Stockfish NN release (NNUE)

[JohnS]

If the binary runs on your machine, and the GUI is UCI compliant (i.e issues 'ucinewgame')
there's no logical explanation for it to not work.
...
May I ask what happens when you double click on the executable?

Hi kranium

The binary runs fine. Bench test passes. Most likely the problem is really in:

Some on sf-nnue discord have suggested Chessbase/Fritz are not 100% UCI compatible...

Ok in command mode, crashes in Chessbase 15 if I try to use it for analysis - causes exception.

Hi John, Roman-
sf-nnue run (halfkp-256x2-32-32) x64-popc.exe

Per discussion on sf-nnue discord,
it's seems likely that Fritz/Chessbase can't handle long file names...
you might try shortening it

Still doesnt work in Chessbase if shorten name. Any chance of changing name in source before compile- maybe sf-nnue-date.

[kranium]

Still doesnt work in Chessbase if shorten name. Any chance of changing name in source before compile- maybe sf-nnue-date.

SF NNUE + NET - working under Fritz
https://workupload.com/file/ggEUrvNVgmH

I have not tested this John...
someone in sf-nnue discord posted it

I believe he made the UCI name shorter in order to get it to work...

[JohnS]

Still doesnt work in Chessbase if shorten name. Any chance of changing name in source before compile- maybe sf-nnue-date.

SF NNUE + NET - working under Fritz
https://workupload.com/file/ggEUrvNVgmH

I have not tested this John...
someone in sf-nnue discord posted it

I believe he made the UCI name shorter in order to get it to work...

Many thanks kranium - works well in Chessbase now.

[Rom77]

SF NNUE + NET - working under Fritz
https://workupload.com/file/ggEUrvNVgmH

I have not tested this John...
someone in sf-nnue discord posted it

Wow, how was he able to get such a high speed? On SF NN popcnt, the speed is 43% of SF dev! A little more and overtake bmi2 . Maybe it can be compiled even faster?

[Rom77]

POPCNT compilations speed test

CPU: Intel i5 760 (Nehalem), 2.95 GHz
Command: bench 128 1 20
Net: Gek 27-06-2020

nodchip
===========================
Total time (ms) : 109051
Nodes searched : 53875054
Nodes/second : 494035

FireFather
===========================
Total time (ms) : 121672
Nodes searched : 53875054
Nodes/second : 442789

joergoster
===========================
Total time (ms) : 110207
Nodes searched : 53875054
Nodes/second : 488853

pleomati
===========================
Total time (ms) : 93655
Nodes searched : 53875054
Nodes/second : 575250

SF dev
===========================
Total time (ms) : 46159
Nodes searched : 60669708
Nodes/second : 1314363

[amchess]

I created a fork from Joergoster repository and updated to the latest stockfish patch:

https://github.com/amchess/Stockfish-NNUE/

Let me know if it's all good.
Andrea

[peter]

I created a fork from Joergoster repository and updated to the latest stockfish patch:

https://github.com/amchess/Stockfish-NNUE/

How about a BrainLearn- or ShashChess- NNUE, Andrea?

Difficult to add learning- code to SF NNUE?

[amchess]

Yes.
BrainLearn=search with the learning system and the rest Stockfish
StockfishNNUE=eval NN and the rest Stockfish
So, we can surely think about a
BrainLearn NNUE = search with the learning system + eval NN + the rest Stockfish

[peter]

Yes.
BrainLearn=search with the learning system and the rest Stockfish
StockfishNNUE=eval NN and the rest Stockfish
So, we can surely think about a
BrainLearn NNUE = search with the learning system + eval NN + the rest Stockfish

Sounds great.
Guess Pleonati at discord is working on something similiar, if I got him right.
https://discord.com/channels/4359437104 ... 4093398056

[carldaman]

@amchess,
I second that - it would be totally awesome!