Evert wrote:I have now updated the release to SjaakII 1.4.1.
This fixes a bug that could corrupt the board by passing invalid moves in the PV array, as well as a number of other minor bugs. It also significantly changes piece values for Crazyhouse, and is is considerably stronger in self-play than 1.4.0.
It can be downloaded from
http://www.eglebbk.dds.nl/program/chess-download.html
Evert, is there a way to prevent very early repetitions?
I have done a little test match with the test version from yesterday against Pychess.
Of course the score was overwhelmingly in favour of SjaakII.
IIRC it was 78W 1D 1L. No illegal moves or other disturbancies in WB appeared.
Code: Select all
[Event "RWBC"]
[Site "CAPPUCCINO"]
[Date "2016.11.29"]
[Round "24"]
[White "PyChess_20161118"]
[Black "SjaakII_141-64"]
[Result "1/2-1/2"]
[TimeControl "40/180"]
[Variant "crazyhouse"]
[Number "24"]
[Annotator "3. -0.04 3... -0.05"]
1. d4 d5 2. Nf3 Bf5 3. Nh4 {-0.04/4 2.3} Bc8 {-0.05/10 4} 4. Nf3
{+0.00/4 1.8} Bf5 {-0.10/10 2.8} 5. Nh4 {+0.00/4 1.5} Bc8 {+0.00/13 2.7} 6.
Nf3 {+0.00/4 1.8}
{3-fold repetition} 1/2-1/2
Here the only loss for curiosity:
Code: Select all
[Event "RWBC"]
[Site "CAPPUCCINO"]
[Date "2016.11.29"]
[Round "9"]
[White "SjaakII_141-64"]
[Black "PyChess_20161118"]
[Result "0-1"]
[TimeControl "40/180"]
[Variant "crazyhouse"]
[Number "9"]
[Annotator "2. -0.13 1... +0.20"]
1. d3 d5 {+0.20/5 4} 2. Nf3 {-0.13/11 4} Nc6 {+0.14/4 5} 3. d4 {-0.14/12 5}
Bf5 {+0.12/4 2.1} 4. a3 {+0.00/10 2.5} Qd6 {+0.45/4 4} 5. e3 {+0.00/11 4}
O-O-O {+0.55/3 5} 6. Bd3 {-0.05/11 5} Bxd3 {+1.09/3 5} 7. cxd3
{-0.07/10 2.9} e5 {+0.81/3 4} 8. O-O {-0.04/9 5} B@g4 {+0.39/2 5} 9. dxe5
{+0.36/10 4} Nxe5 {+0.73/2 5} 10. Nxe5 {+0.84/9 2.5} Bxd1 {+2.70/2 5} 11.
Nxf7 {+1.10/9 4} Qg6 {-1.62/2 5} 12. B@h3+ {+1.65/8 4} P@g4 {-1.40/2 6} 13.
Nxh8 {+2.20/9 3} Qxd3 {-2.67/2 5} 14. R@a8+ {+0.75/8 2.9} Q@b8
{-6.85/3 2.8} 15. Rxb8+ {+0.82/10 5} Kxb8 {-7.85/2 5} 16. P@e2 {+0.42/9 5}
Bxe2 {-3.41/2 1.7} 17. N@f4 {-1.08/8 5} R@d1 {-3.19/2 5} 18. Q@e1
{-2.36/10 3} Rxe1 {+2.16/2 2.0} 19. Rxe1 {-3.44/9 3} gxh3 {+0.56/2 5} 20.
P@e7 {-3.73/8 7} Nxe7 {+4.47/1 1.1} 21. Nf7 {-5.17/8 5} P@a2 {+8.37/2 5}
22. Nc3 {-4.48/7 2.9} Qxc3 {+9.48/2 2.3} 23. Rxe2 {-6.15/8 4} Q@d1+
{+14.85/2 2.8} 24. R@f1 {-3.53/7 2.9} Qcxc1 {+9.04/2 2.1} 25. Rxc1
{+1.30/9 4} Qxc1 {+3.21/2 4} 26. Rxc1 {+0.66/8 3} R@b1 {+3.25/2 5} 27. Ree1
{+1.40/8 3} Rxc1 {+3.82/3 4} 28. Rxc1 {+1.32/8 5} P@d2 {+3.73/2 1.2} 29.
Rd1 {-0.91/6 4} R@e1+ {+5.06/2 4} 30. R@f1 {-0.54/8 6} hxg2 {+4.79/2 5} 31.
Kxg2 {-0.03/8 7} B@e4+ {+5.05/2 4} 32. f3 {-0.11/7 6} N@h4+ {+5.20/2 4} 33.
Kh3 {-0.31/7 6} Rxf1 {+5.47/2 1.5} 34. Rxf1 {-0.83/7 7} P@g2 {+1.66/2 4}
35. Nxg2 {-0.63/6 4} P@e2 {+2.52/2 2.2} 36. Ra1 {-0.94/6 5} Nxg2
{+0.71/2 2.6} 37. fxe4 {-0.94/4 3} e1=Q {+1.05/2 4} 38. B@f2 {-3.35/6 7}
R@f3+ {+12.92/1 1.5} 39. B@g3 {-6.27/5 5} Rxf7 40. Bxe1 {-3.87/7 9} dxe1=R
{+13.31/1 0.8} 41. Rxe1 {-6.27/6 4} Nxe1 {+15.27/3 11} 42. Bxc7+
{-6.36/6 5} Kxc7 {+20.03/3 15} 43. Q@a5+ {-6.66/4 2.4} R@b6 {+19.76/3 6}
44. R@c1+ {-7.01/5 2.6} N@c6 {+19.35/2 15} 45. Qxe1 {-6.47/7 5} N@g5+
{+20.41/3 12} 46. Kg4 {-7.04/6 5} Nxe4 {+18.31/2 12} 47. N@e6+
{-7.05/5 2.5} Kd7 {+18.67/3 5} 48. Nf4 {-8.71/6 5} B@f5+ {+24.12/2 11} 49.
Kf3 {-10.38/7 5} Rxb2 {+25.76/3 10} 50. P@c2 {-11.36/7 3} P@d2 {+29.44/3 5}
51. Qxd2 {-16.73/7 5} Nxd2+ {+36.50/3 3} 52. Ke2 {-159.90/6 2.8} Nb3
{+39.40/2 14} 53. Rd1 {-18.05/3 4} a1=Q {+43.03/2 5} 54. Rxa1 {-18.05/3 3}
Rxc2+ {+43.03/2 2.7} 55. P@d2 {-159.96/2 0.1} Q@b5+ {+43.26/2 3} 56. Kf2
{-19.41/7 4} Nxa1 {+43.34/2 2.4} 57. P@e2 {-19.57/7 3} Rxd2 {+48.74/1 9}
58. Q@c3 {-19.90/7 5} R@a2 {+48.89/2 4} 59. Kg1 {-21.03/6 3} P@e5
{+50.20/2 7} 60. P@b2 {-23.79/7 5} Rd1+ {+52.77/2 5} 61. Kf2 {-29.22/6 5}
{Xboard adjudication} 0-1
@Ferdinand & Tamas (hopefully reading here too?)
Is there a way to tell pypy which runs pychess, a max hash size?
I noticed that it used around 800MB when I came home. In the beginning
it was perhaps 200MB or so.
Guenther
fortunately this only happens when using pypy. If I use python 2.7 it will not leak.