Proteus Chess SF v004

Proteus Chess SF v004 is a free GPLv3 Stockfish derivative. I have made some fine tuning to make Stockfish more aggressive. In the next releases I will add patterns for sacrificing material for positional advantage and Castling attacks schemas.

You find latest builds for Windows 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

Best regards,
Alessandro Morales
AlexChess

Hi Alex
Thanks. Very nice and strong engine wish you success in further stages too until you achieve all your goals about your project.
Best regards

hola alex gracias por compartir este modulo,es el mas fuerte que he probado.

Just to remind that ProteusChess is renamed (fair will be said it's a clone) of SF-PB-250222, with some "amateur" value changes, for pawns, pawn penalties and KingOnFile. Some people beleve that they are smarter that Stockfish team.
AMD Ryzen 7 1800X, 8 cores / 16 threads, 64-bit, 3.60 GHz, 16 GB RAM, Windows 10 pro
GUI: Cutechess 1.2 /RR / 1 core per engine / Hash 256 / TC: 3min / book: 8moves_V3 / Concurrency 8 / Ponder OFF /
For GUI syzygy 5 for draw adjudication.

Games https://pixeldrain.com/u/uCzpakX5

ChessFan1 wrote:

Just to remind that ProteusChess is renamed (fair will be said it's a clone) of SF-PB-250222, with some "amateur" value changes, for pawns, pawn penalties and KingOnFile. Some people beleve that they are smarter that Stockfish team.  
AMD Ryzen 7 1800X, 8 cores / 16 threads, 64-bit, 3.60 GHz, 16 GB RAM, Windows 10 pro
GUI: Cutechess 1.2 /RR / 1 core per engine / Hash 256 / TC: 3min / book: 8moves_V3 / Concurrency 8 / Ponder OFF /
For GUI syzygy 5 for draw adjudication.

Code:

# PLAYER                     :  RATING  ERROR  PLAYED   (%)    W     D    L  D(%)  CFS(%)
1 SF-PB-250222_x64_AVX2      :    3202      1    4016  50.5  153  3752  111  93.4     100
2 ProteusChessSFv004-avx2    :    3198      1    4016  49.5  111  3752  153  93.4     ---

Games https://pixeldrain.com/u/uCzpakX5

Hi chess fan1
At first thank you very much for your SF pb which is the basis for proteus too. I think here no one has the claim that is smarter than anyone else. The subject is that as you know better than me stockfish nnue is not so much aggressive in playing as it was before. Some chess fans who know about chess programming too are trying to make it a little more aggressive. For example Dorze with blue marlin, Mehmet with kayra, and now Alex with proteus. They don’t have any claim that are smarter than stockfish team. They just want to create different playing style that would be more lovely according to chess fans. And I don’t see any problem here. You are one of the main members in OCF too. So it is not necessary that I remind you that proteus has very brilliant results in test sections there. So let give time and opportunity to this kind of engines and see what happens in future.

Best regards

Smarter than sf team :
- nnue authors
- alphazero team
- private chess engines authors
- authors of chess engines with learning function
- book creators
- etc.

Homayoun wrote:

]
Hi chess fan1
At first thank you very much for your SF pb which is the basis for proteus too. I think here no one has the claim that is smarter than anyone else. The subject is that as you know better than me stockfish nnue is not so much aggressive in playing as it was before. Some chess fans who know about chess programming too are trying to make it a little more aggressive. For example Dorze with blue marlin, Mehmet with kayra, and now Alex with proteus. They don’t have any claim that are smarter than stockfish team. They just want to create different playing style that would be more lovely according to chess fans. And I don’t see any problem here. You are one of the main members in OCF too. So it is not necessary that I remind you that proteus has very brilliant results in test sections there. So let give time and opportunity to this kind of engines and see what happens in future.

Best regards

Hi Homayoun, i understand your point of view, but don't you agree that before someone begin making changes in the source code, he must to have knowledge in C++ programing? Making some random changes doesn't lead anywhere, engine become only worst, but most important is that all this kind of "amateur" engines confusing people, and they spend they time for nothing (like i did in above test, only 4016 games and regression -4 elo). Regarding brilliant results - i didn't see any of them. If you are talking about engine's "tactics solving" abilities, there is a much better engines (made by professionals), eg Crystal, Black Diamond etc.
If people want to use engines like proteus, that's their choice. Just next time, let Alex take the original code and not from SF-PB, SF-PB is made in memory of Brainfish.
Regards.

deeds wrote:

Smarter than sf team :
- nnue authors
- alphazero team
- private chess engines authors
- authors of chess engines with learning function
- book creators
- etc.

Hello my friend.
- nnue authors created network, not the Stockfish.
- alphazero team who came out of nowhere, made the test versus older Stockfish (with unknown conditions) and claimed that "alphazero" is the strongest engine in the world. In the meantime, we see every time from hundreds of tests how Lc0 loses every time to Stockfish.
- Regarding - private chess engines authors, Khalid is good programmer, but Eman is not strong enough, besides it is 90% Stockfish.
About other private engines, i'll better not to speak. So far, none of these authors have written an engine from scratch that will be better than the original.
- Learning function is also not engine, let's not confuse this things. Same for the "opening books", they have nothing to do with Stockfish.
Regards.

deeds wrote:

Smarter than sf team :
- nnue authors
- alphazero team
- private chess engines authors
- authors of chess engines with learning function
- book creators
- etc.

I forgot endgame tablebases creators too.

All of them used smarter techniques with higher elo gain than sf team. First nets based on the NNUE technology already brought tens of elo when sf team were stuck on ridiculous elo gains. The A0 team has demonstrated that by self-training from scratch the chess engine was much stronger in few hours/days than sf team while years. No interest to use a chess engine that does the same errors because he can't learn.

ChessFan1 wrote:

Homayoun wrote:

]
Hi chess fan1
At first thank you very much for your SF pb which is the basis for proteus too. I think here no one has the claim that is smarter than anyone else. The subject is that as you know better than me stockfish nnue is not so much aggressive in playing as it was before. Some chess fans who know about chess programming too are trying to make it a little more aggressive. For example Dorze with blue marlin, Mehmet with kayra, and now Alex with proteus. They don’t have any claim that are smarter than stockfish team. They just want to create different playing style that would be more lovely according to chess fans. And I don’t see any problem here. You are one of the main members in OCF too. So it is not necessary that I remind you that proteus has very brilliant results in test sections there. So let give time and opportunity to this kind of engines and see what happens in future.

Best regards

Hi Homayoun, i understand your point of view, but don't you agree that before someone begin making changes in the source code, he must to have knowledge in C++ programing? Making some random changes doesn't lead anywhere, engine become only worst, but most important is that all this kind of "amateur" engines confusing people, and they spend they time for nothing (like i did in above test, only 4016 games and regression -4 elo). Regarding brilliant results - i didn't see any of them. If you are talking about engine's "tactics solving" abilities, there is a much better engines (made by professionals), eg Crystal, Black Diamond etc.
If people want to use engines like proteus, that's their choice. Just next time, let Alex take the original code and not from SF-PB, SF-PB is made in memory of Brainfish.
Regards.

I completely agree with you that when a person enters in this field schould have the knowledge of c programming and even should study and learn more during the project and random changes in sieve code of an engine can just be confusing for chess fans and engine users. It is completely correct idea. But I want to add a point. About the proteus results that I mentioned because I don’t have the habit to speak without reference let me defend from the point. You can reffer to test sections in OCF (Oliprog and Ghppn tests) and see again that proteus has won the latest Zeus, Ai, Kayra,and latest SFpb. And you have rated positive to the posts too.

Regards

deeds wrote:

deeds wrote:

Smarter than sf team :
- nnue authors
- alphazero team
- private chess engines authors
- authors of chess engines with learning function
- book creators
- etc.

I forgot endgame tablebases creators too.

All of them used smarter techniques with higher elo gain than sf team. First nets based on the NNUE technology already brought tens of elo when sf team were stuck on ridiculous elo gains. The A0 team has demonstrated that by self-training from scratch the chess engine was much stronger in few hours/days than sf team while years. No interest to use a chess engine that does the same errors because he can't learn.

Well, allow me to disagree, again. TBs along with the all the mentioned above has nothing to do with evolution of the any of the engines. Simply put, what are the networks, learning experience files, TBs etc, by them selfs? Nothing, because nothing of mentioned can work without the engine it self, but engine can work without all this.
Regarding creation of TBs, networks and so on, for each one of them programmers are using separate type of programing (techniques and required knowledge).

Homayoun wrote:

About the proteus results  that I mentioned because I don’t have the habit to speak without reference let me defend from the point. You can reffer to test sections in OCF (Oliprog and Ghppn tests) and see again that proteus has won the latest Zeus, Ai, Kayra,and latest SFpb. And you have rated positive to the posts too.

Regards

Yes, i have see all those tests, but there is a huge difference between "Tests for fun" (like Oliprog's) and professional tests. Tests for fun with small number of games can't be taken as serious, due to huge number of error bars. We can't claim that, engine A is stronger than engine B after 200, 300 or 600 games (where each engine will play 100, 150, 300 games - switching colors). It's like saying that chess are limited only to 10 or 15 opening lines, and nothing else can't be played, while only after first 4 moves there is 170.000 possible combinations.
Regards.

no need any engine to request moves from experience files, endgame tablebases or books. And all of them solved more parts of chess than any engine ! There are tons of people smarter than sf team and their technologies often bring more elo than tons of daily sf dev releases. Private chess engines authors don't wait for sf team find +0.89 elo after 80k games. Even the sf nets are based on lc0 data. There are tons of motherboards that handles 7 and more graphic cards so it's a joke to see lc0 on limited hardwares in order to pleasure sf community.

ChessFan1 wrote:

Homayoun wrote:

About the proteus results  that I mentioned because I don’t have the habit to speak without reference let me defend from the point. You can reffer to test sections in OCF (Oliprog and Ghppn tests) and see again that proteus has won the latest Zeus, Ai, Kayra,and latest SFpb. And you have rated positive to the posts too.

Regards

Yes, i have see all those tests, but there is a huge difference between "Tests for fun" (like Oliprog's) and professional tests. Tests for fun with small number of games can't be taken as serious, due to huge number of error bars. We can't claim that, engine A is stronger than engine B after 200, 300 or 600 games (where each engine will play 100, 150, 300 games - switching colors). It's like saying that chess are limited only to 10 or 15 opening lines, and nothing else can't be played, while only after first 4 moves there is 170.000 possible combinations.
Regards.

Yes exactly. And that’s the reason I suggested that let’s see what happens in future. Proteus may be a successful project or may be not. It depends on many factors including author knowledge, correct and scientific tests and,...etc. Let everything continue normally and see what would happen in future. I personally wish success for every new attempts and ideas.
Best regards

people don't need to be smarter than any team to success too

deeds wrote:

no need any engine to request moves from experience files, endgame tablebases or books. And all of them solved more parts of chess than any engine ! There are tons of people smarter than sf team and their technologies often bring more elo than tons of daily sf dev releases. Private chess engines authors don't wait for sf team find +0.89 elo after 80k games. Even the sf nets are based on lc0 data. There are tons of motherboards that handles 7 and more graphic cards so it's a joke to see lc0 on limited hardwares in order to pleasure sf community.

A TB isn't even the same kind of development as a chess engine. A tablebase tells you explicitly the move order when the game simplifies to a certain number of pieces. It's a specialized tool specifically for endgame situations. Of course a tablebase specifically meant for resolving endgames is going to be better than an engine that calculates through the lines. It's supposed to be.

This is like saying the airbag manufacturer is smarter than the engineers working on the v8 engine, because the engine isn't as good at protecting you from a crash. They're completely different tools meant for different purposes. There's a reason why SF has tablebase options that allow you to reference your tablebase.

ChessFan1 wrote:

Just to remind that ProteusChess is renamed (fair will be said it's a clone) of SF-PB-250222, with some "amateur" value changes, for pawns, pawn penalties and KingOnFile. Some people beleve that they are smarter that Stockfish team.  
AMD Ryzen 7 1800X, 8 cores / 16 threads, 64-bit, 3.60 GHz, 16 GB RAM, Windows 10 pro
GUI: Cutechess 1.2 /RR / 1 core per engine / Hash 256 / TC: 3min / book: 8moves_V3 / Concurrency 8 / Ponder OFF /
For GUI syzygy 5 for draw adjudication.

Code:

# PLAYER                     :  RATING  ERROR  PLAYED   (%)    W     D    L  D(%)  CFS(%)
1 SF-PB-250222_x64_AVX2      :    3202      1    4016  50.5  153  3752  111  93.4     100
2 ProteusChessSFv004-avx2    :    3198      1    4016  49.5  111  3752  153  93.4     ---

Games https://pixeldrain.com/u/uCzpakX5

Hi Chessfan1,

Just a little invidious of my results, uh?

Maybe you are able to compare software with a tool, but sadly you ignore the meaning and value of Open Source. I can FREELY use SF-PB made by Skynet from Venus that hiddens his name (good idea for trollers and haters, less to moderators) Stockfish+ (thank Norman Schmidt) , Stockfish Polyglot (thanks Massimiliano Goi) and every source if they are GPLv3 and I release as GPLv3

...And about Stockfish Team that I admire and respect (differenty from people not releasing their source, in the interest of the whole chess community)

From Stockfish official page:

Get Involved
Whether you are a programming wizard or just a fan, there’s something you can do to help make Stockfish even better.

Fishtest
Fishtest lets you volunteer your computer’s processing power to help test future versions of Stockfish. No coding required! All you need to do is install the Fishtest worker, and your computer will automatically start playing chess games using future and current versions of the engine.

Best regards, Alessandro Morales
Alexchess (working now on ProteusChessSF v005 for my friends and fans)

Chessfan1-Skynet I wait for your own original engine.

PS: You have a strange attitude to offend people:
Here your OT demential humor against Ukrainian Chessman, while you block members simply thanking Archimedes for his hard work. MZ (that's a very good guy) doesn't deserve to have people like you administering his forum.

http://outskirts.altervista.org/forum/viewtopic.php?p=40519&view=single_post#p40519

Homayoun wrote:

ChessFan1 wrote:

Homayoun wrote:

About the proteus results  that I mentioned because I don’t have the habit to speak without reference let me defend from the point. You can reffer to test sections in OCF (Oliprog and Ghppn tests) and see again that proteus has won the latest Zeus, Ai, Kayra,and latest SFpb. And you have rated positive to the posts too.

Regards

Yes, i have see all those tests, but there is a huge difference between "Tests for fun" (like Oliprog's) and professional tests. Tests for fun with small number of games can't be taken as serious, due to huge number of error bars. We can't claim that, engine A is stronger than engine B after 200, 300 or 600 games (where each engine will play 100, 150, 300 games - switching colors). It's like saying that chess are limited only to 10 or 15 opening lines, and nothing else can't be played, while only after first 4 moves there is 170.000 possible combinations.
Regards.

Yes exactly. And that’s the reason I suggested that let’s see what happens in future. Proteus may be a successful project or may be not. It depends on many factors including author knowledge, correct and scientific tests and,...etc. Let everything continue normally and see what would happen in future. I personally wish success for every new attempts and ideas.
Best regards

Thank you dear friend Homayoun, and thank you Oliprog for you very professional and appreciated tests. v004 is only a fine tuning (but already with very good results ) I'm studying SF-dev in depth with very skilled and known developers that I don't want t reveal to avoid to expose them to Skynet silly attacks. We have a lot of ideas for improving his positional knowledge in LT analysis but also to make it more aggressive.

Kind regards, Alex

AlexChess wrote:

Hi Chessfan1,

Just a little invidious of my results, uh?  

LOL, only a kid (with your IQ) will believe in your results.. Keep live in your world.

AlexChess wrote:

Maybe you are able to compare software with a tool, but sadly you ignore the meaning and value of Open Source. I can FREELY use SF-PB made by Skynet from Venus that hiddens his name (good idea for trollers and haters, less to moderators) Stockfish+ (thank Norman Schmidt) , Stockfish Polyglot (thanks Massimiliano Goi) and every source if they are GPLv3 and I release as GPLv3

...And about Stockfish Team that I admire and respect (differenty from people not releasing their source, in the interest of the whole chess community)

From Stockfish official page:

Get Involved
Whether you are a programming wizard or just a fan, there’s something you can do to help make Stockfish even better.

Fishtest
Fishtest lets you volunteer your computer’s processing power to help test future versions of Stockfish. No coding required! All you need to do is install the Fishtest worker, and your computer will automatically start playing chess games using future and current versions of the engine.

Do whatever you want, next time you'll be updating the code by yourself, with your unique skills.

AlexChess wrote:

Chessfan1-Skynet I wait for your own original engine.
 
PS: You have a strange attitude to offend people:
Here your OT demential humor against Ukrainian Chessman, while you block members simply thanking Archimedes for his hard work. MZ (that's a very good guy) doesn't deserve to have people like you administering his forum.

http://outskirts.altervista.org/forum/viewtopic.php?p=40519&view=single_post#p40519

Chessman is my friend for more than 6 years.. What you posted here, once again proves that you are a MORON.. Where did you saw the "dementia humor"?? Learn english first.
Every forum has some rules, we are following them, if you want to thank someone, PRESS "Thank you" button instead of flooding and spamming, you still don't get it don't ya? Btw, how do you know that i'm the person who is banning people on altervista? Do you have any evidence? Oh no, you don't have any, i forgot that you are just SPAMMER and TROLL.
People say - "some people are born idiots, some people become idiots". In your case, you was and you still an idiot. Next time take your pills before you write nonsense again, because asylum is waiting your arrival.

AlexChess wrote:

I don't want t reveal to avoid to expose them to Skynet silly attacks..

Oh no, baby boy, you was attacked again?
Good luck with your projects.

ChessFan1 wrote:

AlexChess wrote:

I don't want t reveal to avoid to expose them to Skynet silly attacks..

Oh no, baby boy, you was attacked again?
Good luck with your projects.

When you will learn Italian

https://context.reverso.net/traduzione/inglese-italiano/demential

Then thanks Chessman for publishing SF-PB for the whole chess community.

Your are right, your it's not even humor more likely grudge against the World. Be constructive, not destructive.

If you think that being a simple moderator without other skills is interesting, go on.

And YOU are a baby, I'm 57.

Kind regards,
Alessandro Morales, aka AlexChess

...luckily not Skynet from Third Dimension or Chessfan1 from Saturn
Are you ashamed of your real name or about your behavior permanently blocking members you offend, to be sure that they cannot reply??

@ChessFan1

You are offending even @Oliprog :


"Yes, i have see all those tests, but there is a huge difference between "Tests for fun" (like Oliprog's) and professional tests. Tests for fun with small number of games can't be taken as serious, due to huge number of error bars. We can't claim that, engine A is stronger than engine B after 200, 300 or 600 games (where each engine will play 100, 150, 300 games - switching colors)"

What kind of bad guy are you, SKYNET?

Katabatic wrote:

It's a specialized tool specifically for endgame situations. Of course a tablebase specifically meant for resolving endgames is going to be better than an engine that calculates through the lines. It's supposed to be.

Tablebases are not only used at the end of games ! Several private chess engines fully analyze (by Brute Force or not) openings to store the best moves. From what I've seen, we're already at +80 elo from SF14.1, this system is 10 years ahead of the SF team.