Winter Released

[jorose]

While it did originate as a university project, it had long ceased to be by the time I added the mixture model and EM algorithms. That being said, here are some links you might find useful.

While implementing the EM algorithm I myself relied on these notes from the University of California Irvine.

I recommend first reading into the K-means algorithm, if you are unaware of that, before getting into that. This link from Stanford explains the K-means algorithm. I am not content with how Wikipedia describes GMMs as I feel the lacking visual representation would make it much easier. Instead I recommend reading some of the slides from this set of slides from an MIT lecture. Note that I only briefly skimmed both of these links now as I learned about these algorithms well before working on Winter and I don't think the slides presented in those lectures are actually publicly available.

[jorose]

I added a Windows version today. It would be cool if some of you could confirm it is indeed working.

There are some caveats under windows unfortunately:

In my brief testing I was getting at least 50% more NPS on Linux compared to Windows. A bit of a performance penalty was expected, but losing that much feels extreme. Perhaps someone else has a suggestion?

The Windows version is hilariously huge, just comparing the zip file for OSX (138 KB) with the Windows one (4.8 MB) it's fairly extreme. I imagine this is due to compiling with -static, but I am not sure how to circumvent this. I suppose this point and the previous one are linked.

It seems unicode characters are apparently still too state of the art for out of the box Windows support, so unfortunately my print commands are quite useless. I suspect Windows users won't spend much time in a console however, so I don't intend to invest any energy on this in the near future.

Finally a point I only just realized was that my perft_test command doesn't work on my Windows machine. I am not certain why this doesn't work, but as long as everything else works I am fine with that, as it is mostly just a feature for people compiling the engine themselves.

[Guenther]

I added a Windows version today. It would be cool if some of you could confirm it is indeed working.

There are some caveats under windows unfortunately:

In my brief testing I was getting at least 50% more NPS on Linux compared to Windows. A bit of a performance penalty was expected, but losing that much feels extreme. Perhaps someone else has a suggestion?

The Windows version is hilariously huge, just comparing the zip file for OSX (138 KB) with the Windows one (4.8 MB) it's fairly extreme. I imagine this is due to compiling with -static, but I am not sure how to circumvent this. I suppose this point and the previous one are linked.

It seems unicode characters are apparently still too state of the art for out of the box Windows support, so unfortunately my print commands are quite useless. I suspect Windows users won't spend much time in a console however, so I don't intend to invest any energy on this in the near future.

Finally a point I only just realized was that my perft_test command doesn't work on my Windows machine. I am not certain why this doesn't work, but as long as everything else works I am fine with that, as it is mostly just a feature for people compiling the engine themselves.

Well some people still use cmd sometimes on Win ;-)
Here is the raw output from initial position with default settings for 'go depth 15' on my old quadcore with Win7-64 Ultimate:
(the result in this run equals ~ 378.76 kn/s)

Code: Select all

uci
id name Winter Beta 0.1
id author Jonathan Rosenthal
option name Hash type spin default 32 min 1 max 104576
uciok
go depth 15
info  depth 1 seldepth 1 time 0 nodes 21 score cp 23 pv e2e3
info  depth 2 seldepth 3 time 3 nodes 162 score cp 4 pv d2d4 d7d5
info  depth 3 seldepth 7 time 6 nodes 454 score cp 19 pv d2d4 d7d5 g1f3
info  depth 4 seldepth 10 time 12 nodes 1576 score cp 4 pv d2d4 d7d5 g1f3 g8f6
info  depth 5 seldepth 12 time 22 nodes 3679 score cp 14 pv d2d4 d7d5 g1f3 g8f6
b1c3
info  depth 6 seldepth 13 time 44 nodes 10271 score cp 4 pv d2d4 d7d5 g1f3 g8f6
b1c3 b8c6
info  depth 7 seldepth 19 time 138 nodes 43942 score cp 10 pv d2d4 d7d5 b1c3 g8f
6 c1f4 g7g5 f4g5
info  depth 8 seldepth 19 time 263 nodes 89198 score cp 4 pv d2d4 d7d5 b1c3 g8f6
 g1f3 b8c6 c1f4 c8f5
info  depth 9 seldepth 21 time 463 nodes 163684 score cp 15 pv d2d4 d7d5 b1c3 g8
f6 g1f3 b8c6 c1f4 c8f5 e2e3
info  depth 10 seldepth 22 time 1028 nodes 378670 score cp 6 pv d2d4 d7d5 b1c3 b
8c6 e2e4 g8f6 f2f3 g7g6 e4e5 f6h5
info  depth 11 seldepth 24 time 1978 nodes 743982 score cp 14 pv d2d4 d7d5 b1c3
g8f6 c1f4 b8c6 e2e3 a7a6 f1d3 g7g6 g1f3
info  depth 12 seldepth 24 time 4027 nodes 1535005 score cp 6 pv d2d4 d7d5 b1c3
g8f6 c1f4 c7c5 g1f3 c5d4 f3d4 b8c6 h2h3 d8a5
info  depth 13 seldepth 27 time 7910 nodes 3019184 score cp 13 pv d2d4 d7d5 b1c3
 g8f6 c1f4 c7c5 g1f3 b8c6 e2e3 c5c4 f1e2 d8b6 d1c1
info  depth 14 seldepth 32 time 23677 nodes 8951954 score cp 7 pv d2d4 d7d5 c2c4
 d5c4 e2e4 c7c6 f1c4 g8f6 c4d3 e7e5 d4e5 f6g4 b1c3 g4e5
info  depth 15 seldepth 34 time 60326 nodes 22849233 score cp 13 pv d2d4 d7d5 c2
c4 d5c4 g1f3 g8f6 b1c3 c7c6 c1f4 b7b5 a2a3 g7g6 e2e3 h7h5 g2g3
bestmove d2d4

BTW as you can see it still calls itself 0.1 beta and not 0.11 (beta).
Thanks for the 'old' cpu release!

Guenther

[BrendanJNorman]

I added a Windows version today. It would be cool if some of you could confirm it is indeed working.

There are some caveats under windows unfortunately:

In my brief testing I was getting at least 50% more NPS on Linux compared to Windows. A bit of a performance penalty was expected, but losing that much feels extreme. Perhaps someone else has a suggestion?

The Windows version is hilariously huge, just comparing the zip file for OSX (138 KB) with the Windows one (4.8 MB) it's fairly extreme. I imagine this is due to compiling with -static, but I am not sure how to circumvent this. I suppose this point and the previous one are linked.

It seems unicode characters are apparently still too state of the art for out of the box Windows support, so unfortunately my print commands are quite useless. I suspect Windows users won't spend much time in a console however, so I don't intend to invest any energy on this in the near future.

Finally a point I only just realized was that my perft_test command doesn't work on my Windows machine. I am not certain why this doesn't work, but as long as everything else works I am fine with that, as it is mostly just a feature for people compiling the engine themselves.

I like to test new engines against weak opponents to gauge their "style" (the manner in which they smash their weak opponent).

Winter is really unique in its playing style, and seems to be quite a slow searcher.

This game is very interesting. g3 and then Be2 had me a bit confused, but apart from that, the playing style seemed pretty human-like!

[pgn][Event "Computer chess game"]
[Site "BRENDANNORMD8A2"]
[Date "2018.01.23"]
[Round "?"]
[White "Winter"]
[Black "Clueless 1.4"]
[Result "*"]
[BlackElo "2200"]
[ECO "A46"]
[Opening "Torre Attack"]
[Time "22:17:56"]
[Variation "3...Be7"]
[WhiteElo "2000"]
[TimeControl "180+2"]
[Termination "unterminated"]
[PlyCount "102"]
[WhiteType "program"]
[BlackType "program"]

1. Nf3 Nf6 2. d4 e6 3. Bg5 Be7 {(Bf8-e7 e2-e3 h7-h6 Bg5xf6 Be7xf6 Bf1-d3
c7-c5 O-O Nb8-c6 c2-c3) +0.18/10 5} 4. Nc3 {(Nb1-c3 O-O e2-e3 d7-d6 Bf1-d3
Bc8-d7 O-O Nb8-c6 Nf3-d2 Nc6-b4 Bg5xf6 Be7xf6) +0.17/12 6} Nc6 {(Nb8-c6
e2-e3) +0.14/9 5} 5. e3 {(e2-e3 h7-h6 Bg5xf6 Be7xf6 Bf1-d3 O-O Ke1-f1 d7-d5
Kf1-g1 Bc8-d7 a2-a3 Bf6-h4) +0.06/12 6} h6 {(h7-h6 Bg5xf6 Be7xf6 Bf1-b5
a7-a6 Bb5xc6 d7xc6 O-O O-O Nc3-e4 Bf6-e7) +0.09/10 5} 6. Bxf6 {(Bg5xf6
Be7xf6 Bf1-d3 d7-d5 O-O Bc8-d7 g2-g3 O-O a2-a3 a7-a6 Kg1-g2 g7-g6 Rf1-h1)
+0.10/13 6} Bxf6 {(Be7xf6 Bf1-b5 Bf6-e7 O-O O-O) +0.05/11 5} 7. g3 {(g2-g3
O-O Bf1-e2 d7-d5 O-O Bc8-d7 Kg1-g2 a7-a6 Qd1-d2 g7-g5 h2-h3 g5-g4 h3xg4)
+0.11/13 6} O-O {(O-O Bf1-b5 d7-d5 e3-e4 Nc6-b4 e4-e5 Bf6-e7 O-O c7-c5
Nc3-e2) +0.47/10 5} 8. Be2 {(Bf1-e2 Rf8-e8 Ke1-f1 e6-e5 Kf1-g2 e5-e4 Nf3-d2
d7-d5 Be2-b5 Bc8-d7 Qd1-h5 a7-a6 Bb5xc6) +0.10/13 6} d5 {(d7-d5 O-O a7-a5
Rf1-e1) +0.50/10 5} 9. O-O {(O-O Qd8-d6 Qd1-d3 Rf8-d8 a2-a3 a7-a6 e3-e4
Nc6xd4 Nf3xd4 d5xe4 Nc3xe4 Qd6xd4) +0.06/12 6} b6 {(b7-b6 Qd1-d2 a7-a6
Rf1-d1 b6-b5 a2-a3) +0.53/9 5} 10. Kg2 {(Kg1-g2 Bc8-b7 a2-a3 e6-e5 Rf1-g1
e5-e4 Nf3-d2 Qd8-d6 Be2-b5 a7-a6 Bb5-e2) +0.16/11 5} Bb7 {(Bc8-b7 Kg2-g1
e6-e5 b2-b3 e5-e4 Nf3-h4 Nc6-e7) +0.70/9 5} 11. a3 {(a2-a3 Qd8-d7 Qd1-d2
e6-e5 d4xe5 Nc6xe5 Nf3xe5 Bf6xe5 f2-f4 d5-d4+ Be2-f3) +0.10/11 5} e5
{(e6-e5 Kg2-g1 e5-e4 Nf3-e1 Bb7-c8 Ne1-g2) +0.75/9 5} 12. Bb5 {(Be2-b5
e5-e4 Nf3-d2 a7-a6 Bb5xc6 Bb7xc6 f2-f3 Bf6-g5 Qd1-e2 f7-f5 f3xe4 d5xe4)
+0.07/12 5} exd4 {(e5xd4 e3xd4 a7-a6 Bb5-e2) +0.84/9 5} 13. exd4 {(e3xd4
Rf8-e8 Qd1-d3 a7-a6 Bb5-a4 Qd8-d7 Rf1-e1 a6-a5 Ba4-b5 g7-g6 Qd3-d2 Kg8-h7
Nf3-e5) +0.10/13 5} Re8 {(Rf8-e8 Kg2-g1 Re8-e7 Nc3-e2 a7-a6 Bb5-d3) +0.80/9
4} 14. Qd3 {(Qd1-d3 a7-a6 Bb5-a4 Qd8-d7 Rf1-e1 a6-a5 Ba4-b5 g7-g6 Qd3-d2
Kg8-g7 h2-h3 g6-g5 Nf3-e5) +0.07/13 5} Qd6 {(Qd8-d6 Rf1-e1 a7-a6 Bb5xc6
Qd6xc6 Kg2-g1) +0.90/9 4} 15. Rfe1 {(Rf1-e1 a7-a6 Bb5xc6 Bb7xc6 Re1xe8+
Ra8xe8 Qd3xa6 Bc6-d7 Qa6-d3 c7-c6 h2-h3) +0.01/11 5} Re6 {(Re8-e6 Re1xe6
f7xe6 Kg2-h1 a7-a6 Bb5-a4 e6-e5 Ba4xc6) +0.61/8 4} 16. Rxe6 {(Re1xe6 Qd6xe6
Ra1-e1 Qe6-d6 Qd3-e3 Nc6-a5 Bb5-d3 Na5-c4 Qe3-e8+ Qd6-f8 Qe8-d7 Nc4xb2
Qd7xc7) +0.34/13 5} fxe6 {(f7xe6 Bb5-a4 a7-a6) +0.59/8 4} 17. Re1 {(Ra1-e1
Nc6-e7 Bb5-a6 Bb7xa6 Qd3xa6 Ne7-f5 Nc3-b5 Qd6-d7 g3-g4 Nf5-d6 Nb5xd6 c7xd6
c2-c3) +0.46/13 5} a6 {(a7-a6 Bb5xc6 Bb7xc6 h2-h3) +0.51/9 4} 18. Bxc6
{(Bb5xc6 Bb7xc6 Nf3-e5 Bc6-e8 Ne5-g4 Bf6-e7 f2-f4 b6-b5 Qd3-e2 Be8-g6
Qe2xe6+ Qd6xe6 Re1xe6 Kg8-f7) +0.46/14 5} Bxc6 {(Bb7xc6 Kg2-h1 Bc6-e8)
+0.59/10 4} 19. Ne5 {(Nf3-e5 Bc6-e8 Ne5-g4 Bf6-e7 f2-f4 h6-h5 Ng4-e5 Be7-f6
Re1-f1 b6-b5 f4-f5 Be8-d7 f5xe6 Qd6xe6) +0.41/14 5} Be8 {(Bc6-e8 Qd3-e3
Be8-h5) +0.65/9 4} 20. Ng4 {(Ne5-g4 Bf6-e7 f2-f4 h6-h5 Ng4-e5 Be7-f6 Re1-f1
b6-b5 b2-b4 Be8-f7 h2-h4 Bf7-e8 Rf1-e1 Be8-f7) +0.40/14 5} Be7 {(Bf6-e7
Qd3-e2 Be8-d7 Ng4-e5 Bd7-c8 Kg2-h1 c7-c5 Ne5-g6) +0.61/8 4} 21. Re3
{(Re1-e3 c7-c5 d4xc5 b6xc5 Nc3xd5 Be7-f8 Nd5-f4 Qd6xd3 c2xd3 Be8-c6+ f2-f3
Ra8-b8) +0.45/12 4} c5 {(c7-c5 d4xc5 b6xc5 Qd3-e2 Be8-d7 Kg2-h1 Ra8-b8
Ng4-e5 Rb8xb2 Ne5xd7) +1.00/9 4} 22. dxc5 {(d4xc5 b6xc5 Nc3xd5 Be8-c6
Ng4xh6+ g7xh6 Qd3-g6+ Kg8-h8 Qg6xh6+ Kh8-g8 Qh6-g6+ Kg8-h8 Qg6-h5+ Kh8-g7)
+0.31/14 4} bxc5 {(b6xc5 Qd3-e2 Be8-d7 Kg2-h1 Ra8-b8 Ng4-e5 Rb8xb2 Ne5xd7
Qd6xd7 Re3xe6) +1.00/9 4} 23. Nxd5 {(Nc3xd5 Be8-c6 Ng4xh6+ g7xh6 Qd3-g6+
Kg8-h8 Qg6xh6+ Kh8-g8 Qh6-g6+ Kg8-h8 Qg6-h5+ Kh8-g7 Qh5-g4+) +0.31/13 4}
exd5 {(e6xd5 Re3xe7 Be8-g6 Qd3-e2 Bg6xc2 Re7-e5 Bc2-b1 Kg2-h1 Ra8-b8)
+0.34/9 4} 24. Rxe7 {(Re3xe7 Qd6xe7 Qd3xd5+ Kg8-h7 Qd5xa8 Qe7-e6 Kg2-g1
Qe6-e1+ Kg1-g2 Qe1-e6 Kg2-g1 Qe6-e1+ Kg1-g2 Qe1-e6) 0.00/14 4} Bg6 {(Be8-g6
Re7-d7 Qd6xd7 Qd3xg6 Kg8-f8 Ng4-e5 Qd7-e7 Qg6-f5+ Kf8-e8 f2-f4) +0.49/9 4}
25. Qe3 {(Qd3-e3 Bg6-e4+ Re7xe4 d5xe4 Qe3xe4 Ra8-f8 Ng4-e5 Rf8-e8 f2-f4
Qd6-d2+ Kg2-h3 Re8-e6 f4-f5) +0.56/13 4} Bxc2 {(Bg6xc2 Re7-e6 Qd6-d7 f2-f3)
+0.65/9 3} 26. Re6 {(Re7-e6 Qd6-f8 f2-f3 d5-d4 Qe3-d2 Bc2-f5 Re6-f6 g7xf6
Ng4xh6+ Kg8-h7 Nh6xf5 Ra8-b8 g3-g4 d4-d3) +0.37/14 4} Qf8 {(Qd6-f8 Qe3-e5
Qf8-f5 f2-f3 d5-d4) +0.71/9 3} 27. f3 {(f2-f3 d5-d4 Qe3-d2 Bc2-f5 Re6-f6
g7xf6 Ng4xh6+ Kg8-h7 Nh6xf5 Ra8-b8 g3-g4 d4-d3 Qd2xd3) +0.37/13 4} Bf5
{(Bc2-f5 Re6-c6 Ra8-c8 Rc6xa6 Rc8-b8 Qe3-e2) +1.06/9 3} 28. Rf6 {(Re6-f6
g7xf6 Ng4xh6+ Kg8-h8 Nh6xf5 Ra8-b8 Qe3-d2 d5-d4 g3-g4 d4-d3 Qd2xd3 Rb8xb2+
Kg2-g3 Qf8-a8) +0.29/14 4} gxf6 {(g7xf6 Ng4xh6+ Kg8-h7 Nh6xf5 Ra8-e8 Qe3-d3
c5-c4 Qd3-c2) +1.89/12 3} 29. Nxh6+ {(Ng4xh6+ Kg8-h8 Nh6xf5 Ra8-e8 Qe3-d2
Re8-e5 g3-g4 d5-d4 b2-b4 Re5xf5 g4xf5 Kh8-g7 a3-a4 Qf8-d6 b4xc5 Qd6xc5)
+0.29/16 4} Kh7 {(Kg8-h7 Nh6xf5 Ra8-b8 Qe3-d2 d5-d4 h2-h4) +1.86/11 3} 30.
Nxf5 {(Nh6xf5 Ra8-b8 Qe3-d2 d5-d4 g3-g4 Rb8-b7 h2-h4 Qf8-d8 h4-h5 d4-d3
Qd2-h6+ Kh7-g8 Qh6-g6+ Kg8-f8) +0.32/14 4} Rb8 {(Ra8-b8) +1.77/10 3} 31.
Qd2 {(Qe3-d2 d5-d4 b2-b4 c5xb4 a3xb4 Rb8xb4 Nf5xd4 Qf8-c5 Nd4-f5 Qc5xf5
Qd2xb4 Qf5-c2+ Kg2-h3 Kh7-g6) +0.27/14 4} a5 {(a6-a5 g3-g4 d5-d4 h2-h4
a5-a4 h4-h5 Kh7-h8 h5-h6 Rb8-b3 Qd2-c1 d4-d3 Kg2-g3 Qf8-b8+) +1.85/10 3}
32. g4 {(g3-g4 d5-d4 h2-h4 Qf8-f7 h4-h5 Rb8-b3 Qd2-c2 d4-d3 Qc2-d2 Rb3-b8
Kg2-h3 Rb8-b3 Qd2xa5 Rb3xb2) +0.29/14 4} d4 {(d5-d4 h2-h4 a5-a4 h4-h5
Rb8-b3 Kg2-f2 d4-d3 Qd2-c1) +1.87/10 3} 33. h4 {(h2-h4 Qf8-f7 h4-h5 Rb8-b3
Qd2-c2 d4-d3 Qc2-d2 c5-c4 Kg2-h3 Rb3-b6 Qd2-h6+ Kh7-g8 Qh6-f4 Qf7-e6)
+0.22/14 4} a4 {(a5-a4 h4-h5 Kh7-h8 Kg2-g1 Kh8-g8 Kg1-f2 Qf8-d8) +1.85/11
3} 34. Qc2 {(Qd2-c2 d4-d3 Qc2xd3 Rb8xb2+ Kg2-h3 Rb2-b3 Qd3-d5 Rb3xa3 h4-h5
Ra3-b3 h5-h6 c5-c4 Qd5xc4 Rb3xf3+) +0.03/14 4} Kh8 {(Kh7-h8 h4-h5 d4-d3
Qc2-c1 Rb8-b7 Kg2-f2 Qf8-e8 Qc1-c3 Qe8-e2+ Kf2-g3 d3-d2 Qc3xf6+ Kh8-h7)
+1.94/10 3} 35. h5 {(h4-h5 Rb8-b3 Qc2-c1 d4-d3 Qc1-d2 c5-c4 Qd2-c1 Rb3-b5
Qc1-c3 Rb5-e5 Qc3xc4 d3-d2 Qc4xa4 Re5-e1) +0.03/14 4} Rb7 {(Rb8-b7 Kg2-f2
Qf8-b8) +1.87/10 3} 36. Qd2 {(Qc2-d2 Rb7-b3 Qd2-c1 d4-d3 Qc1-d2 c5-c4
Qd2-c1 Kh8-h7 Kg2-h3 Kh7-h8 h5-h6 Qf8-a8 Kh3-g3 Qa8-d5) +0.03/14 4} Qd8
{(Qf8-d8 Kg2-g3 d4-d3 f3-f4 Rb7-b6) +1.89/9 3} 37. Qh6+ {(Qd2-h6+ Kh8-g8
Qh6-g6+ Kg8-h8 Qg6-h6+ Kh8-g8 Qh6-g6+ Kg8-h8 Qg6-h6+ Kh8-g8 Qh6-g6+ Kg8-h8
Qg6-h6+ Kh8-g8) 0.00/14 3} Kg8 {(Kh8-g8 Qh6-c1 Qd8-d5 Kg2-f2 d4-d3 Kf2-e3)
+1.98/11 3} 38. Qg6+ {(Qh6-g6+ Kg8-h8 Qg6-h6+ Kh8-g8 Qh6-g6+ Kg8-h8 Qg6-h6+
Kh8-g8 Qh6-g6+ Kg8-h8 Qg6-h6+ Kh8-g8 Qh6-g6+ Kg8-h8 Qg6-h6+) 0.00/15 3} Kf8
{(Kg8-f8 Qg6-h6+ Kf8-e8 Qh6-c1 Qd8-b6 Kg2-g3) +1.87/11 3} 39. h6 {(h5-h6
Qd8-c7 b2-b4 c5xb4 h6-h7 Qc7xh7 Qg6xf6+ Qh7-f7 Qf6-h8+ Qf7-g8 Qh8-h6+
Kf8-e8 Nf5-d6+ Ke8-d7 Nd6xb7) +1.50/15 3} Qc7 {(Qd8-c7 h6-h7 Rb7xb2+ Kg2-f1
Rb2-b1+ Kf1-g2) 0.00/10 3} 40. b4 {(b2-b4 c5xb4 h6-h7 Qc7xh7 Qg6xf6+ Qh7-f7
Qf6-h8+ Qf7-g8 Qh8-h6+ Kf8-e8 Nf5-d6+ Ke8-d7 Nd6xb7 Qg8-a2+ Kg2-g3)
+1.68/15 3} cxb4 {(c5xb4 Nf5xd4 Rb7-b6 a3xb4 Qc7-f4 Qg6-g7+ Kf8-e8 Qg7-g6+
Ke8-e7 Qg6-g7+) +0.77/8 3} 41. h7 {(h6-h7 Qc7xh7 Qg6xf6+ Qh7-f7 Qf6-h8+
Qf7-g8 Qh8-h6+ Rb7-g7 Qh6-d6+ Kf8-e8 Qd6-e5+ Ke8-d7 Qe5xd4+ Kd7-c8 Nf5xg7)
+2.54/15 3} Qxh7 {(Qc7xh7 Qg6xf6+ Qh7-f7 Qf6-d8+ Qf7-e8 Qd8xe8+ Kf8xe8
Nf5-d6+ Ke8-e7 Nd6xb7 b4xa3 Nb7-a5 d4-d3 Kg2-g3 Ke7-f7) -1.68/11 3} 42.
Qxf6+ {(Qg6xf6+ Qh7-f7 Qf6-h8+ Qf7-g8 Qh8-h6+ Kf8-f7 Nf5-d6+ Kf7-e7 Nd6xb7
Qg8-a2+ Kg2-g3 b4xa3 Qh6-d6+ Ke7-f7 Qd6xd4 Qa2-c2) +2.48/16 3} Qf7
{(Qh7-f7) -2.07/10 3} 43. Qh8+ {(Qf6-h8+ Qf7-g8 Qh8-h6+ Kf8-f7 Nf5-d6+
Kf7-e7 Nd6xb7 Qg8-a2+ Kg2-g3 b4xa3 Qh6-g7+ Qa2-f7 Qg7xd4 Qf7-e6 Qd4-c5+
Ke7-f6) +2.80/16 3} Qg8 {(Qf7-g8 Qh8-h6+ Kf8-f7 Nf5-d6+ Kf7-e7) -1.66/11 3}
44. Qh6+ {(Qh8-h6+ Kf8-f7 Nf5-d6+ Kf7-e7 Nd6xb7 Qg8-a2+ Kg2-g3 b4xa3
Qh6-g7+ Qa2-f7 Qg7xd4 Qf7-e6 Qd4-c5+ Ke7-f6 Qc5xa3 Qe6-e1+) +2.80/16 3} Kf7
{(Kf8-f7) -2.38/10 3} 45. Nd6+ {(Nf5-d6+ Kf7-e7 Nd6xb7 Qg8-a2+ Kg2-g3
Qa2-e6 Qh6-g7+ Ke7-e8 Qg7xd4 Qe6-e1+ Kg3-g2 Qe1-e2+ Kg2-g1 Qe2-e1+ Kg1-h2)
+2.76/15 3} Ke7 {(Kf7-e7 Nd6xb7 Qg8-a2+ Kg2-g3 Qa2xa3 Qh6-g7+ Ke7-e6
Nb7-c5+ Ke6-d5) -2.33/12 3} 46. Nxb7 {(Nd6xb7 Qg8-a2+ Kg2-g3 Qa2-e6 Qh6-g5+
Ke7-f8 Qg5-d8+ Kf8-f7 Nb7-d6+ Kf7-g6 a3xb4 Qe6-e1+ Kg3-g2 Qe1-e2+ Kg2-g3)
+2.80/15 3} Qa2+ {(Qg8-a2+) -2.33/10 2} 47. Kg3 {(Kg2-g3 b4xa3 Qh6-g5+
Ke7-d7 Qg5-g7+ Kd7-c6 Nb7-d8+ Kc6-d6 Qg7-f6+ Kd6-c5 Qf6-e7+ Kc5-b6 Qe7-d6+
Kb6-b5 Qd6xd4) +2.97/15 3} Qxa3 {(Qa2xa3 Qh6-g7+ Ke7-e8 Qg7xd4 b4-b3
Nb7-d6+ Ke8-f8 Nd6-c4 Qa3-a2 Qd4-b2 Qa2xb2 Nc4xb2) -2.97/10 2} 48. Qh7+
{(Qh6-h7+ Ke7-f6 Qh7-f5+ Kf6-g7 Nb7-d8 Qa3-e3 Nd8-e6+ Qe3xe6 Qf5xe6 a4-a3
Qe6-e7+ Kg7-g6 Qe7xb4 a3-a2 Qb4-d6+) +7.94/15 3} Kf6 {(Ke7-f6 Qh7-h8+
Kf6-g6 Qh8xd4 Qa3-c3 Qd4xc3 b4xc3 Nb7-c5 c3-c2 Nc5-d3 Kg6-f7) -3.39/11 2}
49. Qf5+ {(Qh7-f5+ Kf6-e7 Qf5-e5+ Ke7-f8 Nb7-d6 Qa3-d3 Nd6-f5 Qd3-e3 Nf5xe3
d4xe3 Qe5-c5+ Kf8-g7 Qc5xb4 a4-a3 Qb4-e7+ Kg7-g8) +8.45/16 3} Ke7 {(Kf6-e7
Qf5-e5+ Ke7-f8) -3.49/11 2} 50. Qe5+ {(Qf5-e5+ Ke7-f8 Nb7-d6 Qa3-d3 Nd6-f5
Qd3xf5 Qe5xf5+ Kf8-e7 Qf5-e4+ Ke7-f7 Qe4-d5+ Kf7-e8 Qd5xd4 a4-a3 Qd4xb4
a3-a2 Qb4-a4+) +9.70/17 3} Kd7 {(Ke7-d7 Qe5-d6+ Kd7-e8 Nb7-c5 Ke8-f7)
-4.14/11 2} 51. Qd6+ {(Qe5-d6+ Kd7-e8 Qd6-e6+ Ke8-f8 Nb7-d6 Qa3xf3+ Kg3xf3
Kf8-g7 Nd6-f5+ Kg7-h7 Qe6-d7+ Kh7-g8) +10.85/12 3} Ke8 {(Kd7-e8 Nb7-c5
Ke8-f7 Qd6-d5+ Kf7-g6) -4.14/11 1} *[/pgn]

[jorose]

That sounds about right. On Linux with nothing else running on my laptop with an i7-7600U (base frequency 2.8GHz iirc, but single threaded I assume it should be faster) I am getting 1.5 MNps. As I mentioned before I am getting at least 50% more nodes on linux, I imagine an old quadcore should have a lower frequency and finally for the "old cpu" compile the only architecture flag I added to the compile was -m64 so that is connected with a big penalty.

[jorose]

Be2 and g3 does indeed look a bit unorthodox to put it mildly I feel that often times Winter will go a bit overboard with its ideas, which results in it sometimes under-performing against weaker programs. For the longest time I thought it was rated around 2200, because I tested it against Cinnamon 2.0 and it was only performing on that level, then I tested it against stronger engines and the rating performance improved quite a bit.

I had never really thought of Winter's play as human, but looking at that game I can see what you mean =)

Today I let it play some games against Pawny 0.3.0 and before I started being productive again I let it play a minimatch with a time control of 20 minutes with a 5 second increment. So these two games show what it looks like against a relatively even opponent when it is not really time constrained.

[pgn][Event "Computer Chess Game"]
[Site "jonathan-ThinkPad-T470s"]
[Date "2018.01.25"]
[Round "1"]
[White "Winter Beta 0.1"]
[Black "Pawny 0.3 (x64)"]
[Result "1-0"]
[TimeControl "1200+5"]
[Annotator "1. +0.07 1... -0.12"]

1. d4 {+0.07/16} Nf6 {-0.12/17 50} 2. Nc3 {+0.11/16 34} e6 {-0.10/18 1:09}
3. e4 {+0.29/15 33} d5 {-0.14/18 1:13} 4. e5 {+0.35/16 33} Nfd7
{-0.05/18 1:07} 5. Qg4 {+0.41/15 32} c5 {+0.15/16 25} 6. Nb5 {+0.26/15 31}
cxd4 {+0.24/15 34} 7. Nf3 {+0.29/15 31} Nc6 {+0.19/16 48} 8. Bf4
{+0.23/15 30} f5 {+0.27/15 1:13} 9. Qg3 {+0.48/15 29} Bb4+ {-0.05/15 31}
10. c3 {+0.56/17 29} dxc3 {-0.38/17 45} 11. bxc3 {+0.64/17 28} Bf8
{-0.49/17 29} 12. Nd6+ {+0.95/18 28} Bxd6 {-0.39/17 20} 13. Qxg7
{+0.70/17 27} Rf8 {-0.44/18 46} 14. exd6 {+0.75/17 26} Qf6 {-0.41/18 46}
15. Qxf6 {+0.65/18 26} Rxf6 {-0.30/19 1:01} 16. Kd2 {+0.57/18 25} e5
{-0.18/19 23} 17. Re1 {+0.59/18 25} Kf7 {-0.36/19 38} 18. Nxe5+
{+0.69/19 24} Ncxe5 {-0.32/19 30} 19. Bxe5 {+0.59/19 24} Nxe5 {-0.35/19 18}
20. Rxe5 {+0.58/19 23} Be6 {-0.31/18 19} 21. Bd3 {+1.05/17 23} b6
{-0.81/17 1:07} 22. h3 {+1.19/16 23} f4 {-0.92/15 17} 23. Bxh7
{+1.23/16 22} Rd8 {-0.89/17 27} 24. h4 {+1.18/16 22} Rxd6 {-0.87/15 13} 25.
h5 {+1.14/16 21} Kg7 {-0.85/16 12} 26. Bd3 {+1.29/16 21} Kh6 {-0.78/17 28}
27. Rh4 {+1.32/17 20} a5 {-0.74/17 39} 28. g3 {+1.46/16 20} fxg3
{-0.67/17 14} 29. fxg3 {+1.17/18 20} Rf2+ {-0.69/17 14} 30. Be2
{+0.98/18 19} Rg2 {-0.64/17 13} 31. Re3 {+0.94/18 19} Bc8 {-0.17/17 9} 32.
Rd4 {+0.71/18 19} Ba6 {-0.35/17 11} 33. g4 {+0.79/19 18} Rf6 {-0.37/18 18}
34. Rxd5 {+0.55/18 18} Bxe2 {-0.39/19 17} 35. g5+ {+0.69/19 18} Rxg5
{-0.41/18 9} 36. Rxg5 {+0.70/20 17} Kxg5 {-0.51/19 8} 37. Re5+
{+0.70/20 17} Kf4 {-0.54/19 9} 38. Rxe2 {+0.70/18 17} Rh6 {-0.61/20 19} 39.
Rh2 {+0.21/18 16} Rd6+ {-0.68/19 15} 40. Kc2 {+0.79/16 16} Rh6
{-0.84/20 22} 41. c4 {+0.97/17 16} Kg3 {-0.84/20 9} 42. Rh1 {+0.97/18 16}
Kg2 {-0.97/20 11} 43. Rb1 {+0.97/18 15} Kf3 {-1.03/20 8} 44. Rb5
{+0.97/18 15} Ke4 {-1.09/19 7} 45. Kb3 {+1.09/17 15} Kf4 {-1.14/18 7} 46.
a4 {+1.20/19 15} Kg4 {-1.65/21 10} 47. c5 {+1.34/19 14} bxc5 {-1.85/21 16}
48. Rxc5 {+1.35/19 14} Rb6+ {-1.84/21 7} 49. Kc3 {+1.35/20 14} Ra6
{-1.98/22 15} 50. Kc4 {+1.73/19 14} Kf4 {-2.02/22 9} 51. Kb5 {+1.74/19 13}
Ra8 {-2.03/21 7} 52. h6 {+1.74/18 13} Rh8 {-2.04/21 12} 53. Rh5
{+2.20/19 13} Kg4 {-5.67/20 6} 54. Rh1 {+3.18/20 13} Rh7 {-5.84/20 5} 55.
Kxa5 {+4.99/21 13} Kf3 {-6.27/21 6} 56. Kb6 {+5.10/21 12} Kg2 {-7.80/21 6}
57. Rh5 {+6.46/21 12} Kg3 {-9.55/22 11} 58. a5 {+6.46/20 12} Kg4
{-9.57/21 6} 59. Rh1 {+9.23/21 12} Rh8 {-9.76/21 11} 60. h7 {+9.23/20 12}
Kg5 {-10.04/21 12} 61. Kb7 {+9.43/20 11} Kg6 {-10.46/22 6} 62. a6
{+9.48/21 11} Rxh7+ {-1000.10/19 23} 63. Rxh7 {+9.61/23 11} Kxh7
{-1000.08/20 6} 64. a7 {+1000.10/28 11} Kh6 {-1000.08/21 5} 65. Kc6
{+1000.09/29 11} Kg5 {-1000.07/19 5} 66. Kd5 {+1000.08/31 11} Kf4
{-1000.06/34 5} 67. a8=Q {+1000.07/33 11} Ke3 {-1000.05/50 4} 68. Qf8
{+1000.06/37 10} Kd3 {-1000.05/50 0.1} 69. Qf3+ {+1000.05/42 10} Kd2
{-1000.04/50 0.7} 70. Kc4 {+1000.04/46 10} Ke1 {-1000.03/50 0.1} 71. Qg2
{+1000.03/50 10} Kd1 {-1000.02/50 0.1} 72. Kd3 {+1000.02/54 10} Kc1
{-1000.01/50 0.1} 73. Qc2# {+1000.01/60 10}
{Xboard adjudication: Checkmate} 1-0

[Event "Computer Chess Game"]
[Site "jonathan-ThinkPad-T470s"]
[Date "2018.01.25"]
[Round "2"]
[White "Pawny 0.3 (x64)"]
[Black "Winter Beta 0.1"]
[Result "1/2-1/2"]
[TimeControl "1200+5"]
[Annotator "1. +0.17 1... -0.04"]

1. e4 {+0.17/17} d5 {-0.04/15 35} 2. exd5 {+0.21/17 34} e6 {-0.04/15 34} 3.
Bb5+ {+0.74/17 41} c6 {-0.04/16 33} 4. dxc6 {+0.83/17 27} Nxc6
{-0.07/16 33} 5. Nf3 {+0.83/17 1:01} Bd6 {-0.14/16 32} 6. Nc3 {+0.78/16 31}
Ne7 {-0.07/15 31} 7. O-O {+0.86/16 50} O-O {-0.09/15 31} 8. d4
{+0.86/15 36} a6 {-0.12/15 30} 9. Ba4 {+0.84/15 52} b5 {-0.05/15 29} 10.
Bb3 {+0.82/15 24} Bb7 {-0.07/15 29} 11. Re1 {+0.77/15 23} Nf5 {-0.16/14 28}
12. d5 {+1.12/15 24} Na5 {-0.18/15 28} 13. dxe6 {+1.27/16 39} Nxb3
{-0.28/16 27} 14. exf7+ {+1.24/15 21} Rxf7 {-0.29/16 26} 15. axb3
{+1.04/16 34} Rd7 {+0.00/15 26} 16. Qe2 {+1.25/15 25} Re7 {-0.21/15 25} 17.
Qf1 {+1.03/15 24} Bxf3 {+0.00/15 25} 18. Rxe7 {+0.96/16 53} Qxe7
{+0.00/16 24} 19. gxf3 {+0.92/16 21} Nd4 {+0.00/16 24} 20. Qd3
{+0.68/16 46} Qh4 {-0.11/16 23} 21. f4 {+0.83/16 40} Re8 {+0.00/16 23} 22.
h3 {+0.87/16 27} Bxf4 {+0.00/16 23} 23. Bxf4 {+0.22/16 39} Qxf4
{+0.00/18 22} 24. Rd1 {+0.02/16 26} Nf3+ {+0.00/18 22} 25. Kf1
{+0.00/16 42} Rf8 {+0.00/18 21} 26. Qd5+ {+0.00/17 18} Kh8 {+0.00/19 21}
27. Qc5 {+0.00/17 39} Nh2+ {+0.00/18 20} 28. Kg1 {+0.00/17 19} Nf3+
{+0.00/19 20} 29. Kf1 {+0.00/22 18} Nh2+ {+0.00/19 20} 30. Kg1
{+0.00/20 41} Nf3+ {+0.00/20 19} 31. Kf1 {+0.00/24 1:23}
{Draw by repetition} 1/2-1/2
[/pgn]