Winter is a chess engine that relies on the UCI protocol to communicate with a GUI.
Contributors: Jonathan Rosenthal, Jonathan Maurer, Jonas Kuratli.
Rating JCER=3142
v.1.0:
First major Winter release in over two years! In regular chess it is expected to be roughly on paar with the previous release v0.9. It is expected to be stronger than Winter 0.9.5 at Fischer random chess, which was tested at CCRL. Fischer random is a newly supported variant and was not supported in v0.9.0.
Huge thank you to Seer author Connor McMonigle, who generated the Windows Winter compiles for this version. Without this, there is a good chance I would procrastinate the release for another couple of weeks!
Also thank you to Andrew Grant and the contributors to OpenBench, without whom it would not be possible to test changes to Winter with any statistical significance.
Major changes:
Chess960 support.
Evaluation function revision.
Eval is now based on small net on top of raw board state as opposed to handcrafted features.
Net weights are handled with incbin.h. This is standard amongst strong C and C++ AlphaBeta engines. There is no cross compatibility with other engines at this time.
Training code is rewritten from scratch and will likely get released at a later time.
Training data consists of some 350k CCRL Fischer random games. Additionally there are another 300k or so self play games, mostly DFRC, with a small number of regular FRC games being the exception. As only a small number of games are (coincidentally) from the regular start position, Winter cannot overfit to learn regular chess openings.
Android support
Hasn't been tested in a while, may need to fix.
Minor changes:
Removing cutnode/allnode expectations
Depth for singular extension may now be lower as long depth of entry is large enough
Program exit via uci quit command should be more stable
QS Move ordering changes
Code simplifications and removing a lot of legacy code.
History Heuristic tweaks
UCI code handling improvements
Winter 1.0 - download
Individual statistics: Winter 0.8 - 148 games (+ 70,= 34,- 44), 58.8 %
Schooner 2.2 : 2 (+ 0,= 0,- 2), 0.0 %
Arasan 22.0 : 4 (+ 0,= 2,- 2), 25.0 %
Vajolet2 2.8 : 2 (+ 1,= 0,- 1), 50.0 %
BlackMamba 2.0 : 2 (+ 0,= 2,- 0), 50.0 %
Ice4 853 : 2 (+ 2,= 0,- 0), 100.0 %
SmarThink 1.98 : 2 (+ 0,= 0,- 2), 0.0 %
Sctr 1.1f : 2 (+ 0,= 2,- 0), 50.0 %
Protector 1.9.0 : 2 (+ 0,= 2,- 0), 50.0 %
Equinox 3.30 : 2 (+ 1,= 1,- 0), 75.0 %
Deuterium 2019.2.37.73 : 2 (+ 1,= 1,- 0), 75.0 %
Ippos 1.27 : 4 (+ 1,= 2,- 1), 50.0 %
Rodent IV : 2 (+ 0,= 2,- 0), 50.0 %
Fritz 17 : 10 (+ 0,= 2,- 8), 10.0 %
Fat Fritz (in lc0) cpu : 4 (+ 2,= 0,- 2), 50.0 %
Xiphos 0.6.1 : 2 (+ 0,= 0,- 2), 0.0 %
Booot 6.4 : 2 (+ 1,= 0,- 1), 50.0 %
ComStock 3 : 2 (+ 0,= 2,- 0), 50.0 %
Monolith 2 : 2 (+ 1,= 0,- 1), 50.0 %
Gogobello 2.2 : 2 (+ 0,= 1,- 1), 25.0 %
FabChess 1.15 : 2 (+ 1,= 0,- 1), 50.0 %
Marvin 3.6.0 : 2 (+ 1,= 0,- 1), 50.0 %
ProDeo 2.92 : 8 (+ 7,= 1,- 0), 93.8 %
Bagatur 2.2 : 2 (+ 2,= 0,- 0), 100.0 %
Eubos 1.0.8 : 2 (+ 2,= 0,- 0), 100.0 %
Weiss 1.0 : 10 (+ 7,= 3,- 0), 85.0 %
Devel 3.2.9 : 6 (+ 6,= 0,- 0), 100.0 %
Francesca 0.29 : 2 (+ 0,= 2,- 0), 50.0 %
Sapeli 1.90 : 6 (+ 6,= 0,- 0), 100.0 %
Knightx 2.1b : 4 (+ 2,= 2,- 0), 75.0 %
Sargon 1978 1.00 : 6 (+ 6,= 0,- 0), 100.0 %
MagiC 1.0 : 6 (+ 6,= 0,- 0), 100.0 %
Rocket 14 : 6 (+ 0,= 1,- 5), 8.3 %
RubiChess 1.7.3 : 6 (+ 2,= 0,- 4), 33.3 %
Vafra 16.0 : 2 (+ 2,= 0,- 0), 100.0 %
Minic 2.35 : 4 (+ 0,= 2,- 2), 25.0 %
Stash 17.0 : 4 (+ 4,= 0,- 0), 100.0 %
NoC 4.20 : 2 (+ 2,= 0,- 0), 100.0 %
Raubfisch X42d3 : 2 (+ 0,= 1,- 1), 25.0 %
Wasp 4.0 : 2 (+ 0,= 1,- 1), 25.0 %
Igel 2.5.0 : 2 (+ 0,= 1,- 1), 25.0 %
Raubfisch GTZ21e : 2 (+ 0,= 0,- 2), 0.0 %
Brainfish 120620 : 2 (+ 0,= 0,- 2), 0.0 %
Ceibo 0.5 : 2 (+ 2,= 0,- 0), 100.0 %
FracTal 1.0 : 2 (+ 2,= 0,- 0), 100.0 %
Slowchess 2.2 : 2 (+ 0,= 1,- 1), 25.0 %
Comments
Post a Comment