Proteus Chess SF - developing my Stockfish 14.1 derivative.

Hi chess engines entusiasts!

That will be my development page with notes, suggestions and request by friends.

You can download always updated builds for Windows, Mac and Android here:

https://www.chess.com/forum/view/game-analysis/freeware-against-commercial-chess-software-arena-3-5-1-vs-chessbase-fritz-17?page=1


First independent and positive result against the ultra-strong AI 23.4 from @Oliprog !!!

AlexChess wrote:

Hi chess engines entusiasts!

That will be my development page with notes, suggestions and request by friends.

You can download always updated builds for Windows, Mac and Android here:

https://www.chess.com/forum/view/game-analysis/freeware-against-commercial-chess-software-arena-3-5-1-vs-chessbase-fritz-17?page=1


First independent and positive result against the ultra-strong AI 23.4 from @Oliprog !!!

http://microsmeta.com/microsmeta.com/public/Proteus5-AI243.png

AlexChess wrote:

Thanks Oliprog for your matches! Is very important to have third part results

ProteusChessSFv005 vs CorChess 220316

Strangely, against CorChess Proteus Chess SF v005 is stronger on long time control



ALL PGN GAMES

Proteus Chess SF v005 vs Zeus 24.3



All games in PGN: https://onedrive.live.com/?cid=B0F797063D1EB745&id=B0F797063D1EB745%2110924&parId=B0F797063D1EB745%2110618&o=OneUp

Thanks again to Oliprog and to all my friends on Outskirts

Respect to Zeus 24.3 Team!

(8 CPUs - 7 Mn/s on starting position, TB Syzygy 6 man. Opening book of 3 full moves)



ALL GAMES IN PGN: https://onedrive.live.com/?cid=B0F797063D1EB745&id=B0F797063D1EB745%2110927&parId=B0F797063D1EB745%2110657&o=OneUp

NOTES:
-Igel 3.0.5 (also if I have added net in config) probably is playing WITHOUT NNUE.
-Ethereal 13.00 github is playing without NNUE due to author's choice, making it commercial and only for Windows AVX2 SSSE3.
-Stockfish and derivatives are still a bit stronger than originals, but Koivisto 8.2 and Fire 8.NN.MC.3 aren't still available.
-Sugar 2.4.0 isn't the latest version - I can't find 2.5.0 anymore (project closed by Andrea Manzo)

AlexChess wrote:

(8 CPUs - 7 Mn/s on starting position, TB Syzygy 6 man. Opening book of 3 full moves)



ALL GAMES IN PGN: https://onedrive.live.com/?cid=B0F797063D1EB745&id=B0F797063D1EB745%2110927&parId=B0F797063D1EB745%2110657&o=OneUp

NOTES:
-Igel 3.0.5 (also if I have added net in config) probably is playing WITHOUT NNUE.
-Ethereal 13.00 github is playing without NNUE due to author's choice, making it commercial and only for Windows AVX2 SSSE3.
-Stockfish and derivatives are still a bit stronger than originals, but Koivisto 8.2 and Fire 8.NN.MC.3 aren't still available.
-Sugar 2.4.0 isn't the latest version - I can't find 2.5.0 anymore (project closed by Andrea Manzo)

The link doesn't work. Can I have a mirror?!

Pixeldrain
https://pixeldrain.com/

AlexChess wrote:

Proteus Chess SF v005 vs Zeus 24.3



All games in PGN: https://onedrive.live.com/?cid=B0F797063D1EB745&id=B0F797063D1EB745%2110924&parId=B0F797063D1EB745%2110618&o=OneUp

Thanks again to Oliprog and to all my friends on Outskirts

Respect to Zeus 24.3 Team!

This link doesn't work either

AlexChess wrote:

ProteusChessSFv005 vs CorChess 220316

Strangely, against CorChess Proteus Chess SF v005 is stronger on long time control  



ALL PGN GAMES

This link doesn't work either

lisandr wrote:

AlexChess wrote:

ProteusChessSFv005 vs CorChess 220316

Strangely, against CorChess Proteus Chess SF v005 is stronger on long time control  



ALL PGN GAMES

This link doesn't work either

mirror:

Top Engines Mac M1 Blitz 3+2.pgn.zip
https://pixeldrain.com/u/yRzpuRKW

Engmatch 9 -1456.pgn.zip
https://pixeldrain.com/u/5GgUk4RF

ProteusChessSFv05-CorChess220323-Active30min.pgn.zip
https://pixeldrain.com/u/uaUS5rTp

Congratulations to Mehmet and thanks to Oliprog and all my friends!

(Already started ProteusChessSFv006 . Now it would be interesting to know the performances of Kayra 1.3 vs AI and Zeus)

Kind regards, Alex

Good luck with yoir project alex..
I can't wait to see how strong is it ProteusChess V006 really are ..

I mean sinced even the previous version are still standing among the strongest of engines list

Zippo wrote:

Good luck with yoir project alex..
I can't wait to see how strong is it ProteusChess V006 really are ..

I mean sinced even the previous version are still standing among the strongest of engines list

A Long night of tests against Kayra 1.3 are waiting my Mac mini M1

I will test deeply with Mehmet engine, CorChess and Stockfish before releasing it

Proteus Chess SF v006 preparation to @Oliprog 's tournaments




To get reliable results, this is

Biltz tournament vs Kayra 1.3 3 mins + 2 secs / all on mac M1 6.5 Mn/s TBs 6 man TCEC7 pgn book 612 positions switching colors.



Active 30 mins + 5 secs tournaments on Android 8 CPUs vs SF220313 & CorChess 3 220330 500 Kn/s TCEC 7 pgn book 612 positions switching colors.

Sadly I cannot test private AI 24.x and Zeus 24.3 If positive results I will compile v006 for Windows, in doubt I will repeat the 3
tests using dorsz DOT4 or DOT5 umbalanced openings test suite.

https://www.cpubenchmark.net/cpu.php?cpu=AMD+A10-7850K+APU&id=2133

https://www.cpubenchmark.net/cpu.php?cpu=Apple+M1+8+Core+3200+MHz&id=4104

Best regards my friends!

Alex

@Oliprog how many MN/s calculates Proteus Chess SF v005 on your AMD A10 on starting position? So I can calibrate my tests to your hardware speed.

Alex

Hi Alex when ur done with ur Proteus V006 project, please don't forget to release it in Android platfotm either.. sinced Im one the Proteus biggest fans..

Thank u & Good luck

First impressions OK Proteus Chess SF v006 play style is still active but less risky

Just seen AI results against Kayra 1.3 It Is very difficult to evaluate engines strenght!

AlexChess wrote:

Just seen AI results against Kayra 1.3 It Is very difficult to evaluate engines strenght!

It seems that an important part of the strength of engines like AI or Zeus is related to their specific nnue nets(Zeus nets). As mehmet mentioned before providing neural networks needs to have very sophisticated hardwares. So it’s very difficult to compete with these kinds of engines. Like Zeus,AI,and fat titz. And needs to work very hard on project. And may be need a complete teamwork. I wish you success Alex.
Best regards

Homayoun wrote:

AlexChess wrote:

Just seen AI results against Kayra 1.3 It Is very difficult to evaluate engines strenght!

It seems that an important part of the strength of engines like AI or Zeus is related to their specific nnue nets(Zeus nets). As mehmet mentioned before providing neural networks needs to have very sophisticated hardwares. So it’s very difficult to compete with these kinds of engines. Like Zeus,AI,and fat titz. And needs to work very hard on project. And may be need a complete teamwork. I wish you success Alex.
Best regards

ProteusChessSF beats Zeus and AI for many times in Oliprog tests. At last Oliprog tests ProteusChessSF beat AI, Kayra beat ProteusChessSF and AI beat Kayra. It isnt possible to say with a clear result which of the engine is the most powerful.
But dont forget Kayra and ProteusChessSF are new engines, they are open source engines and they don' t violate GPL rules.

mehmet1921 wrote:

Homayoun wrote:

AlexChess wrote:

Just seen AI results against Kayra 1.3 It Is very difficult to evaluate engines strenght!

It seems that an important part of the strength of engines like AI or Zeus is related to their specific nnue nets(Zeus nets). As mehmet mentioned before providing neural networks needs to have very sophisticated hardwares. So it’s very difficult to compete with these kinds of engines. Like Zeus,AI,and fat titz. And needs to work very hard on project. And may be need a complete teamwork. I wish you success Alex.
Best regards

ProteusChessSF beats Zeus and AI for many times in Oliprog tests. At last Oliprog tests ProteusChessSF beat AI, Kayra beat ProteusChessSF and AI  beat Kayra. It isnt possible to say with a clear result which of the  engine  is the most powerful.
But dont forget Kayra and ProteusChessSF are new engines,  they are open source engines and they don' t violate GPL rules.

Would be interesting an all top engines @Oliprog tournament. And yes I have released it as GPLv3, removing from my tests not compliant engines to avoid problems and to thank Stockfish Team that changed chess engines market forever, with builds for all operating systems (I will add Linux build soon) and hardwares, not like Ethereal and Zahak that support only AVX2 and SSSE3, ignoring fans with older hardware on purpose

Zippo wrote:

Hi Alex when ur done with ur Proteus V006 project, please don't forget to release it in Android platfotm either.. sinced Im one the Proteus biggest fans..

Thank u & Good luck

Obviously... And is doing already well on Snapdragon 870 8 CPUs 1.6 Mn/s bullet against Kayra 1.3. dors DOTS4 unbalanced book 400 positions, switched colors to have the most reliable results    

mehmet1921 wrote:

Homayoun wrote:

It seems that an important part of the strength of engines like AI or Zeus is related to their specific nnue nets(Zeus nets). As mehmet mentioned before providing neural networks needs to have very sophisticated hardwares. So it’s very difficult to compete with these kinds of engines. Like Zeus,AI,and fat titz. And needs to work very hard on project. And may be need a complete teamwork. I wish you success Alex.
Best regards

Finally I have found a way to test engines in a more scientific way, using DOTS4 unbalanced pgn 400 positions opening test suite and TCEC 7 balanced 306 positions (2014, so after 8 years consumer hardware probably matches TOP PC speed requisites) switching sides for each position. I will test Proteus Chess SF v006 against Fat Titz 2 both popcnt sse41 con Windows 11 ARM64 with these opening books. Now Polyglot is almost useless for me, since BSG let switching sides only using PGN and .bin often suddenly stops loading on BSG after crashes.

Right now i prefer handcrafted fine tuning Proteus instead using Fishtest or other tools, but I'm studying them. I'm also adding learning to next Proteus releases.