tue Oct 20, 2020 5:24 am The next match will be Stockfisk Classical Vs LCZero. [...]
Last SuFi was a tie after 35 games: 4-4-27 (
https://tcec-chess.com/#div=sf&game=1&season=19 )

I predict Lc0 will clobber SF Classical. +15 at least! 🙂
Match underway, with a strong start by Lc0 ( https://tcec-chess.com/), +5 after 24 games:

Code: Select all

# PLAYER                            :  RATING  ERROR  PLAYED   (%)   CFS    W    D   L   D(%)
1 LCZero v0.26.3_T60.SV.JH.92-330   :    81.6   90.0    24   60.42    96    7   15   2  62.50
2 StockfishClassical 202007311012   :     0.0   ----    24   39.58   ---    2   15   7  62.50

Code: Select all

                                             W H I T E                                  B L A C K

            Name                   Gms     Win    Draw    Lose       %    ::  Gms     Win    Draw    Lose       %

LCZero v0.26.3_T60.SV.JH.92-330 :   12  :   6+  :   6=  :   0-  :  75.0%  ::   12  :   1+  :   9=  :   2-  :  45.8%
StockfishClassical 202007311012 :   12  :   2+  :   9=  :   1-  :  54.2%  ::   12  :   0+  :   6=  :   6-  :  25.0%
Top
MMarco
Posts: 195
Joined: Sun Apr 12, 2020 1:09 am
Full name: Marc-O Moisan-Plante

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

Post by MMarco »

MMarco wrote: ↑Sat Nov 07, 2020 6:47 pm
MMarco wrote: ↑Tue Oct 20, 2020 8:07 pm
Chessqueen wrote: ↑Tue Oct 20, 2020 5:24 am The next match will be Stockfisk Classical Vs LCZero. [...]
Last SuFi was a tie after 35 games: 4-4-27 (
https://tcec-chess.com/#div=sf&game=1&season=19 )

I predict Lc0 will clobber SF Classical. +15 at least! 🙂
Match underway, with a strong start by Lc0 ( https://tcec-chess.com/), +5 after 24 games:

Code: Select all

# PLAYER                            :  RATING  ERROR  PLAYED   (%)   CFS    W    D   L   D(%)
1 LCZero v0.26.3_T60.SV.JH.92-330   :    81.6   90.0    24   60.42    96    7   15   2  62.50
2 StockfishClassical 202007311012   :     0.0   ----    24   39.58   ---    2   15   7  62.50

Code: Select all

                                             W H I T E                                  B L A C K

            Name                   Gms     Win    Draw    Lose       %    ::  Gms     Win    Draw    Lose       %

LCZero v0.26.3_T60.SV.JH.92-330 :   12  :   6+  :   6=  :   0-  :  75.0%  ::   12  :   1+  :   9=  :   2-  :  45.8%
StockfishClassical 202007311012 :   12  :   2+  :   9=  :   1-  :  54.2%  ::   12  :   0+  :   6=  :   6-  :  25.0%
Update after 62 games:

Code: Select all

   # PLAYER                             :  RATING  ERROR  PLAYED    (%)   CFS    W    D    L   D(%)
   1 LCZero v0.26.3_T60.SV.JH.92-330    :    87.7   54.1      62  60.48   100   19   37    6  59.68
   2 StockfishClassical 202007311012    :     0.0   ----      62  39.52   ---    6   37   19  59.68

White advantage = 144.67 +/- 28.02
Draw rate (equal opponents) = 83.80 % +/- 9.82

Code: Select all

                                             W H I T E

           Name                 :  Gms     Win    Draw    Lose       %  

LCZero v0.26.3_T60.SV.JH.92-330 :   31  :  18+  :  13=  :   0-  :  79.0%
StockfishClassical 202007311012 :   31  :   6+  :  24=  :   1-  :  58.1%

                                             B L A C K

           Name                 :  Gms     Win    Draw    Lose       %
LCZero v0.26.3_T60.SV.JH.92-330 :   31  :   1+  :  24=  :   6-  :  41.9%
StockfishClassical 202007311012 :   31  :   0+  :  13=  :  18-  :  21.0%
Top
MMarco
Posts: 195
Joined: Sun Apr 12, 2020 1:09 am
Full name: Marc-O Moisan-Plante

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

Post by MMarco »

Final standings: +18 for Lc0, undefeated as white:

Code: Select all

   # PLAYER                             :  RATING  ERROR  PLAYED    (%)   CFS    W    D    L   D(%)
   1 LCZero v0.26.3_T60.SV.JH.92-330    :    74.0   33.8     100  59.00   100   28   62   10  62.00
   2 StockfishClassical 202007311012    :     0.0   ----     100  41.00   ---   10   62   28  62.00

White advantage = 137.76 +/- 19.42
Draw rate (equal opponents) = 86.85 % +/- 7.80

Code: Select all

                                             W H I T E                   

           Name                   Gms     Win    Draw    Lose       %   

LCZero v0.26.3_T60.SV.JH.92-330 :   50  :  27+  :  23=  :   0-  :  77.0%
StockfishClassical 202007311012 :   50  :  10+  :  39=  :   1-  :  59.0%

                                              B L A C K

           Name                   Gms     Win    Draw    Lose       % 

LCZero v0.26.3_T60.SV.JH.92-330 :   50  :   1+  :  39=  :  10-  :  41.0%
StockfishClassical 202007311012 :   50  :   0+  :  23=  :  27-  :  23.0%
Top

Return to “Computer Chess Club: General Topics”