Has a significantly larger network (1536x4 -> 2048x16), a few bugfixes, and some QOL additions like pretty printing.
For the sake of my sanity (and because .NET can't/doesn't autovectorize NNUE code), this requires at least AVX2 to be performant, but should still function without it.
Picking a binary:
Start with the non-aot binary, and pick ...-512 if your processor has Avx512.
If you want a slightly slower but more compact binary, start with ...-aot-v4 and use the first one that doesn't crash:
- v4 enables Avx512,
- v3 enables Avx256,
- v2 is a fallback for older CPU's
STC:
Elo | 25.03 +- 7.17 (95%)
Conf | 8.0+0.08s Threads=1 Hash=32MB
Games | N: 2016 W: 328 L: 183 D: 1505
Penta | [6, 126, 609, 251, 16]
http://somelizard.pythonanywhere.com/test/1954/
LTC:
Elo | 16.69 +- 6.95 (95%)
Conf | 60.0+0.60s Threads=1 Hash=256MB
Games | N: 1000 W: 89 L: 41 D: 870
Penta | [0, 28, 398, 72, 2]
http://somelizard.pythonanywhere.com/test/1958/
DFRC:
Elo | 108.26 +- 12.12 (95%)
Conf | 8.0+0.08s Threads=1 Hash=32MB
Games | N: 2004 W: 873 L: 268 D: 863
Penta | [15, 116, 353, 285, 233]
http://somelizard.pythonanywhere.com/test/1955/