Gull 3 Linux+Syzygy and Fathom released.

[basil00]

No we have missed the boat for stage 2. Maybe stage 3, or the super final
I still do not have a suitable Windows machine for BMI builds.

[stuwph]

No we have missed the boat for stage 2. Maybe stage 3, or the super final
I still do not have a suitable Windows machine for BMI builds.

I will try to do a working BMI2 + PGO compile with mingw gcc version 5.4.0 (Rev1, Built by MSYS2 project)

would you like your last 2 Fathom patches "Fix en passant in do_move()" and Use Ronald's "slightly more efficient" fix for #10 to be integrated?

[basil00]

would you like your last 2 Fathom patches "Fix en passant in do_move()" and Use Ronald's "slightly more efficient" fix for #10 to be integrated?

Yes please.

[MikeB]

A new optimized build for Linux has been tested, see here:
https://github.com/basil00/Gull/commit/ ... ffa9ff10d8

Code: Select all

LTC (60"+0.6", 1-thread, 6-man syzygy):

    Score: +1728-1351=6921
    Number of games: 10000
    Winning fraction: 0.51885
    Elo difference: +13.1045
    LOS: 1

STC (10"+0.1", 1-thread, 6-man syzygy):

    Score: +2040-1791=6169
    Number of games: 10000
    Winning fraction: 0.51245
    Elo difference: +8.65293
    LOS: 0.999971

I will release binaries (Linux and Windows) as soon as I can find a suitable Windows machine to build. The intention is that this will be the "release candidate".

Added Mac exe here, with the "quit" fix.

https://github.com/MichaelB7/Gull

Code: Select all

Pseudo Bench
go depth 16
info depth 16 seldepth 37 score cp 22 nodes 14834252 nps 2032089 tbhits 0 pv e2e4 e7e6 d2d4 d7d5 e4d5 e6d5 g1f3 g8f6 f1d3 f8e7 b1c3 b8c6 e1g1 e8g8 f1e1 c8g4
info nodes 15530173 score cp 22
bestmove e2e4 ponder e7e6

[stuwph]

new bmi2 & pop compile aligned to Basil's repo
http://1drv.ms/1Qb2V48

did quite well in a local 5min blitz Noomen match against Fritz15 (which seemed to promise more fun than against SF or K10)
stopped it at +14 =17 -1

both 8cores / ponder off / 6-men / no book / each position with changed color

already the 1st game proved the tables working
http://www.viewchess.com/cbreader/2016/ ... 05671.html

Gull with black made a somewhat too optimistic exchange sac with 30..Rxc3, and after 62.b8Q I thought it was completely lost. Even worried that Gull's 0.00 eval must be wrong ...
but the tables did the job nicely and F15 was almost loosing on time in the end


despite it apparently works well, compiling requires to remove -fno-exceptions
otherwise it still runs in a "HSHMAX too low" exception while initializing 6-men ...

so these compiles are without -fno-exceptions, but wonder if this is correct?!
tried different mingws, actually used gcc version 5.4.0 (Rev1, Built by MSYS2 project)

[basil00]

new bmi2 & pop compile aligned to Basil's repo
http://1drv.ms/1Qb2V48

Thanks! The preliminary results are encouraging.

despite it apparently works well, compiling requires to remove -fno-exceptions
otherwise it still runs in a "HSHMAX too low" exception while initializing 6-men ...
so these compiles are without -fno-exceptions, but wonder if this is correct?!
tried different mingws, actually used gcc version 5.4.0 (Rev1, Built by MSYS2 project)

That definitely isn't right. The HSHMAX or material hash function should have no dependency on whether exceptions are enabled or not.

[basil00]

Added Mac exe here, with the "quit" fix.
https://github.com/MichaelB7/Gull

Thanks. I have been neglecting Mac since I no longer have a machine to build on.
Also thanks for pointing out the "quit" problem.

[Nordlandia]

i7-5960X 3.9GHz
6-Men Syzygy
Fritz 15 last patch
2048 MB allocated for each engine
Houdini settings: default
Gull settings: default
[pgn][Event "EXPERT-PC, Rapid 12m+12s"]
[Site "EXPERT-PC"]
[Date "2016.06.14"]
[Round "1"]
[White "Gull3 BF_sy"]
[Black "Houdini 4 Pro x64"]
[Result "1/2-1/2"]
[ECO "B95"]
[PlyCount "133"]

1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Bg5 e6 7. Qf3 Nbd7 8.
O-O-O h6 9. Bd2 Qc7 10. Qg3 b5 11. a3 Rb8 12. f3 Nc5 13. Qe1 Be7 14. g4 Nfd7
15. Kb1 Ne5 16. Bc1 Bd7 17. h4 b4 18. axb4 Rxb4 19. Rh3 Qb7 20. b3 O-O 21. Ba3
Ned3 22. Bxd3 Rxd4 23. Qe3 e5 24. Qe2 Be6 25. Bxc5 dxc5 26. Nd5 Bxd5 27. exd5
Qxd5 28. Rhh1 Rd8 29. c4 Qd6 30. Bc2 a5 31. Rxd4 cxd4 32. Qd3 Kf8 33. g5 hxg5
34. hxg5 g6 35. Rh8+ Kg7 36. Rxd8 Bxd8 37. Qd2 Be7 38. Bd3 Qd8 39. Qe2 Qc7 40.
Qc2 Qb7 41. Be4 Qb6 42. Bd5 Qb4 43. c5 Qe1+ 44. Kb2 Bxg5 45. Ka2 Qe3 46. Be4 f5
47. c6 Bd8 48. c7 Bxc7 49. Qxc7+ Kh6 50. Qxe5 Qd2+ 51. Kb1 fxe4 52. f4 Qd1+ 53.
Ka2 Qc2+ 54. Ka1 Qc3+ 55. Ka2 Kh7 56. Qe7+ Kh8 57. Qf8+ Kh7 58. Qe7+ Kg8 59.
Qxe4 Qd2+ 60. Kb1 Qd1+ 61. Kb2 Kh7 62. Qe7+ Kh6 63. Qg5+ Kg7 64. Qe5+ Kf7 65.
Qd5+ Kf8 66. Qd8+ Kg7 67. Qe7+ 1/2-1/2

[/pgn]