New J.A Compiles

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

Moderator: Ras

User avatar
Jim Ablett
Posts: 2214
Joined: Fri Jul 14, 2006 7:56 am
Location: London, England
Full name: Jim Ablett

Re: New J.A Compiles

Post by Jim Ablett »

Image
https://github.com/Lacovipo/Anubis-chess-engine
Anubis 3.01 by José Carlos Martínez Galán

Anubis 3.01 64 JA

Windows avx512 & bmi2 builds

Proton link:
https://drive.proton.me/urls/P33ZCY141C#GsyiX7Q68qoZ

Smash link:
https://fromsmash.com/Anubis301JA

Jim.
User avatar
Jim Ablett
Posts: 2214
Joined: Fri Jul 14, 2006 7:56 am
Location: London, England
Full name: Jim Ablett

Re: New J.A Compiles

Post by Jim Ablett »

Image
https://github.com/ruicoelhopedro/pawn
Pawn 4.0.4 Dev by Pedro Rui Coelho

Pawn 4.0.4 Dev 64 JA

Windows/Linux & Android profiled builds

Proton link:
https://drive.proton.me/urls/XNT2STDQ3M#9GnrwmvtY8NP

Smash link:
https://fromsmash.com/Pawn404DevJA

Jim.
Joseph
Posts: 40
Joined: Tue Jul 02, 2024 1:53 pm
Full name: Joseph Masa

Re: New J.A Compiles

Post by Joseph »

Hi Jim,

I was testing the Android build of PlentyChess 6.0.15 JA and noticed what seems to be a bug. The engine played a self-match at 1+1 time control, and the moves looked highly irregular and uncharacteristic for its strength.

Here’s the PGN of the game:

[Event "?"]
[Site "?"]
[Date "2025.08.07"]
[Round "?"]
[White "PlentyChess 6.0.15 JA"]
[Black "PlentyChess 6.0.15 JA"]
[Result "*"]
[TimeControl "1+1"]

1. e4 e5 2. Qh5 Nf6 3. Bc4 Qe7 4. d3 Nxe4 5. dxe4 g6 6. Bxf7+ Kxf7 7. Qe2 Qe6
8. Nf3 Kg7 9. b3 Be7 10. Qd2 h5 11. Bb2 d6 12. Qc3 h4 13. Nbd2 Na6 14. b4 Rh5
15. b5 Nc5 16. g4 hxg3 17. fxg3 Bd7 18. Qa5 *

The early queen sortie, repeated odd pawn pushes, and self-blocking moves don’t match the engine’s usual style. I suspect the Android compilation might not be functioning correctly (possibly a parameter or NNUE load issue), since the PC build doesn’t produce similar play.

Could you check if there’s an initialization or NNUE loading problem in the Android version of 6.0.15 JA?

Thanks for your work on PlentyChess — it’s normally a great engine!


Joseph.
User avatar
Jim Ablett
Posts: 2214
Joined: Fri Jul 14, 2006 7:56 am
Location: London, England
Full name: Jim Ablett

Re: New J.A Compiles

Post by Jim Ablett »

Jim Ablett wrote: Fri Aug 08, 2025 7:33 pm Image
https://github.com/Lacovipo/Anubis-chess-engine
Anubis 3.01 by José Carlos Martínez Galán

Anubis 3.01 64 JA

Windows avx512 & bmi2 builds

Proton link:
https://drive.proton.me/urls/P33ZCY141C#GsyiX7Q68qoZ

Smash link:
https://fromsmash.com/Anubis301JA

Jim.
I've ported the msvc src code to gcc & clang and made all new profiled builds including sse4 & sse3.

Windows & Linux profiled builds

Proton link:
https://drive.proton.me/urls/6YJ9YPS6TR#2smy96g9ETIj

Smash link:
https://fromsmash.com/ANUBIS30164JA

Jim.
User avatar
Jim Ablett
Posts: 2214
Joined: Fri Jul 14, 2006 7:56 am
Location: London, England
Full name: Jim Ablett

Re: New J.A Compiles

Post by Jim Ablett »

Joseph wrote: Sat Aug 09, 2025 3:13 pm Hi Jim,

I was testing the Android build of PlentyChess 6.0.15 JA and noticed what seems to be a bug. The engine played a self-match at 1+1 time control, and the moves looked highly irregular and uncharacteristic for its strength.

Here’s the PGN of the game:

[Event "?"]
[Site "?"]
[Date "2025.08.07"]
[Round "?"]
[White "PlentyChess 6.0.15 JA"]
[Black "PlentyChess 6.0.15 JA"]
[Result "*"]
[TimeControl "1+1"]

1. e4 e5 2. Qh5 Nf6 3. Bc4 Qe7 4. d3 Nxe4 5. dxe4 g6 6. Bxf7+ Kxf7 7. Qe2 Qe6
8. Nf3 Kg7 9. b3 Be7 10. Qd2 h5 11. Bb2 d6 12. Qc3 h4 13. Nbd2 Na6 14. b4 Rh5
15. b5 Nc5 16. g4 hxg3 17. fxg3 Bd7 18. Qa5 *

The early queen sortie, repeated odd pawn pushes, and self-blocking moves don’t match the engine’s usual style. I suspect the Android compilation might not be functioning correctly (possibly a parameter or NNUE load issue), since the PC build doesn’t produce similar play.

Could you check if there’s an initialization or NNUE loading problem in the Android version of 6.0.15 JA?

Thanks for your work on PlentyChess — it’s normally a great engine!


Joseph.
Hi Joseph,

I can confirm. It's playing badly for me too. I'll take a look.

Jim.
User avatar
Jim Ablett
Posts: 2214
Joined: Fri Jul 14, 2006 7:56 am
Location: London, England
Full name: Jim Ablett

Re: New J.A Compiles

Post by Jim Ablett »

Jim Ablett wrote: Sun Aug 10, 2025 1:34 am
Joseph wrote: Sat Aug 09, 2025 3:13 pm Hi Jim,

I was testing the Android build of PlentyChess 6.0.15 JA and noticed what seems to be a bug. The engine played a self-match at 1+1 time control, and the moves looked highly irregular and uncharacteristic for its strength.

Here’s the PGN of the game:

[Event "?"]
[Site "?"]
[Date "2025.08.07"]
[Round "?"]
[White "PlentyChess 6.0.15 JA"]
[Black "PlentyChess 6.0.15 JA"]
[Result "*"]
[TimeControl "1+1"]

1. e4 e5 2. Qh5 Nf6 3. Bc4 Qe7 4. d3 Nxe4 5. dxe4 g6 6. Bxf7+ Kxf7 7. Qe2 Qe6
8. Nf3 Kg7 9. b3 Be7 10. Qd2 h5 11. Bb2 d6 12. Qc3 h4 13. Nbd2 Na6 14. b4 Rh5
15. b5 Nc5 16. g4 hxg3 17. fxg3 Bd7 18. Qa5 *

The early queen sortie, repeated odd pawn pushes, and self-blocking moves don’t match the engine’s usual style. I suspect the Android compilation might not be functioning correctly (possibly a parameter or NNUE load issue), since the PC build doesn’t produce similar play.

Could you check if there’s an initialization or NNUE loading problem in the Android version of 6.0.15 JA?

Thanks for your work on PlentyChess — it’s normally a great engine!


Joseph.
Hi Joseph,

I can confirm. It's playing badly for me too. I'll take a look.

Jim.
I've re-compiled the Android builds >

Proton link:
https://drive.proton.me/urls/WH6GXCWPC8#vZvQIquhfRhS

Smash link:
https://fromsmash.com/Plentychess6015JAAndroid

Jim.
User avatar
Jim Ablett
Posts: 2214
Joined: Fri Jul 14, 2006 7:56 am
Location: London, England
Full name: Jim Ablett

Re: New J.A Compiles

Post by Jim Ablett »

Image
https://github.com/FireFather/fire
Fire 10 by Norman Schmidt

Fire 10 64 JA

Windows/Linux & Android profiled builds

Proton link:
https://drive.proton.me/urls/YAG660J03R#Z4PBHej8KkAT

Smash link:
https://fromsmash.com/Fire10JA

Jim.
Jose Carlos
Posts: 153
Joined: Wed Mar 08, 2006 10:09 pm
Location: Murcia (Spain)
Full name: José C. Martínez Galán

Re: New J.A Compiles

Post by Jose Carlos »

Jim Ablett wrote: Sun Aug 10, 2025 1:11 am
Jim Ablett wrote: Fri Aug 08, 2025 7:33 pm Image
https://github.com/Lacovipo/Anubis-chess-engine
Anubis 3.01 by José Carlos Martínez Galán

Anubis 3.01 64 JA

Windows avx512 & bmi2 builds

Proton link:
https://drive.proton.me/urls/P33ZCY141C#GsyiX7Q68qoZ

Smash link:
https://fromsmash.com/Anubis301JA

Jim.
I've ported the msvc src code to gcc & clang and made all new profiled builds including sse4 & sse3.

Windows & Linux profiled builds

Proton link:
https://drive.proton.me/urls/6YJ9YPS6TR#2smy96g9ETIj

Smash link:
https://fromsmash.com/ANUBIS30164JA

Jim.
Big big thanks, Jim. I'll have a look and try to learn something. :)
__________________________
José Carlos Martínez Galán
(Averno & Anubis)
User avatar
Jim Ablett
Posts: 2214
Joined: Fri Jul 14, 2006 7:56 am
Location: London, England
Full name: Jim Ablett

Re: New J.A Compiles

Post by Jim Ablett »

Jose Carlos wrote: Sun Aug 10, 2025 8:03 pm
Jim Ablett wrote: Sun Aug 10, 2025 1:11 am
Jim Ablett wrote: Fri Aug 08, 2025 7:33 pm Image
https://github.com/Lacovipo/Anubis-chess-engine
Anubis 3.01 by José Carlos Martínez Galán

Anubis 3.01 64 JA

Windows avx512 & bmi2 builds

Proton link:
https://drive.proton.me/urls/P33ZCY141C#GsyiX7Q68qoZ

Smash link:
https://fromsmash.com/Anubis301JA

Jim.
I've ported the msvc src code to gcc & clang and made all new profiled builds including sse4 & sse3.

Windows & Linux profiled builds

Proton link:
https://drive.proton.me/urls/6YJ9YPS6TR#2smy96g9ETIj

Smash link:
https://fromsmash.com/ANUBIS30164JA

Jim.
Big big thanks, Jim. I'll have a look and try to learn something. :)
I updated the builds to use embedded net.

Proton link the same.

New Smash link:
https://fromsmash.com/Anubis301JAEmbed

Jim.