My list updated

[Sylwy]

Only engines I tested. There are around 3-4 untested.

[Carlos777]

Thanks for sharing this, Sylwy!

Edit: I think Nemorino 6 and Fire use SF-NNUE, Nemorino 6.05 uses his own net

[Sylwy]

Edit: I think Nemorino 6 and Fire use SF-NNUE, Nemorino 6.05 uses his own net

Yes, of course ! In general, my list shows the situation from the first appearance. Of course it is perfectible.

[jdart]

Not quite sure what "own NNUE" means exactly. Arasan uses the Stockfish 13 network topology, its own reading code, the nodchip tuner, its own training data, and its own net.

[Sylwy]

Not quite sure what "own NNUE" means exactly. Arasan uses the Stockfish 13 network topology, its own reading code, the nodchip tuner, its own training data, and its own net.

In general its own training data. But things are evolving. Even SF uses Lc0 databases for the networks........

[Guenther]

Your Coiled info is wrong.

[Sylwy]

Your Coiled info is wrong.

???

From version 0.8 Coiled uses red_neuronal.nnue. The author's answer to this problem is below. Until proven otherwise, I give him credit.

[RubiChess]

If you would have read just two entries further you would have found this: forum3/viewtopic.php?f=2&t=76209&start=670#p907028

[Sylwy]

If you would have read just two entries further you would have found this: forum3/viewtopic.php?f=2&t=76209&start=670#p907028

If you had read only five entries above you would have found that my list is perfectible. Like anything in this world!