ProteusChessSF - Developing my Stockfish 14.1 Polyglot MTCS derivative

[AlexChess]

Blitz 3 mins + 1 sec / all - Good performance of Proteus Chess SF v004 against Ethereal 13.50 github (no NNUE)

Android Snapdragon 870 8 threads 2 Mn/s versus 7 Mn/s



All games in PGN: http://www.microsmeta.com/public/Proteu ... _Blitz.zip

[AlexChess]

ProteusChessSF v004 qualified as sparring partner for SF derivatives vs Original engines gauntlet!!!!



Blitz 3 mins + 1 sec / all on Android Snapdragon 870 8 threads - 2 Mn/s

All PGN games: http://www.microsmeta.com/public/Proteu ... 220225.zip

[Guenther]

ProteusChessSF v004 qualified as sparring partner for SF derivatives vs Original engines gauntlet!!!!



Blitz 3 mins + 1 sec / all on Android Snapdragon 870 8 threads - 2 Mn/s

All PGN games: http://www.microsmeta.com/public/Proteu ... 220225.zip

You are a 'great' tester, bravo... +5.25 first move out of book.
(and you really did random selection of the openings w/o at least using the positions pairwise for White/Black,
so only SF had to play that known busted opening with Black)

[pgn][Event "Chess for Android Tournament"]
[Site "Italia"]
[Date "2022.03.01"]
[Round "5"]
[White "ProteusChessSF v004 260222"]
[Black "Stockfish 220225"]
[Result "1-0"]
[Device "RMX3370"]
[ECO "A40"]
[TimeControl "180+1"]

1.d4 {Book} 1...b5 {Book}
2.e4 {Book} 2...Bb7 {Book}
3.Bd3 {Book} 3...f5 {Book}
4.exf5 {Book} 4...Bxg2 {Book}
5.Qh5+ {+5.25/23 8.575}
1-0[/pgn]

original output before being transformed to more standard like pgn:

Code: Select all

[Event "Chess for Android Tournament"]
[Site "Italia"]
[Date "2022.03.01"]
[Round "5"]
[White "ProteusChessSF v004 260222"]
[Black "Stockfish 220225"]
[Result "1-0"]
[Device "RMX3370"]
[ECO "A40"]
[TimeControl "180+1"]

1.d4 { Risposta di apertura 0.0 } 1...b5 { Risposta di apertura 0.0 } 2.e4
{ Risposta di apertura 0.0 } 2...Bb7 { Risposta di apertura 0.0 } 3.Bd3
{ Risposta di apertura 0.0 } 3...f5 { Risposta di apertura 0.001 } 4.exf5
{ Risposta di apertura 0.0 } 4...Bxg2 { Risposta di apertura 0.0 } 5.Qh5+
{ [23/41]525/g7g6 8.575 }...
1-0

Of course SF agrees with 'itself'

Code: Select all

SF 22022508-64 SSSE3 on RWBC-CAPPUCCINO Win7U64 Q8200 2.33Ghz

>> setoption name Threads value 4
>> setoption name Hash value 512
>> position fen rn1qkbnr/p1ppp1pp/8/1p3P2/3P4/3B4/PPP2PbP/RNBQK1NR w KQkq - 0 5
>> go infinite
<< info string NNUE evaluation using nn-6877cd24400e.nnue enabled
...
<< info depth 18 seldepth 31 multipv 1 score cp 422 nodes 691906 nps 469088 hashfull 8 tbhits 0 time 1475 pv d1h5 g7g6 f5g6 f8g7 g6h7 e8f8 g1f3 g8f6 h5g6 g2f3 h1g1 h8h7 g6g3 f3e4 d3e4 f6e4 g3f3 f8g8 f3e4 b8c6 c1e3 d8e8 b1d2 e7e5 d4e5 e8e5
<< info depth 19 seldepth 28 multipv 1 score cp 442 nodes 735195 nps 467087 hashfull 8 tbhits 0 time 1574 pv d1h5 g7g6 f5g6 f8g7 g6h7 e8f8 g1f3 g8f6 h5g6 g2f3 h1g1 h8h7 g6g3 f3e4 d3e4 f6e4 g3f3 f8g8 f3e4 d7d5 e4e6 g8h8 b1c3 d8d7 e6d5 d7d5 c3d5 b8c6
<< info depth 20 seldepth 36 multipv 1 score cp 439 nodes 887609 nps 461095 hashfull 9 tbhits 0 time 1925 pv d1h5 g7g6 f5g6 f8g7 g6h7 e8f8 g1f3 g8f6 h5g6 g2f3 h1g1 h8h7 g6g3 f3e4 d3e4 f6e4 g3f3 f8g8 f3e4 b8c6 d4d5 c6a5 b1a3 a5b7 a3b5 a7a6 b5c3
<< info depth 21 seldepth 33 multipv 1 score cp 432 nodes 1015316 nps 462558 hashfull 9 tbhits 0 time 2195 pv d1h5 g7g6 f5g6 f8g7 g6h7 e8f8 g1f3 g8f6 h5g6 g2f3 h1g1 h8h7 g6g3 f3e4 d3e4 f6e4 g3f3 f8g8 f3e4 b8c6 d4d5 c6a5 b1c3 g8h8 c3b5 d8f8 c1e3 a8b8 b5c3 b8b2 c3e2
<< info depth 22 seldepth 40 multipv 1 score cp 406 upperbound nodes 1749437 nps 457608 hashfull 16 tbhits 0 time 3823 pv d1h5 g7g6
<< info depth 22 seldepth 40 multipv 1 score cp 444 nodes 3194843 nps 441154 hashfull 36 tbhits 0 time 7242 pv d1h5 g7g6 f5g6 f8g7 g6h7 e8f8 g1h3 g2h1 h3g5 g8h6 b1c3 b8c6 c1e3 b5b4 c3e2 e7e5 e1c1 h1d5 d4e5 c6e5 e3c5 d7d6 e2f4 d8d7 f4d5 d6c5 d5c7 d7g4 c7e6 f8e7 e6g7 g4h5 g7h5 e5d3 d1d3 a8f8 d3e3 e7d7
<< info depth 23 seldepth 40 multipv 1 score cp 464 lowerbound nodes 6243935 nps 446314 hashfull 77 tbhits 0 time 13990 pv d1h5
<< info depth 23 seldepth 40 multipv 1 score cp 492 lowerbound nodes 6982773 nps 445870 hashfull 85 tbhits 0 time 15661 pv d1h5
<< info depth 23 seldepth 40 multipv 1 score cp 487 nodes 7170832 nps 445172 hashfull 87 tbhits 0 time 16108 pv d1h5 g7g6 f5g6 f8g7 g6h7 e8f8 g1h3 g2h1 h3g5 g8h6 b1c3 b8c6 c1e3 b5b4 e1c1 b4c3 d1h1 e7e5 d4e5 d8e7 h5f3 f8e8 e5e6 e7b4 b2b3 h8f8 f3d5 b4d6 d3g6 e8d8 e6e7 d6e7 e3c5 d7d6 h1g1
<< info depth 24 seldepth 44 multipv 1 score cp 518 lowerbound nodes 10670321 nps 448521 hashfull 129 tbhits 0 time 23790 pv d1h5
<< info depth 24 seldepth 53 multipv 1 score cp 501 nodes 12147046 nps 447092 hashfull 149 tbhits 0 time 27169 pv d1h5 g7g6 f5g6 f8g7 g6h7 e8f8 g1h3 g2h1 h3g5 g8h6 b1c3 b8c6 c1e3 e7e6 e1c1 h1d5 d3g6 c6e7 h5h6 d8e8 c3d5 e8g6 h6g6 e7g6 d5c7 a8b8 d1g1 f8e7 g5e6 d7e6 g1g6 h8h7 g6e6 e7f8
<< info depth 25 seldepth 43 multipv 1 score cp 538 lowerbound nodes 16549825 nps 445270 hashfull 201 tbhits 0 time 37168 pv d1h5
<< info depth 25 seldepth 53 multipv 1 score cp 496 nodes 16954706 nps 444003 hashfull 205 tbhits 0 time 38186 pv d1h5 g7g6 f5g6 f8g7 g6h7 e8f8 g1h3 g2h1 h3g5 g8h6 b1c3 b8c6 c1e3 e7e6 e1c1 h1d5 d1g1 d8f6 c3d5 e6d5 g5f3 f6f7 d3g6 f7e6 c1b1 h6f7 g6f5 e6f6 g1g6 f6d8 f3h4 c6e7 g6g4 g7f6 e3h6 f7h6
<< info depth 26 seldepth 48 multipv 1 score cp 538 nodes 20450771 nps 441482 hashfull 246 tbhits 0 time 46323 pv d1h5 g7g6 f5g6 f8g7 g6h7 e8f8 g1h3 g2h1 h3g5 g8h6 b1c3 e7e5 c1d2 b8c6 e1c1 b5b4 c3e4 d8e8 h5h3 h1e4 d3e4 c6d4 e4a8 e8a8 h3d7 a7a5 d2b4 a5b4 d1d4 a8e8 g5e6 e8e6 d7e6 e5d4 e6c4 h6f5 c4b4 f8f7 b4b3 f7g6 a2a4 h8h7 h2h3 h7h5
>> stop
<< info depth 27 seldepth 48 multipv 1 score cp 538 nodes 25607714 nps 442320 hashfull 308 tbhits 0 time 57894 pv d1h5 g7g6 f5g6 f8g7 g6h7 e8f8 g1h3 g2h1 h3g5 g8h6 b1c3 e7e5 c1d2 b8c6 e1c1 b5b4 c3e4 d8e8 h5h3 h1e4 d3e4 c6d4 e4a8 e8a8 h3d7 a7a5 d2b4 a5b4 d1d4 a8e8 g5e6 e8e6 d7e6 e5d4 e6c4 h6f5 c4b4 f8f7 b4b3 f7g6 a2a4 h8h7 h2h3 h7h5
<< bestmove d1h5 ponder g7g6

[Sopel]

lmao the game was adjudicated one ply after book

this also means that in this testing setup an engine that fakes uci output to always be above 600 will lose every game instantly

[Guenther]

lmao the game was adjudicated one ply after book

this also means that in this testing setup an engine that fakes uci output to always be above 600 will lose every game instantly

It lasted longer, I just removed the rest of the pgn after move 1 as it was useless for my post, here is the full not converted original pgn:
(BTW I find GUIs 'funny' that write 'XYZ resigned' when uci engines don't resign, but the GUI adjudicates on given evals)

Faking output will only help, if the GUI/user setting does not have checking evals for both sides (or just the losing side).

[pgn][Event "Chess for Android Tournament"]
[Site "Italia"]
[Date "2022.03.01"]
[Round "5"]
[White "ProteusChessSF v004 260222"]
[Black "Stockfish 220225"]
[Result "1-0"]
[Device "RMX3370"]
[ECO "A40"]
[TimeControl "180+1"]

1.d4 { Risposta di apertura 0.0 } 1...b5 { Risposta di apertura 0.0 } 2.e4
{ Risposta di apertura 0.0 } 2...Bb7 { Risposta di apertura 0.0 } 3.Bd3
{ Risposta di apertura 0.0 } 3...f5 { Risposta di apertura 0.001 } 4.exf5
{ Risposta di apertura 0.0 } 4...Bxg2 { Risposta di apertura 0.0 } 5.Qh5+
{ [23/41]525/g7g6 8.575 } 5...g6 { [16/30]-223/f5g6 0.708 } 6.fxg6
{ [23/51]526/f8g7 4.142 } 6...Bg7 { [22/52]-549/g6h7 17.039 } 7.gxh7+
{ [22/51]553/e8f8 7.552 } 7...Kf8 { [20/37]-539/g1h3 0.661 } 8.Nh3
{ [22/48]572/g2h1 4.531 } 8...Bxh1 { [23/43]-544/h3g5 8.101 } 9.Ng5
{ [22/52]563/g8h6 4.486 } 9...Nh6 { [24/52]-529/b1c3 15.216 } 10.Nc3
{ [25/42]576/b8c6 11.889 } 10...e5 { [25/42]-538/c1d2 33.592 } 11.Bd2
{ [21/52]593/a7a5 6.356 } 11...Nc6 { [22/52]-588/e1c1 12.109 } 12.O-O-O
{ [25/57]590/b5b4 4.328 } 12...b4 { [23/49]-623/c3e4 7.163 } 13.Nce4
{ [25/59]588/d8e8 5.215 } 13...Qe8 { [24/51]-600/h5h3 15.59 } 14.Qh3
{ [26/61]604/h1e4 6.529 } 14...Bxe4 { [24/40]-638/d3e4 6.487 } 15.Bxe4
{ [26/55]618/h6f7 12.025 } 15...Nf7 { [26/35]-637/e4c6 16.228 } 16.Bxc6
{ [21/47]650/f7g5 3.849 } 16...Nxg5 { [23/49]-642/d2b4 2.628 } 17.Bxb4+
{ [22/40]677/d7d6 7.125 } 17...d6 { [21/49]-663/h3g2 3.066 } 18.Qg2
{ [21/41]676/e8h5 5.935 } 18...Qh5 { [21/44]-655/c6a8 3.807 } 19.Bxa8
{ [23/43]688/g5e6 4.894 } 19...Rxh7 { [26/42]-689/d1e1 17.734 } 20.Re1
{ [22/42]688/g7h6 4.23 } 20...Bh6 { [21/43]-681/c1b1 4.074 } 21.Kb1
{ [24/48]726/h7e7 17.479 } 21...Re7 { [23/39]-662/d4e5 4.754 } 22.dxe5
{ [20/27]757/e7e5 3.82 } 22...Rxe5 { [19/39]-677/f2f3 3.181 } 23.Rg1
{ [19/38]746/g5h3 4.984 } 23...Nh3 { [19/32]-618/g1f1 3.126 } 24.Qg8+
{ [20/38]812/f8e7 3.794 } 24...Ke7 { [14/30]-756/g1h1 0.605 } 25.Rf1
{ [19/46]860/h5e2 4.011 } 25...Qe2 { [23/42]-861/a8g2 16.5 } 26.Bg2
{ [21/39]845/h3f4 4.292 } 26...Ng5 { [22/46]-854/a2a3 10.336 } 27.a3
{ [22/52]857/c7c5 4.385 } 27...Ne6 { [20/38]-862/f1e1 1.276 } 28.Re1
{ [23/49]849/e2h5 6.134 } 28...Qb5 { [19/41]-845/e1e5 3.157 } 29.Rxe5
{ [24/41]862/b5e5 3.376 } 29...Qxe5 { [18/39]-869/g2c6 2.494 } 30.Bc6
{ [26/49]864/e6f8 3.991 } 30...Nf8 { [17/35]-877/g8g4 1.47 } 31.Qg4
{ [23/37]861/e5e6 3.381 } 31...Qe6 { [18/35]-873/g4d4 1.346 } 32.Qd4
{ [23/37]869/f8d7 5.996 } 32...Nd7 { [21/37]-861/b4c3 1.845 } 33.Qh4+
{ [22/39]882/e6f6 3.188 } 33...Qf6 { [15/35]-858/h4e4 1.133 } 34.Qg4
{ [25/37]883/f6e6 5.865 } 34...Qe6 { [17/40]-841/g4h4 1.994 } 35.Qh4+
{ [19/41]887/e6f6 3.361 } 35...Qf6 { [17/36]-835/h4e4 0.932 } 36.Qe4+
{ [22/46]876/d7e5 4.16 } 36...Ne5 { [18/38]-864/c6a4 4.501 } 37.Ba4
{ [23/52]882/c7c5 6.737 } 37...c5 { [16/29]-862/b4c3 0.663 } 38.Bc3
{ [20/34]882/f6f4 4.166 } 38...Qxf2 { [16/31]-869/c3e5 1.561 } 39.Qh7+
{ [21/50]930/e5f7 7.746 } 39...Nf7 { [15/36]-877/h7e4 0.703 } 40.Qe4+
{ [19/43]936/f7e5 2.743 } 40...Ne5 { [14/24]-903/c3e5 0.623 } 41.Bxe5
{ [20/41]930/d6e5 4.319 } 41...dxe5 { [14/30]-898/e4h7 0.978 } 42.Qh7+
{ [17/31]958/f2f7 2.297 } 42...Qf7 { [14/28]-912/h7h6 0.544 } 43.Qxh6
{ [19/32]975/f7f5 2.822 } 43...c4 { [18/26]-937/h6h4 3.847 } 44.Qh4+
{ [19/38]1012/e7d6 3.937 } 44...Kd6 { [16/31]-963/h4d8 1.779 } 45.Qd8+
{ [21/31]1046/d6e6 2.864 } 45...Ke6 { [15/25]-994/a4b5 0.552 } 46.Bb5
{ [21/30]1056/f7f1 2.249 } 46...Kf5 { black resigned:adjudication } 1-0[/pgn]

[AlexChess]

lmao the game was adjudicated one ply after book

this also means that in this testing setup an engine that fakes uci output to always be above 600 will lose every game instantly

Thank you Sopel. I check this game right now. Mea culpa, I haven't controlled the games overnight. Obviously I would not have published the games but only the crosstable knowing that is a wrong adjudication.

Unfortunately here I cannot edit/remove my post, but I replay the game and update results.

PS: I have used the internal Chess for Android 6.2.1 Abik book (4 full plies) I disable it and replay the full match. For Polyglot supporting engines I'm using Perfect 2021 bin & abk up to 8 full moves. Is it reliable? I'd prefer to completely disable openings book, but engines play very often the same boring Spanish with 3 ...Nf6

Best regards, Alex

[AlexChess]

@Sopel

Luckily, updating both pgn and crosstable with corrected results on my server, it has replaced the wrong data!

After 10 blitz games, the results WITHOUT openings book is now 5-5 (all draws)

So I will play Two Gauntlets against Original engines availables for Android against Group B and Group C:

Sparring partners: Stockfish dev and ProteusChessSF

All draws between Top 13 engines are becoming a little boring (You can follow all my tournaments at the link on bottom)

Thank you for spending your time watching my games!

PS: As I have already told you privately, I admire your programming and compiling skills and funny humor (Fat Titz and his specs are wonderful!) and, if even a myth like you is criticized by trolls and haters (dangi12012), I can easily accept my own critics

Kind regards, Alex

[AlexChess]

Proteus Chess SF v004 Eigenmann 111 Rapid Engine Tests



Hardware:
Windows 11 ARM64 4 CPUs under Parallels 17 on Mac Mini M1
(M1 calculates only 2 Mn/s at starting position, while today best mass-market hardware is 30 times faster https://openbenchmarking.org/test/pts/stockfish-1.3.0)

Positions and solutions:
https://glarean-magazin.ch/2017/03/05/c ... test-eret/

[AlexChess]

Differences between SF-PB and ProteusChessSF v004 analysis on starting position



My purpose is NOT to make a SF clone to feed my ego, but to modify it to create an extremely aggressive engine. I have released the source code like GPLv3 imposes, so everyone can verify my changes. It's not a simple rename.

I'd like to discuss ideas to implement patterns for material sacrifices (like Carlsen and top IM & GMs do) and castling attacks with all members interested, like I'm already doing here in PM with very skilled friends

I'll answer to all interesting and constructive comments

[AlexChess]

Blitz 3 mins + 1 sec ProteusChessSFv004 vs BrainLearn 16 & Crystal 220304 (No books)

Android Snapdragon 870 (2 Mn/s on starting position)



ALL PGN GAMES: http://microsmeta.com/public/ProteusChe ... -1.pgn.zip



ALL PGN GAMES: http://microsmeta.com/public/ProteusChe ... -1.pgn.zip

[pgn][Event "Chess for Android Tournament"]
[Site "Italia"]
[Date "2022.03.06"]
[Round "28"]
[White "ProteusChessSF v004 260222"]
[Black "Crystal 220304"]
[Result "1-0"]
[Device "RMX3370"]
[ECO "B90"]
[TimeControl "180+1"]

1.e4 { [26/35]45/c7c6 4.145 } 1...c5 { [22/31]-22/g1f3 16.746 } 2.Nf3
{ [22/32]54/d7d6 2.487 } 2...d6 { [23/29]-23/d2d4 10.921 } 3.d4
{ [22/30]51/g8f6 3.2 } 3...cxd4 { [24/29]-27/f3d4 5.996 } 4.Nxd4
{ [23/30]58/g8f6 2.314 } 4...Nf6 { [22/30]-32/b1c3 1.71 } 5.Nc3
{ [26/30]61/a7a6 2.372 } 5...a6 { [20/30]-23/f1e2 3.628 } 6.Be3
{ [28/33]60/e7e5 6.33 } 6...e5 { [21/31]-22/d4b3 4.948 } 7.Nb3
{ [26/32]57/c8e6 3.597 } 7...Be7 { [21/32]-19/f1e2 3.801 } 8.f3
{ [24/33]80/c8e6 4.618 } 8...Be6 { [22/37]-39/d1d2 13.68 } 9.Qd2
{ [23/30]76/e8g8 3.803 } 9...h5 { [21/32]-40/c3d5 2.928 } 10.Be2
{ [25/33]62/d8c7 4.156 } 10...O-O { [25/32]-42/c3d5 17.027 } 11.Nd5
{ [25/32]68/e6d5 12.186 } 11...Bxd5 { [20/32]-40/e4d5 2.854 } 12.exd5
{ [23/31]66/d8c7 2.09 } 12...Qc7 { [22/34]-35/a2a4 13.295 } 13.O-O-O
{ [27/39]62/b8d7 8.809 } 13...Nbd7 { [20/36]-48/c2c4 6.291 } 14.c4
{ [24/37]83/b7b5 3.87 } 14...b5 { [20/36]-53/c1b1 3.072 } 15.Kb1
{ [23/30]85/d7b6 4.0 } 15...Nb6 { [21/35]-37/d2a5 4.397 } 16.Bxb6
{ [24/29]89/c7b6 3.435 } 16...Qxb6 { [22/39]-37/b3a5 3.227 } 17.Na5
{ [22/31]100/f8e8 4.114 } 17...Rfb8 { [21/37]-36/a5c6 4.197 } 18.Nc6
{ [23/27]105/b8b7 4.463 } 18...Rb7 { [20/34]-27/h1f1 3.187 } 19.Rc1
{ [23/27]99/e7f8 3.027 } 19...Bf8 { [22/30]-44/h1d1 15.138 } 20.Rhd1
{ [22/30]85/g7g6 3.743 } 20...Nd7 { [19/36]-47/g2g4 3.107 } 21.g4
{ [19/29]110/b6f2 7.786 } 21...Qf2 { [20/32]-42/g4h5 13.356 } 22.gxh5
{ [21/31]105/f2h4 5.602 } 22...Qh4 { [21/30]-42/f3f4 19.457 } 23.Rc2
{ [22/29]100/b5c4 9.008 } 23...bxc4 { [21/32]-57/c2c4 10.769 } 24.Rxc4
{ [22/33]108/h4h2 5.084 } 24...Qxh2 { [18/35]-44/d2e1 7.58 } 25.Rcc1
{ [23/35]106/a6a5 6.121 } 25...a5 { [19/35]-48/d1h1 7.494 } 26.a3
{ [28/30]75/a5a4 21.573 } 26...a4 { [18/27]-56/d1h1 6.24 } 27.Rh1
{ [23/37]127/h2g2 2.498 } 27...Qg2 { [17/29]-41/c1g1 1.146 } 28.Rcg1
{ [25/35]95/g2f2 9.017 } 28...Qf2 { [18/25]-43/g1f1 2.311 } 29.Rf1
{ [23/30]87/f2g2 2.831 } 29...Qg2 { [17/32]-46/h1g1 2.381 } 30.Rfg1
{ [24/32]92/g2f2 10.811 } 30...Qf2 { [18/35]-40/g1f1 1.433 } 31.Rf1
{ [22/33]83/f2g2 4.101 } 31...Qg2 { [18/28]-60/b1a2 4.345 } 32.Qc2
{ [24/35]88/g8h8 9.675 } 32...Kh8 { [12/25]-40/h1g1 0.467 } 33.Rh4
{ [22/29]106/d7f6 4.351 } 33...Nf6 { [12/21]-32/f1d1 0.513 } 34.Rd1
{ [18/29]67/g2f2 17.271 } 34...Qf2 { [16/25]-38/h4c4 1.589 } 35.Rc4
{ [22/32]83/h8g8 3.39 } 35...Kg8 { [15/29]-30/b1a1 1.204 } 36.Rd2
{ [23/33]110/f6h5 3.134 } 36...Qe3 { [14/25]-12/e2d1 0.593 } 37.Rd3
{ [22/40]114/e3f2 2.334 } 37...Qg1+ { [15/32]-11/c2d1 0.715 } 38.Ka2
{ [23/36]117/f6h5 2.743 } 38...Nxh5 { [14/28]-90/c4a4 2.208 } 39.Rxa4
{ [22/42]145/a8a4 1.599 } 39...Rxa4 { [16/29]-119/c2a4 1.047 } 40.Qxa4
{ [23/32]168/h5f4 1.675 } 40...Nf4 { [17/27]-117/d3d2 1.795 } 41.Rd2
{ [20/37]173/g1e3 1.998 } 41...Qe3 { [16/28]-77/a4d1 1.336 } 42.Qd1
{ [21/41]211/g7g5 8.086 } 42...g6 { [17/35]-96/e2a6 2.695 } 43.Ba6
{ [21/28]246/b7d7 2.675 } 43...Rc7 { [17/34]-111/d2c2 1.12 } 44.Rc2
{ [19/29]265/g8g7 1.079 } 44...e4 { [17/30]-107/c2c3 1.638 } 45.Rc3
{ [17/29]318/e3b6 0.394 } 45...Qb6 { [14/29]-138/a6c4 0.162 } 46.Bc4
{ [17/27]346/e4e3 0.54 } 46...Rb7 { [18/25]-133/c3b3 2.192 } 47.Rb3
{ [19/32]387/b6c5 0.867 } 47...Qc5 { [15/32]-127/f3e4 0.381 } 48.Rxb7
{ [20/41]355/c5c4 1.302 } 48...Qxc4+ { [12/29]-52/b7b3 0.445 } 49.Rb3
{ [20/32]371/e4f3 0.727 } 49...exf3 { [14/32]-234/d1f3 1.397 } 50.Qxf3
{ [19/42]424/g6g5 0.805 } 50...f5 { [18/34]-284/c6a5 2.664 } 51.Na5
{ [18/28]418/c4d4 0.87 } 51...Qd4 { [15/26]-257/f3g3 0.453 } 52.Nc6
{ [23/33]459/d4c4 3.599 } 52...Qc4 { [17/40]-325/c6a5 1.984 } 53.Na5
{ [21/42]414/c4d4 1.491 } 53...Qd4 { [14/36]-292/f3g3 0.492 } 54.Rb4
{ black resigned ([14/24]-275/f3b3) } 1-0[/pgn]