ProteusChessSF - Developing my Stockfish 14.1 Polyglot MTCS derivative

AlexChess · Post by **AlexChess** » Thu Feb 17, 2022 10:30 am

Good results of Proteus Chess SF v003 AVX2 on blitz games 3 mins + 1 sec /all with powerful hardware: Ryzen 9 3900x!!!

With my last changes, v003 is rock solid against latest Stockfish-dev (no loses)

There is still a huge strength difference between SF & derivatives and best original engines (Ethereal 13.50 github has NNUE disabled by design)

Zahak 9.0-Android (unofficial derivative) seems a little stronger on my tests (Android games at signature link), probably due to a better optimization compiling it.

AlexChess · Post by **AlexChess** » Thu Feb 17, 2022 11:14 am

ProteusChess v004 will have more modifications over the original (trying to make it more aggressive, but I'm already happy with it

)

AlexChess · Post by **AlexChess** » Thu Feb 17, 2022 11:15 am

AlexChess · Post by **AlexChess** » Thu Feb 17, 2022 11:16 am

Differences between ProteusChessSF v003 and latest Stockfish-dev with new Sopel NNUE

AlexChess · Post by **AlexChess** » Sat Feb 19, 2022 9:15 am

Hi!
Image links on all my previous posts are strangely broken and since my edit time is over, I cannot replace them.
So for new posts I'll use only my server not editable, like on 4 previous posts.
You can find broken images and posts of v001 & v002 builds here: https://banksiagui.com/forums/viewtopic.php?p=124#p124

Note: Differently from what somebody thinks, here I don't only publish test games, but also developing ideas a technical issues.
If you don't like what I doing, please do not read and do not spam my topic, simply ignore it!

I'll answer to all meaningful comments and suggestions

I'm interested in deep search because I'd like to add patterns to pawn and pieces sacrifices to gain a superior or winning position like Carsen and GMs do and schemas to attack castles. Sometimes I perfectly see a draw with my 1900 ELO while even top engines continue to play until 6 man Syzygy tablebases determine that the game is over.

AlexChess · Post by **AlexChess** » Sat Feb 19, 2022 7:31 pm

Very interesting game between Proteus Chess SF v003 and dorsz's Blue Marlin 14.8

Black Marlin attacks Proteus castle that counterattacks on the other side. Nobody can overbear

[Event "Long Time Best Engines-BanksiaGui"] 
[Site "Rome"] 
[Date "2022.02.19"] 
[Round "1"] 
[White "ProteusChessSF v003 130222"] 
[Black "Blue Marlin 14.8"] 
[Result "1/2-1/2"] 
[Board "174"] 
[ECO "C50"] 
[Opening "Giuoco Pianissimo"] 
[Termination "repetition"] 
[Time "18:35:16"] 
[TimeControl "7200+30"] 
 
1.e4 e5 { +0.21/45 624919 1005529840 } 2.Nf3 Nc6 { +0.06/38 67129 111522526 } 
3.Bc4 Nf6 { +0.07/37 86896 145700700 } 4.d3 Bc5 { +0.01/42 96996 160110101 } 
5.Nc3 { C50: Giuoco Pianissimo, Italian four knights variation } 5...h6 
{ +0.00/44 157766 260438351 } 6.h3 a6 { +0.00/42 117286 192481289 } 7.a4 d6 
{ +0.00/44 112471 184573256 } 8.O-O Be6 { -0.26/40 288244 474330615 } 9.Bxe6 
{ +0.06/36 100375 183887822 } 9...fxe6 { -0.48/37 73305 122038098 } 10.Ne2 
{ +0.17/41 124665 230696066 } 10...g5 { -0.32/38 112075 185990809 } 11.c3 
{ +0.00/41 184735 348017975 } 11...g4 { -0.47/39 99385 170743943 } 12.hxg4 
{ +0.00/49 141957 272721146 } 12...Nxg4 { -0.36/39 130150 218686468 } 13.b4 
{ +0.00/46 161375 310768832 } 13...Bb6 { +0.00/46 191808 326161811 } 14.d4 
{ +0.00/49 248392 489081008 } 14...Qf6 { +0.00/47 199192 314956723 } 15.a5 
{ +0.00/47 142191 236374799 } 15...Ba7 { +0.00/48 152810 260967254 } 16.Qd3 
{ +0.00/46 168174 330200438 } 16...Kd7 { +0.00/48 191794 329006697 } 17.b5 
{ +0.00/50 145250 289407105 } 17...axb5 { +0.00/49 165540 301728577 } 18.Qxb5 
{ +0.00/47 138376 285680832 } 18...Rhg8 { +0.00/50 149195 268870022 } 19.a6 
{ +0.00/46 147837 308671859 } 19...Bb6 { +0.00/51 175201 278878706 } 20.d5 
{ +0.00/49 136450 285616749 } 20...Ke7 { +0.00/52 125016 221635654 } 21.dxc6 
{ +0.00/50 156491 323831965 } 21...bxa6 { +0.00/59 236906 394333139 } 22.Rxa6 
{ +0.00/57 198096 343134363 } 22...Rxa6 { +0.00/58 375438 509048056 } 23.Qxa6 
{ +0.00/57 279364 213020215 } 23...Qg6 { +0.00/56 126985 93077571 } 24.Nh4 
{ +0.00/53 150694 80062489 } 24...Qf6 { +0.00/58 296199 162557881 } 25.Nf3 
{ +0.00/60 176255 160598739 } 25...Qg6 { +0.00/62 108761 143306614 } 26.Nh4 
{ +0.00/62 159270 195218970 } 26...Qh5 { +0.00/61 134990 142437064 } 27.Nf3 
{ +0.00/67 329805 425116722 } 27...Nf6 { +0.00/161 592578 707276219 } 28.Ne1 
{ +0.00/61 145424 113351037 } 28...Ng4 { +0.00/52 141262 180582382 } 29.Nf3 
{ +0.00/65 192082 283729200 } 29...Nf6 { +0.00/60 1341343 1882898403 } 30.Ne1 
{ +0.00/65 1284645 1506436871 } 30...Qh3 { +0.00/54 65517 94503562 } 31.Qa2 
{ +0.00/56 465821 629084609 } 31...h5 { +0.00/58 65173 49647018 } 32.Qd2 
{ +0.00/54 182145 132090222 } 32...Kf7 { +0.00/55 67491 55838179 } 33.Qa2 
{ +0.00/48 108845 123553389 } 33...h4 { +0.00/56 109378 142123559 } 34.Be3 
{ +0.00/51 106598 163543381 } 34...Bxe3 { +0.00/50 73086 89166861 } 35.fxe3 
{ +0.00/54 185963 285785636 } 35...Qxe3+ { +0.00/52 143328 168631627 } 36.Kh2 
{ +0.00/56 153207 166161726 } 36...Ke7 { +0.00/56 137792 169675864 } 37.Rf3 
{ +0.00/56 117929 154651361 } 37...Qxe4 { +0.00/56 101755 130662979 } 38.Qa6 
{ +0.00/56 103365 127200681 } 38...h3 { +0.00/57 85092 106046238 } 39.gxh3 
{ +0.00/58 781935 221136679 } 39...Ra8 { +0.00/55 71289 54055211 } 40.Qxa8 
{ +0.00/56 76026 89893882 } 40...Qxe2+ { +0.00/54 198954 178878803 } 41.Kg3 
{ +0.00/56 147056 108976906 } 41...Qxe1+ { +0.00/54 57231 72881514 } 42.Kg2 
{ +0.00/55 115991 87442699 } 42...Qe2+ { +0.00/53 93442 72740192 } 43.Rf2 
{ +0.00/67 76161 38403794 } 43...Qe4+ { +0.00/55 98966 23437662 } 44.Kh2 
{ +0.00/58 85599 19961302 } 44...Qe3 { +0.00/52 101595 10640781 } 45.Rg2 
{ +0.00/50 77699 6260879 } 45...Qf4+ { +0.00/50 85255 25218664 } 46.Kh1 
{ +0.00/55 116048 105172886 } 46...Qc1+ { +0.00/56 140213 218689682 } 47.Kh2 
{ +0.00/63 62046 101261276 } 47...Qf4+ { +0.00/59 173849 125767279 } 48.Kg1 
{ +0.00/60 150490 104898944 } 48...Qe3+ { +0.00/56 25708 14794362 } 49.Kh2 
{ +0.00/63 65020 68666710 } 49...Qf4+ { +0.00/66 8955 7284069 } 1/2-1/2

AlexChess · Post by **AlexChess** » Sun Feb 20, 2022 11:25 am

For who continues to say that derivatives will never offer more than Stockfish, here there is an incredible game of Stockfish 14.1 vs Blue Marlin 14.6b21 of my friend dorsz. (I'm fine tuning Proteus Chess SF against his top engines)

[Event "sweras-44"] 
[Site "?"] 
[Date "2021.12.06"] 
[Round "19.2"] 
[White "Stockfish 14.1-modern"] 
[Black "Blue Marlin 14.6b21"] 
[Result "0-1"] 
[EventDate "2021.12.05"] 
[FEN "rnbqkb1r/p2p1ppp/4Pn2/1pp5/2P5/5N2/PP2PPPP/RNBQKB1R b KQkq - 0 1"] 
[SetUp "1"] 
 
1...fxe6 { [%eval 90,26] [%emt 0:00:03] } 2.cxb5 { [%eval 109,25] [%emt 0:00:02] } 2...a6 { [%eval 110,25] [%emt 0:00:02] } 3.e3 { [%eval 133,25] [%emt 0:00:01] } 3...d5 { [%eval 105,23] [%emt 0:00:01] (Be7) } 4.Nc3 { [%eval 134,22] [%emt 0:00:02] } 4...Nbd7 { [%eval 100,24] [%emt 0:00:04] } 5.bxa6 { [%eval 133,21] [%emt 0:00:02] } 5...Be7 { [%eval 125,24] [%emt 0:00:02] } 6.a4 { [%eval 150,24] [%emt 0:00:04] (Be2) } 6...O-O { [%eval 108,23] [%emt 0:00:02] } 7.Bb5 { [%eval 139,22] [%emt 0:00:01] } 7...Bxa6 { [%eval 105,24] [%emt 0:00:03] } 8.O-O { [%eval 159,24] [%emt 0:00:01] } 8...Bd6 { [%eval 96,22] [%emt 0:00:01] (Bb7) } 9.b3 { [%eval 163,28] [%emt 0:00:15] } 9...Bb7 { [%eval 85,25] [%emt 0:00:03] (Ne5) } 10.Bb2 { [%eval 160,20] [%emt 0:00:01] } 10...Qe7 { [%eval 85,22] [%emt 0:00:02] (Ne5) } 11.a5 { [%eval 145,25] [%emt 0:00:04] (Ne2) } 11...d4 { [%eval 55,20] [%emt 0:00:01] (Nb8) } 12.exd4 { [%eval 292,19] [%emt 0:00:01] } 12...cxd4 { [%eval 53,22] [%emt 0:00:01] } 13.Nxd4 { [%eval 102,24] [%emt 0: 00:04] (Qxd4) } 13...Bxh2+ { [%eval 0,24] [%emt 0:00:01] } 14.Kxh2 { [%eval 102,24] [%emt 0:00:01] } 14...Ne5 { [%eval 0,27] [%emt 0:00:02] } 15.f4 { [%eval 141,24] [%emt 0:00:01] (Be2) } 15...Nfg4+ { [%eval -217,19] [%emt 0:00:01] } 16.Kg3 { [%eval 147,26] [%emt 0:00:02] } 16...Qf6 { [%eval -150,23] [%emt 0:00:02] } 17.Qe2 { [%eval 133,27] [%emt 0:00:02] (Bd7) } 17...Qh6 { [%eval -608,18] [%emt 0:00:02] (Ne3) } 18.fxe5 { [%eval 421,20] [%emt 0:00:03] } 18...Qh2+ { [%eval -1482,23] [%emt 0:00:01] } 19.Kxg4 { [%eval 502,19] [%emt 0:00:01] } 19...h5+ { [%eval -32741,33] [%emt 0:00:01] (Bxg2) } 20.Kg5 { [%eval 778,18] [%emt 0:00:01] } 20...Qg3+ { [%eval -32743,44] [%emt 0:00:02] } 21.Kxh5 { [%eval 862,19] [%emt 0:00:02] } 21...g6+ { [%eval -32745,48] [%emt 0:00:02] (Bxg2) } 22.Kh6 { [%eval -4165,16] [%emt 0:00:01] } 22...Kh8 { [%eval -32747,51] [%emt 0:00:02] } 23.Qe4 { [%eval -32749,62] [%emt 0:00:04] } 23...Qh2+ { [%eval -32749,57] [%emt 0:00:02] (Bxe4) } 24.Kxg6 { [%eval -32751,59] [%emt 0:00:01] } 24...Qg3+ { [%eval -32751,59] [%emt 0:00:02] (Bxe4+) } 25.Kh5 { [%eval -32753,75] [%emt 0: 00:01] } 25...Rg8 { [%eval -32753,75] [%emt 0:00:02] } 26.Qh7+ { [%eval -32755,245] [%emt 0:00:00] } 26...Kxh7 { [%eval -32755,245] [%emt 0:00:00] } 27.Rf7+ { [%eval -32757,245] [%emt 0:00:00] } 27...Kh8 { [%eval -32757,245] [%emt 0:00:00] } 28.Rh7+ { [%eval -32759,245] [%emt 0:00:00] } 28...Kxh7 { [%eval -32759,245] [%emt 0:00:00] } 29.Bd3+ { [%eval -32761,245] [%emt 0:00:00] } 29...Qxd3 { [%eval -32761,245] [%emt 0:00: 00] } 30.g3 { [%eval -32763,245] [%emt 0:00:00] (Nb1) } 30...Qxg3 { [%eval -32763,245] [%emt 0:00:00] (Qg6+) } 31.b4 { [%eval -32765,245] [%emt 0:00:00] (Nxe6) } 31...Rg5# { [%eval -32765,245] [%emt 0:00:00] } 0-1

AlexChess · Post by **AlexChess** » Sun Feb 20, 2022 1:05 pm

AlexChess wrote: ↑Sun Feb 20, 2022 11:25 am For who continues to say that derivatives will never offer more than Stockfish, here there is an incredible game of Stockfish 14.1 vs Blue Marlin 14.6b21 of my friend dorsz. (I'm fine tuning Proteus Chess SF against his top engines)

Post-game analysis by Stockfish 14.1

[Event "sweras-44"] [Site "?"] [Date "2021.12.06"] [Round "19.2"] [White "Stockfish 14.1-modern"] [Black "Blue Marlin 14.6b21"] [Result "0-1"] [EventDate "2021.12.05"] [FEN "rnbqkb1r/p2p1ppp/4Pn2/1pp5/2P5/5N2/PP2PPPP/RNBQKB1R b KQkq - 0 1"] [SetUp "1"] 1...fxe6 2.cxb5 a6 3.e3 d5 4.Nc3 Nbd7 5.bxa6 Be7 6.a4 O-O 7.Bb5 Bxa6 8.O-O Bd6 9.b3 Bb7 $6 { +1.26/34 } ( 9...Ne5 { +0.93/38 } 10.Ng5 Bc8 11.f4 h6 12.Nh3 Nf7 13.Be2 Be7 14.Nf2 Nd6 15.Nb5 Ba6 16.Nxd6 Bxe2 17.Qxe2 Bxd6 18.e4 Be7 19.e5 Nd7 20.Qg4 Rf5 21.Bd2 Qb6 22.Rab1 Qa6 23.Qh3 Rf7 24.a5 Rb8 25.Qd3 c4 26.bxc4 Rxb1 27.Rxb1 Bc5 ) 10.Bb2 Qe7 11.a5 d4 $2 { +2.05/44 } ( 11...Nb8 { +0.94/39 } 12.Qe2 Na6 13.e4 dxe4 14.Ng5 Be5 15.Bc4 Nc7 16.Rfb1 e3 17.fxe3 Ncd5 18.Nd1 Bxb2 19.Rxb2 h6 20.Nf3 Ne4 21.Nf2 Nd6 22.Ng4 Qc7 23.a6 Nxc4 24.Qxc4 Nb4 25.Qxe6+ Qf7 26.Qxf7+ Rxf7 27.Nge5 Re7 28.Rd2 Bxf3 29.Nxf3 Rxe3 30.Rd6 Re7 31.Ra5 Rea7 32.Rxc5 Rxa6 33.Rxa6 Rxa6 34.Ne5 Rd6 35.Rb5 Nd3 36.Rb8+ Kh7 ) 12.exd4 cxd4 13.Nxd4 Bxh2+ 14.Kxh2 Ne5 15.f4 $2 { 0.00/43 } ( 15.Be2 { +1.41/32 } 15...Nh5 16.g3 Rf4 17.Kg1 Raf8 18.Bxh5 Rh4 19.f4 Qg5 20.Nce2 Rxh5 21.Kf2 Ba6 22.b4 Qg6 23.Ra3 Rh2+ 24.Ke1 Rd8 25.fxe5 Bxe2 26.Nxe2 Rxd1+ 27.Kxd1 Qe4 28.Re1 Qd5+ 29.Bd4 Rxe2 30.Kxe2 Qxd4 31.Kf1 Kf8 32.Ree3 Qc4+ 33.Kf2 Qxb4 34.a6 ) 15...Nfg4+ 16.Kg3 Qf6 17.Qe2 $4 { -9.61/32 } ( 17.Ba3 { 0.00/41 } 17...Rac8 18.Nc6 Ne3 19.Qh5 Nxc6 20.Ne4 Qd4 21.Ng5 h6 22.a6 Qd5 23.Qh3 Nd4 24.axb7 Qxb5 25.bxc8=Q Nxf1+ 26.Rxf1 Rxc8 27.Nxe6 Rc3+ 28.Rf3 Nxf3 29.gxf3 Qe2 30.Qg4 Qe1+ 31.Kh3 Qh1+ 32.Kg3 Qe1+ ) 17...Qh6 18.fxe5 $4 { #-14/61 } ( 18.Qxe5 { -9.41/34 } 18...Nxe5 19.Nce2 g5 20.Kf2 gxf4 21.Bc4 Nxc4 22.bxc4 Qh2 23.Ke1 Bxg2 24.Nf3 Bxf3 25.Rxf3 e5 26.Rc1 Rae8 27.c5 Qh1+ 28.Kf2 Qh4+ 29.Kf1 e4 30.Rxf4 Qh2 31.c6 e3 32.Rxf8+ Rxf8+ 33.Ke1 Qh1+ 34.Ng1 ) 18...Qh2+ 19.Kxg4 h5+ 20.Kg5 Qg3+ 21.Kxh5 g6+ 22.Kh6 Kh8 23.Qe4 Qh2+ 24.Kxg6 Qg3+ 25.Kh5 Rg8 26.Qh7+ Kxh7 27.Rf7+ Kh8 28.Rh7+ Kxh7 29.Bd3+ Qxd3 30.g3 Qxg3 31.b4 Rg5# 0-1your webbrowser and/or your host do not support iframes as required to display the chess board

AlexChess · Post by **AlexChess** » Sat Feb 26, 2022 6:31 pm

Very good performance of Proteus Chess SF v003 vs Dragon 2.6.1 on blitz 3 mins + 1 sec /all, only 1 loss|9 wins / 32 games!

Modifications over previous version are related to pawn structures for a more aggressive play style.

Next test is against latest Stockfish-dev

AlexChess · Post by **AlexChess** » Sat Feb 26, 2022 7:13 pm

AlexChess wrote: ↑Sat Feb 26, 2022 6:31 pm Very good performance of Proteus Chess SF v003 vs Dragon 2.6.1 on blitz 3 mins + 1 sec /all, only 1 loss|9 wins / 32 games!

All Proteus Chess SF v004 versus Dragon 2.6.1 PGN games: https://onedrive.live.com/?cid=B0F79706 ... 18&o=OneUp

ProteusChessSF - Developing my Stockfish 14.1 Polyglot MTCS derivative

Re: ProteusChessSF - Developing my Stockfish 14.1 Polyglot MTCS derivative

Re: ProteusChessSF - Developing my Stockfish 14.1 Polyglot MTCS derivative

Re: ProteusChessSF - Developing my Stockfish 14.1 Polyglot MTCS derivative

Re: ProteusChessSF - Developing my Stockfish 14.1 Polyglot MTCS derivative

Re: ProteusChessSF - Developing my Stockfish 14.1 Polyglot MTCS derivative

Re: ProteusChessSF - Developing my Stockfish 14.1 Polyglot MTCS derivative

Re: ProteusChessSF - Developing my Stockfish 14.1 Polyglot MTCS derivative

Re: ProteusChessSF - Developing my Stockfish 14.1 Polyglot MTCS derivative

Re: ProteusChessSF - Developing my Stockfish 14.1 Polyglot MTCS derivative

Re: ProteusChessSF - Developing my Stockfish 14.1 Polyglot MTCS derivative