Chess variant tournament: Capablanca and similar

[enhorning]

The games from the unbalanced starting position ended with this marathon endgame between Spartacus and Nebiyu:

[Event "Computer Chess Game"]
[Site "ENHORNING"]
[Date "2013.04.09"]
[Round "18"]
[White "NebiyuChess_1.44"]
[Black "Spartacus 1.02"]
[Result "1/2-1/2"]
[TimeControl "40/1200"]
[Variant "capablanca"]
[FEN "rbbnnkcaqr/pppppppppp/10/10/10/10/PPPPPPPPPP/RBBNNKCAQR w KQkq - 0 1"]
[SetUp "1"]

1. d4 {+0.15/17} d5 {-0.09/14 18} 2. Ne3 {+0.05/17 24} Ne6 {-0.28/15 30} 3.
g4 {+0.15/17 28} c6 {+0.11/14 21} 4. c3 {+0.05/17 29} Nf6 {+0.05/14 30} 5.
g5 {+0.05/16 29} Nh5 {+0.13/14 28} 6. Nf3 {+0.00/16 29} Nef4 {+0.16/14 40}
7. Bd2 {+0.05/18 29} Bc7 {+0.05/14 31} 8. Bc2 {+0.00/18 29} g6
{+0.13/14 30} 9. Rd1 {-0.05/15 29} Be6 {+1.16/15 30} 10. Ng4 {+0.35/18 24}
O-O-O {+0.97/15 33} 11. e3 {+0.70/18 23} Bf5 {+0.88/16 28} 12. Bxf5+
{+0.65/22 28} gxf5 {+1.03/17 43} 13. exf4 {+0.25/21 1:17} fxg4
{+0.98/16 30} 14. Cxg4 {+0.20/21 28} h6 {+0.92/14 21} 15. Ne5 {+0.20/18 34}
f6 {+1.40/14 16} 16. i4 {+0.05/17 28} fxe5 {+2.02/14 30} 17. ixh5
{+0.00/17 28} Qc2 {+1.95/14 55} 18. Ce3 {+1.05/17 28} Qxb2 {+1.79/14 30}
19. fxe5 {+1.00/16 22} hxg5 {+1.63/14 29} 20. Qi2+ {+1.10/16 28} Kb8
{+1.87/14 16} 21. Ai3 {+1.40/18 27} Qxa2 {+1.51/14 29} 22. Rg1
{+1.35/18 28} c5 {+2.35/13 15} 23. Cd3 {+2.05/16 22} cxd4 {+1.57/13 32} 24.
cxd4 {+2.05/18 25} Af7 {+1.43/14 31} 25. Bxg5 {+2.25/17 28} Re8
{+1.13/14 30} 26. Bf4 {+2.25/18 29} Cf8 {+0.71/14 19} 27. Bxi7
{+2.20/19 1:40} Ri8 {+0.50/15 37} 28. h6 {+2.20/18 24} Qc4 {+0.59/13 24}
29. Ag4 {+2.65/15 23} Ag5 {+0.38/12 18} 30. Ad7+ {+2.60/16 19} Cxd7
{-0.28/16 19} 31. Rxg5 {+2.50/18 26} Cb6 {-0.37/15 26} 32. Qj3
{+2.35/16 23} Cb2 {-0.54/14 23} 33. Kg2 {+2.60/19 23} Ce2 {-0.54/14 20} 34.
Cf3 {+2.65/16 35} Qa4 {-0.19/14 24} 35. Rd2 {+2.50/18 24} Cc1 {-0.40/15 30}
36. Cg1 {+2.55/20 27} Cxg1+ {-0.66/16 39} 37. Kxg1 Qa1+ {-0.65/16 39} 38.
Kg2 {+2.55/19 8} Qe1 {-0.75/16 44} 39. Qd3 {+2.45/20 40} Qi1 {-0.74/15 27}
40. Qi3 {+2.45/21 1:03} Qe1 {-0.76/17 47} 41. Qe3 {+2.45/21 29} Qi1
{-0.71/16 30} 42. j3 {+2.35/20 29} a6 {-0.47/14 21} 43. Qi3 {+2.60/18 26}
Qc1 {-0.52/15 28} 44. Rd3 {+2.60/19 22} Qf4 {-0.28/16 35} 45. Qe3
{+2.55/20 29} Qj4 {-0.76/15 18} 46. f4 {+2.55/19 23} Reh8 {-0.75/15 18} 47.
Qg3 {+2.75/21 22} Rxh6 {-1.02/16 32} 48. Bxh6 {+2.65/21 29} Qxh6
{-1.17/15 17} 49. Rf3 {+2.60/20 28} Ri7 {-0.85/14 17} 50. Rg6 {+2.60/20 30}
Qh7 {-0.90/16 36} 51. Qg5 {+2.65/21 24} Ka7 {-0.91/15 28} 52. f5
{+3.15/20 24} Bb6 {-1.18/15 29} 53. Rh3 {+2.85/23 29} Qi8 {-1.25/15 17} 54.
Rb3 {+2.75/21 27} e6 {-1.16/15 21} 55. Rxe6 {+2.85/22 28} Bxd4
{-1.16/15 25} 56. Rd3 {+2.80/22 30} Bb6 {-1.17/14 26} 57. Rxd5
{+2.75/21 31} Ri5 {-1.17/15 27} 58. Qg6 {+3.15/25 31} Qj8 {-0.88/16 34} 59.
Rxb6 {+1.85/23 38} Rg5+ {-0.67/16 20} 60. Kf3 {+1.55/28 30} Rxg6
{-0.57/17 23} 61. Rxg6 {+1.25/27 30} Qc1 {-0.57/17 20} 62. Rgd6
{+1.45/27 28} Qf1+ {-0.20/17 27} 63. Kg4 {+1.55/28 24} Qg2+ {-0.09/17 18}
64. Kh5 {+1.40/30 28} Qj5+ {-0.10/17 24} 65. Kg6 {+1.25/28 39} Qg8+
{-0.05/18 40} 66. Kf6 {+1.15/29 31} Qi6+ {-0.04/18 35} 67. Kg5
{+1.10/27 31} Qj5+ {-0.02/18 39} 68. Kf4 {+1.10/27 29} Qf1+ {-0.01/18 32}
69. Kg3 {+1.05/26 32} Qxf5 {+0.05/17 23} 70. e6 {+1.00/26 26} Qg6+
{+0.35/17 27} 71. Kh4 {+0.90/26 38} Qxj3 {+0.53/17 37} 72. Kg5
{+0.85/26 32} Qe3+ {+0.36/17 24} 73. Kg6 {+0.60/27 29} j5 {+0.26/18 43} 74.
Kf7 {+1.05/24 32} Qi7+ {+0.05/18 24} 75. Ke8 {+0.90/25 45} Qh8+
{+0.02/19 38} 76. Kd7 {+1.05/27 31} Kb8 {+0.00/19 1:04} 77. Rc5
{+1.05/23 26} Qh3 {+0.02/19 48} 78. Rdd5 {+1.25/23 28} j4 {+0.02/18 27} 79.
Rf5 {+0.90/24 39} Qd3+ {+0.01/19 27} 80. Ke8 {+0.80/25 35} j3 {+0.61/18 33}
81. Kf8 {+0.80/24 24} Qe3 {+1.50/17 18} 82. e7 {+0.80/26 29} j2
{+1.42/18 33} 83. e8=Q+ {+0.00/26 29} Qxe8+ {+1.90/19 19} 84. Kxe8
{+0.00/28 23} j1=Q {+1.90/18 27} 85. Rh5 {+0.90/25 26} Qg4 {+2.71/16 38}
86. Kf7 {+0.90/23 29} Qf4+ {+2.76/14 19} 87. Kg6 {+0.85/23 23} Ka8
{+2.79/15 34} 88. Rcf5 {+0.95/24 27} Qj4 {+2.72/15 27} 89. h4 {+0.95/23 24}
Qj3 {+2.21/15 21} 90. Kh6 {+0.95/24 29} Ka7 {+2.30/15 36} 91. Kg5
{+0.95/25 30} Qe3+ {+2.38/15 40} 92. Kg6 {+0.95/24 25} Qg3+ {+2.41/15 33}
93. Kf6 {+0.95/26 30} Qg4 {+2.43/15 30} 94. Rd5 {+0.95/25 30} Qf3+
{+2.46/15 29} 95. Kg6 {+0.90/24 29} Qj3 {+2.47/15 19} 96. Re5 {+0.95/25 28}
Qj6+ {+2.43/15 35} 97. Kf5 {+0.90/26 24} Qd6 {+2.54/15 40} 98. Rg5
{+0.90/27 31} Qc6 {+2.56/14 23} 99. h5 {+1.05/24 25} Qc8+ {+2.78/14 34}
100. Ke4 {+1.05/24 25} Qd7 {+2.89/14 24} 101. Rd5 {+1.05/25 32} Qj1
{+2.90/14 29} 102. Ke5 {+1.05/25 32} Qj8 {+2.91/14 19} 103. Rf5
{+1.05/26 29} Ka8 {+2.93/14 31} 104. Ke4 {+1.05/24 32} Qj3 {+2.71/15 39}
105. Rde5 {+1.05/25 32} Qc3 {+2.87/14 31} 106. Kf4 {+1.05/25 32} Qh3
{+2.89/14 19} 107. Kg5 {+1.05/25 27} Qg3+ {+2.24/15 32} 108. Kf6
{+1.05/25 33} Qg8 {+2.31/15 19} 109. Rg5 {+1.05/25 33} Qf8+ {+2.32/16 1:04}
110. Ke6 {+1.05/26 25} Ka7 {+2.78/15 19} 111. Rgf5 {+1.05/26 34} Qg7
{+2.75/15 28} 112. Kd5 {+1.05/26 29} Qg8+ {+2.74/15 38} 113. Ke4
{+1.05/26 35} Qc4+ {+2.86/14 25} 114. Ke3 {+1.05/26 31} Qb3+
{+2.74/15 1:21} 115. Kf4 {+1.05/26 28} Qi3 {+2.77/15 40} 116. Re6
{+1.05/27 38} Qh4+ {+3.11/15 34} 117. Ke5 {+1.05/27 33} a5 {+3.04/14 25}
118. Kd5 {+1.00/23 33} Qg4 {+3.01/14 29} 119. Ref6 {+0.65/22 36} Qe2
{+2.71/14 24} 120. Kd6 {+1.00/25 1:02} b6 {+2.91/15 22} 121. Re5
{+0.95/23 28} Qh2 {+2.97/15 17} 122. Rf7+ {+0.95/25 28} Ka6 {+2.97/16 24}
123. Rf3 {+0.30/23 30} Qd2+ {+3.34/15 25} 124. Ke6 {+0.95/23 29} a4
{+3.26/15 26} 125. Ree3 {+0.80/23 22} Qc2 {+3.33/14 16} 126. Rh3
{+0.75/25 29} Qc6+ {+3.30/15 32} 127. Kf7 {+0.75/25 29} Qh6 {+3.26/15 27}
128. Kg8 {+0.75/26 29} b5 {+3.17/15 19} 129. Re5 {+0.70/24 29} Qj8+
{+3.60/15 33} 130. Kg7 {+0.00/28 28} Qj4+ {+3.54/15 42} 131. Kf6
{+0.00/27 29} Qh6+ {+3.55/14 24} 132. Kf5 {+0.00/26 29} Qf8+ {+3.48/14 31}
133. Kg5 {+0.00/26 23} Qd8+ {+3.47/14 21} 134. Kf4 {+0.00/25 29} Qd4+
{+3.47/14 34} 135. Kf5 {+0.00/29 30} Qd1 {+3.40/14 31} 136. Ree3
{+0.70/24 23} Qd5+ {+2.93/14 36} 137. Kf6 {+0.65/25 30} Qd8+ {+2.95/14 22}
138. Kg7 {+0.65/25 30} Qg5+ {+2.86/14 27} 139. Kf7 {+0.65/25 31} Qi7+
{+2.89/15 42} 140. Ke6 {+0.65/26 27} Qh6+ {+2.90/15 35} 141. Kf5
{+0.65/26 27} Qf8+ {+2.87/15 1:41} 142. Kg4 {+0.50/25 30} Qj4+
{+2.86/14 33} 143. Kf3 {+0.70/23 29} Qh6 {+2.46/15 48} 144. Ke2
{+0.65/24 25} Ka5 {+2.32/14 26} 145. Rd3 {+0.65/23 31} Qe6+ {+2.36/15 28}
146. Kf2 {+0.55/24 31} Qf6+ {+2.38/14 19} 147. Kg2 {+0.60/22 28} Ka6
{+2.34/15 30} 148. h6 {+0.70/25 32} Qg6+ {+1.21/15 23} 149. Kf1
{+0.70/27 32} Qh7 {+1.22/15 1:26} 150. Ke2 {+0.70/24 32} b4 {+1.24/14 16}
151. Rd6+ {+0.55/22 28} Ka7 {+1.13/15 31} 152. Rhd3 {+0.60/23 30} Qh8
{+0.79/14 17} 153. Rd7+ {+0.85/22 1:10} Kb6 {+0.92/15 23} 154. R3d6+
{+0.40/22 30} Kc5 {+0.68/15 16} 155. Rd5+ {+0.45/23 24} Kc4 {+0.66/15 15}
156. Rd4+ {+0.50/24 27} Kb3 {+0.73/15 43} 157. Re4 {+0.65/22 25} Qxh6
{+1.55/15 27} 158. Rd3+ {+0.00/29 26} Ka2 {+1.52/16 50} 159. Rxb4
{+0.00/29 28} Qi6+ {+1.56/14 19} 160. Rf3 {+0.15/28 43} Qc6 {+1.14/14 25}
161. Rff4 {+1.00/25 23} a3 {+1.61/15 30} 162. Rf2 {+0.70/22 29} Qa6+
{+1.73/15 28} 163. Kf3+ {+0.25/24 29} Ka1 {+1.66/16 37} 164. Ri4
{+0.00/23 24} Qd3+ {+1.63/15 32} 165. Kg4 {+0.00/26 23} Qd4+ {+1.65/15 28}
166. Rf4 {+0.00/25 26} Qd7+ {+1.65/14 18} 167. Kh4 {+0.10/22 29} Qd8+
{+1.65/15 38} 168. Kg3 {+0.05/23 29} Qd3+ {+1.65/15 32} 169. Kh2
{+0.00/24 26} Qc2+ {+1.61/15 1:04} 170. Ki3 {+0.00/23 23} a2 {+1.62/16 29}
171. Ra4 {+0.00/26 30} Qc3+ {+1.64/16 31} 172. Kj4 {+0.00/27 30} Qh3
{+1.70/16 29} 173. Rie4 {+0.00/27 30} Qj1+ {+2.17/15 26} 174. Ki3
{+0.00/28 30} Qg1+ {+1.84/15 21} 175. Kj2 {+0.00/29 30} Qh2+ {+1.84/15 36}
176. Kj3 {+0.00/31 30} Qg3+ {+1.82/16 40} 177. Ki4 {+0.00/30 30} Qg6+
{+1.88/15 14} 178. Kh4 {+0.00/27 25} Qf6+ {+1.76/15 28} 179. Kg3
{+0.00/26 30} Qi3+ {+1.79/15 41} 180. Kg2 {+0.00/27 31} Qi2+ {+1.83/15 16}
181. Kf3 {+0.00/28 26} Qh3+ {+1.85/15 15} 182. Ke2 {+0.00/29 24} Qg2+
{+1.91/16 30} 183. Kd3 {+0.00/29 32} Qf1+ {+1.93/16 26} 184. Kd2
{+0.00/30 32} Qj5 {+1.92/15 25} 185. Red4 {+0.00/27 32} Qg2+ {+1.97/15 22}
186. Ke3 {+0.00/27 32} Qg3+ {+1.97/15 24} 187. Ke4 {+0.00/29 29} Qg6+
{+1.97/15 20} 188. Kf4 {+0.00/27 32} Qh6+ {+2.02/15 36} 189. Kf3
{+0.00/26 26} Qh5+ {+2.02/15 1:20} 190. Kf4 {+0.00/28 31} Qh4+
{+2.02/16 45} 191. Ke5 {+0.00/26 28} Qe1+ {+2.18/14 17} 192. Re4
{+0.00/27 28} Qi5+ {+2.17/14 14} 193. Kd4 {+0.00/27 35} Kb1 {+2.50/15 20}
194. Rb4+ {+0.00/31 35} Kc1 {+2.45/16 25} 195. Ra4 {+0.00/33 36} Qg7+
{+2.49/15 52} 196. Kc4 {+0.00/30 36} Qf7+ {+2.49/15 26} 197. Kc5
{+0.00/30 37} Qf2+ {+2.45/14 17} 198. Kd5 {+0.00/30 29} Qf5+ {+2.45/15 17}
199. Kd4 {+0.00/32 34} Qf6+ {+2.37/14 20} 200. Kc5 {+0.00/33 53} Qg5+
{+2.32/15 29} 201. Kc4 {+0.00/31 26} Qg6 {+2.31/15 31} 202. Kd5
{+0.00/30 28} Qg8+ {+2.33/15 21} 203. Ke5 {+0.00/28 22} Qb3 {+2.35/14 16}
204. Kd6 {+0.00/27 26} Qb6+ {+2.31/15 29} 205. Kd5 {+0.00/30 27} Qb5+
{+2.13/15 32} 206. Kd6 {+0.00/32 28} Qb8+ {+1.64/15 24} 207. Kc6
{+0.00/32 28} Qg8 {+1.72/15 15} 208. Rec4+ {+0.00/28 27} Kd1 {+1.86/17 48}
209. Rg4 {+0.00/29 29} Qc8+ {+1.89/15 23} 210. Kd6 {+0.00/32 29} Qd8+
{+1.88/16 33} 211. Ke6 {+0.00/30 29} Qe8+ {+1.85/16 28} 212. Kf6
{+0.00/31 29} Qh8+ {+1.86/15 24} 213. Kf5 {+0.00/30 24} Qi8+ {+1.89/15 29}
214. Kf4 {+0.00/30 30} Ke1 {+1.98/14 19} 215. Ke3 {+0.00/27 30} Qi7+
{+2.08/15 37} 216. Ke4 {+0.00/30 30} Qd7 {+4.16/15 29} 217. Ke3
{+0.00/32 30} Qe6+ {+4.06/15 25} 218. Kf3 {+0.00/31 24} Kf1 {+4.18/15 19}
219. Rgd4 {+0.00/33 24} Qh3+ {+4.49/14 18} 220. Kf4 {+0.00/39 31} Qc3
{+5.00/15 33}
{Xboard adjudication: 50-move rule} 1/2-1/2

Feels to me as if this should have been winnable for Spartacus at some point - but I am not good enough to be sure or tell where a different move might have been better.

So, that concludes the second (out of three) sets of games with the reduced participant list. Cross-table is:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Biha TJch Nebi Spar Nebi Joke Smir Sjaa TSCP Max-
 1. Bihasa v3.4.3             #### 1=1= 1101 1111 1111 1111 1111 1111 1111 1111   94%  34.0 (584.0, 533.5)
 2. TJchess10x8 1.1-x64       0=0= #### 1101 1111 1011 =010 =011 1111 1111 =111   74%  26.5 (614.0, 408.5)
 3. NebiyuChess_1.43          0010 0010 #### 0100 1=10 1101 1111 1=11 1111 1111   67%  24.0 (624.0, 322.5)
 4. Spartacus 1.02            0000 0000 1011 #### 101= 1110 1111 1110 1111 1011   62%  22.5 (630.0, 301.0)
 5. NebiyuChess_1.44          0000 0100 0=01 010= #### 1111 0110 1111 1111 1110   58%  21.0 (636.0, 276.3)
 6. Joker80.np                0000 =101 0010 0001 0000 #### 1001 0011 1110 1111   43%  15.5 (658.0, 204.8)
 7. Smirf BC-173g-X           0000 =100 0000 0000 1001 0110 #### 1100 0001 1011   32%  11.5 (674.0, 158.8)
 8. Sjaak 524                 0000 0000 0=00 0001 0000 1100 0011 #### 0111 101=   31%  11.0 (676.0, 128.0)
 9. TSCPtest                  0000 0000 0000 0000 0000 0001 1110 1000 #### 1111   25%   9.0 (684.0,  81.0)
10. Max-Plus 4.8S             0000 =000 0000 0100 0001 0000 0100 010= 0000 ####   14%   5.0 (700.0,  84.8)

Which gives us these combined standings:

1. Bihasa 3.4.3 - 94.5
2. TJChess10x8 - 82.0
3. Nebiyu 1.43 - 73.0
4. Spartacus - 66.0
5. Nebiyu 1.44 - 65.0
6. Joker - 59.0
7. Sjaak - 40.0
8. Smirf - 39.0
9. TSCP - 32.5
10. Max-Plus - 27.5

PGN can be downloaded from: https://www.dropbox.com/s/bc28pe06a5we6n2/Capa30sCD.pgn

Okay, only one more set of games, starting from the Gothic starting position, left to go!

[enhorning]

Another game where Max-Plus simply does not see the coming check-mate... it must have some bug to cause this:

[Event "Computer Chess Game"]
[Site "ENHORNING"]
[Date "2013.04.09"]
[Round "2"]
[White "Max-Plus 4.8S"]
[Black "TJchess10x8 1.1-x64"]
[Result "0-1"]
[TimeControl "40/1200"]
[Variant "capablanca"]
[FEN "rnbqckabnr/pppppppppp/10/10/10/10/PPPPPPPPPP/RNBQCKABNR w KQkq - 0 1"]
[SetUp "1"]

1. d4 {+0.03/10} d5 {-0.32/12 13} 2. Nc3 {+0.08/9 15} g6 {-0.28/12 16} 3.
g3 {+0.00/9 14} Nh6 {-0.14/11 13} 4. Bxd5 {+0.09/10 23} c6 {-0.27/12 42} 5.
Bf3 {+0.10/10 15} Bxd4 {-0.31/12 23} 6. Cd3 {+0.30/10 14} e5 {-0.20/12 17}
7. e3 {+0.30/10 35} Bf5 {-0.25/12 15} 8. Ne4 {+1.03/11 13} Bb6
{-0.05/13 37} 9. Cxd8 {+0.79/11 14} Cxd8 {-0.09/13 27} 10. Bd2
{+0.79/11 15} Na6 {+0.00/12 14} 11. Nh3 {+0.66/10 26} Bxh3+ {+0.33/12 11}
12. Axh3 {+0.04/12 23} f5 {-0.67/12 12} 13. Nc3 {+0.03/11 1:17} Ac4+
{+3.12/12 18} 14. Ke1 {+0.02/12 34} Cd3+ {+5.27/13 42} 15. cxd3
{+0.01/13 17} Axd3# {+199.99/1 0.1}
{Xboard adjudication: Checkmate} 0-1

[hgm]

Indeed, something is obviously wrong. I am in Berlin now, but when I get back next week I will try to feed this same position to Max-Plus to see if the error reproduces. If it does, it should be possible to debug it. If it doesn't, all we can say is that the hash table is somehow involved.

[enhorning]

So, over with over 800 games played (814, to be precise), the tournament is now over!

First half table:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Biha Biha TJch Nebi Nebi Joke Spar Sjaa Smir TSCP Max- Pair Ches Fair BigL ArcB Here
 1. Bihasa v3.4.3             #### 0=10 1111 1011 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111   95%  60.5 (1934.0, 1737.5)
 2. Bihasa v3.2               1=01 #### 1011 1110 1111 111= 1101 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111   92%  59.0 (1940.0, 1679.5)
 3. TJchess10x8 1.1-x64       0000 0100 #### 1111 1111 111= 1011 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111   87%  55.5 (1954.0, 1469.8)
 4. NebiyuChess_1.43          0100 0001 0000 #### 1111 0010 1111 11== 1111 1111 0111 1111 1111 1111 1111 1111 1111   77%  49.0 (1980.0, 1217.5)
 5. NebiyuChess_1.44          0000 0000 0000 0000 #### 1110 1110 1111 =111 1111 1101 1111 =111 1111 1111 1111 1111   69%  44.0 (2000.0, 971.3)
 6. Joker80.np                0000 000= 000= 1101 0001 #### 0011 1010 1=11 1111 =111 1111 1111 =111 1111 1111 1111   68%  43.5 (2002.0, 1000.0)
 7. Spartacus 1.02            0000 0010 0100 0000 0001 1100 #### 1111 1=11 1111 111= 1111 =111 1111 1111 1111 1111   68%  43.5 (2002.0, 967.0)
 8. Sjaak 524                 0000 0000 0000 00== 0000 0101 0000 #### 0010 1111 1011 0110 =1=1 111= 0111 1=11 1110   45%  29.0 (2060.0, 576.5)
 9. Smirf BC-173g-X           0000 0000 0000 0000 =000 0=00 0=00 1101 #### 1101 1111 1101 1100 ==01 0111 1011 1110   43%  27.5 (2066.0, 537.0)
10. TSCPtest                  0000 0000 0000 0000 0000 0000 0000 0000 0010 #### 1011 0=01 111= 1011 1=11 1111 1111   37%  23.5 (2082.0, 356.0)
11. Max-Plus 4.8S             0000 0000 0000 1000 0010 =000 000= 0100 0000 0100 #### 011= 01== 1111 0011 1111 0111   35%  22.5 (2086.0, 428.8)
12. Pair-o-Max 4.8S           0000 0000 0000 0000 0000 0000 0000 1001 0010 1=10 100= #### 111= 1==0 0100 1110 1111   32%  20.5 (2094.0, 344.5)
13. ChessV_WinBoard           0000 0000 0000 0000 =000 0000 =000 =0=0 0011 000= 10== 000= #### 001= =111 1111 0111   30%  19.0 (2100.0, 327.3)
14. Fairy-Max 4.8S            0000 0000 0000 0000 0000 =000 0000 000= ==10 0100 0000 0==1 110= #### =011 1=01 1101   26%  16.5 (2110.0, 282.8)
15. BigLion80 2.23x WB        0000 0000 0000 0000 0000 0000 0000 1000 1000 0=00 1100 1011 =000 =100 #### 0111 1101   24%  15.5 (2114.0, 254.0)
16. ArcBishop80 1.01 WB       0000 0000 0000 0000 0000 0000 0000 0=00 0100 0000 0000 0001 0000 0=10 1000 #### 1=11   13%   8.5 (2142.0, 125.5)
17. Heretic 0.3               0000 0000 0000 0000 0000 0000 0000 0001 0001 0000 1000 0000 1000 0010 0010 0=00 ####   10%   6.5 (2150.0, 134.3)

Second half table:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Bih TJc Neb Spa Neb Jok Smi Sja TSC Max
 1. Bihasa v3.4.3             ### 1=1 110 111 111 111 111 111 111 111
                              ### =11 111 111 111 111 111 111 111 111   96%  52.0 (1308.0, 1231.5)

 2. TJchess10x8 1.1-x64       0=0 ### 110 111 101 =01 =01 111 111 =11
                              =00 ### 101 111 111 011 110 111 101 111   73%  39.5 (1383.0, 895.3)

 3. NebiyuChess_1.43          001 001 ### 010 1=1 110 111 1=1 111 111
                              000 010 ### 011 0=0 111 1=1 111 111 111   69%  37.0 (1398.0, 754.5)

 4. Spartacus 1.02            000 000 101 ### 101 111 111 111 111 101
                              000 000 100 ### =1= 000 111 011 111 101   57%  31.0 (1434.0, 600.5)

 5. NebiyuChess_1.44          000 010 0=0 010 ### 111 011 111 111 111
                              000 000 1=1 =0= ### 100 011 111 111 001   56%  30.0 (1440.0, 592.5)

 6. Joker80.np                000 =10 001 000 000 ### 100 001 111 111
                              000 100 000 111 011 ### 110 111 011 111   49%  26.5 (1461.0, 519.8)

 7. Smirf BC-173g-X           000 =10 000 000 100 011 ### 110 000 101
                              000 001 0=0 000 100 001 ### 000 101 101   30%  16.0 (1524.0, 350.8)

 8. Sjaak 524                 000 000 0=0 000 000 110 001 ### 011 101
                              000 000 000 100 000 000 111 ### 101 =11   30%  16.0 (1524.0, 259.0)

 9. TSCPtest                  000 000 000 000 000 000 111 100 ### 111
                              000 010 000 000 000 100 010 010 ### 110   24%  13.0 (1542.0, 207.0)

10. Max-Plus 4.8S             000 =00 000 010 000 000 010 010 000 ###
                              000 000 000 010 110 000 010 =00 001 ###   17%   9.0 (1566.0, 210.8)

Total standings for the 10 participants who qualified for the second half:
1. Bihasa 3.4.3 - 112.5
2. TJchess10x8 1.1-x64 - 95.0
3. NebiyuChess_1.43 - 86.0
4. Spartacus 1.02 - 74.5
5. NebiyuChess_1.44 - 74.0
6. Joker80.np - 70.0
7. Sjaak 524 - 45.0
8. Smirf BC-173g-X - 43.5
9. TSCPGothic - 36.5
10. Max-Plus 4.8S - 31.5

Bayeselo gives these ratings and LOS matrix:

Code: Select all

ResultSet-EloRating>ratings
Rank Name                  Elo    +    - games score oppo. draws
   1 Bihasa v3.4.3         640  122  122   118   95%    15    3%
   2 Bihasa v3.2           549  152  152    64   92%   -34    3%
   3 TJchess10x8 1.1-x64   388   87   87   118   81%    36    5%
   4 NebiyuChess_1.43      280   80   80   118   73%    45    5%
   5 NebiyuChess_1.44      181   75   75   118   63%    54    5%
   6 Spartacus 1.02        179   75   75   118   63%    54    4%
   7 Joker80.np            147   75   75   118   59%    57    5%
   8 Sjaak 524             -65   73   73   118   38%    74    7%
   9 Smirf BC-173g-X       -70   74   74   118   37%    75    6%
  10 TSCPtest             -143   76   76   118   31%    81    3%
  11 Max-Plus 4.8S        -180   77   77   118   27%    84    6%
  12 Pair-o-Max 4.8S      -211   99   99    64   32%    13    8%
  13 ChessV_WinBoard      -218   95   95    64   30%    14   16%
  14 Fairy-Max 4.8S       -262   98   98    64   26%    16   14%
  15 BigLion80 2.23x WB   -293  104  104    64   24%    18    5%
  16 ArcBishop80 1.01 WB  -430  118  118    64   13%    27    5%
  17 Heretic 0.3          -490  131  131    64   10%    31    2%
ResultSet-EloRating>los
                     Bi Bi TJ Ne Ne Sp Jo Sj Sm TS Ma Pa Ch Fa Bi Ar He
Bihasa v3.4.3           83 99 99 99 99 99100100100100100100100100100100
Bihasa v3.2          16    97 99 99 99 99 99 99 99 99 99 99100100100100
TJchess10x8 1.1-x64   0  2    97 99 99 99 99 99100100100100100100100100
NebiyuChess_1.43      0  0  2    97 97 99 99 99 99 99 99 99 99 99100100
NebiyuChess_1.44      0  0  0  2    51 74 99 99 99 99 99 99 99 99 99100
Spartacus 1.02        0  0  0  2 48    73 99 99 99 99 99 99 99 99 99100
Joker80.np            0  0  0  0 25 26    99 99 99 99 99 99 99 99 99 99
Sjaak 524             0  0  0  0  0  0  0    53 93 98 99 99 99 99 99 99
Smirf BC-173g-X       0  0  0  0  0  0  0 46    92 98 98 99 99 99 99 99
TSCPtest              0  0  0  0  0  0  0  6  7    76 86 89 97 99 99 99
Max-Plus 4.8S         0  0  0  0  0  0  0  1  1 23    69 73 90 95 99 99
Pair-o-Max 4.8S       0  0  0  0  0  0  0  0  1 13 30    54 77 87 99 99
ChessV_WinBoard       0  0  0  0  0  0  0  0  0 10 26 45    74 86 99 99
Fairy-Max 4.8S        0  0  0  0  0  0  0  0  0  2  9 22 25    66 98 99
BigLion80 2.23x WB    0  0  0  0  0  0  0  0  0  0  4 12 13 33    96 99
ArcBishop80 1.01 WB   0  0  0  0  0  0  0  0  0  0  0  0  0  1  3    76
Heretic 0.3           0  0  0  0  0  0  0  0  0  0  0  0  0  0  0 23

So, from this data, it looks very likely that 1.44 was a regression as far as Capablanca variants was concerned - 97.2% LOS for the 1.43 version.

Games from the first half can be downloaded here: https://www.dropbox.com/s/3fnskpwc5k383zq/Capa30sAB.pgn
and games from the second half can be downloaded here: https://www.dropbox.com/s/px5zmzgu8p3l1 ... 30sCDE.pgn

Thanks to all the participants, and congratulations to Bihasa for a dominating performance.

Next tournament, in a few week's time, will be Makruk (Thai Chess). After that, I'll do Grand Chess, which looks like it will be a very small tournament - I only know of Nebiyu, Sjaak and ChessV which are able to play it. Shame as out of the variants with the Capablanca compounds, that one has been the one I've personally enjoyed the most. My preliminary thought after that is to branch out and run XiangQi - but that's likely to be a couple of months away.

Oh, and if I have spare time, I'll run a mini-tournament somewhere in-between, pitting 6-core Nebiyu against TJChess10x8 and Bihasa, and see if its MP-capability is sufficient to let it beat them.

[Daniel Shawul]

Thanks for the games! I can only wonder what the heck went wrong, that made it loose many games against the previous versions. I didn't do self tests but improvements against bihasa. Well this is a flawed way of testing and it showed here. I will try and revert all changes to 1.43 and work on from there. Don't worry about the 6 core test with this version.

[TonyJH]

Thanks for running this tournament. For your Grand Chess tourney, I have a version of TJchess that plays Grand Chess. I think it's not on my website right now. I'll update it later.

[hgm]

I think the version of Spartacus that participated in the 10x8 tourney should also be able to play Grand Chess. If not, I am sure I have a version somewhere that does. Grand Chess was the reason I modified Spartacus to use a 24x10 board in stead of its original 32x8 board.

Not sure about Max-Plus. Fairy-Max is really married to 8-rank boards, as it uses some very trick code to give bonuses for 6th and 7th-rank Pawns, which derives the bonus from the rank number. I am playing with the idea to equip it with true piece-square tables, though, and apart from a null table and a centralize table I could then make separate tables for the Pawns, including these advance bonuses. That would then allow me to generalize the board size, using 12x10 for Grand Chess in stead of the 16x8 it uses now. Main problem is that Grand Chess has this rule that you can only promote to captured pieces.

As to Makruk, it should be a piece of cake to have HaChu play it. (Of course it will be quite weak, without hash and only limited QS.) It could be interesting to make a version of Pair-o-Max that has a sort of poor-man's version of the counting rules, making the leading side avoid positions without Pawns. (The normal way of discounting end-game scores for lack of Pawns is likely to backfire in variants like Makruk and Shatranj, where promotion hardly gains you anything.) I am aware you won't use these rules in the tourney, but it would still be interesting to see how many of the games would have been draws with this rule.

[Ferdy]

Next tournament, in a few week's time, will be Makruk (Thai Chess). After that, I'll do Grand Chess, which looks like it will be a very small tournament - I only know of Nebiyu, Sjaak and ChessV which are able to play it. Shame as out of the variants with the Capablanca compounds, that one has been the one I've personally enjoyed the most. My preliminary thought after that is to branch out and run XiangQi - but that's likely to be a couple of months away.

Thanks for running tourney.
I will send the makruk engine in about a week. I will also try to create an engine that can play grand chess, I hope to finish it before your planned tourney schedule.

[enhorning]

Oh, and if I have spare time, I'll run a mini-tournament somewhere in-between, pitting 6-core Nebiyu against TJChess10x8 and Bihasa, and see if its MP-capability is sufficient to let it beat them.

... so, I was trying to do this, but was unable to convince Nebiyu to use more than one core?

I changed the "Common engine settings" in WB to 6 cores. My tournament-file looks like this:

Code: Select all

-participants {Bihasa_v3.4.3
TJchess10x8_1.1_64bit
Nebiyu Chess Variants
}
-seedBase 884206568
-tourneyType 0
-tourneyCycles 6
-defaultMatchGames 2
-syncAfterRound false
-syncAfterCycle false
-saveGameFile "C:\WinBoard-4.7.0\Files\Capa60sMP.pgn"
-loadGameFile "C:\WinBoard-4.7.0\Files\CapaSetups.pgn"
-loadGameIndex -2
-loadPositionFile ""
-loadPositionIndex -2
-rewindIndex 0
-usePolyglotBook false
-polyglotBook default_book.bin
-bookDepth 12
-bookVariation 50
-discourageOwnBooks false
-defaultHashSize 1024
-defaultCacheSizeEGTB 64
-ponderNextMove false
-smpCores 6
-mps 40
-tc 40
-inc -1.00
-results "*"

And I went and changed in Nebiyu's .ini-file to this:

Code: Select all

turn logging on and off
log on

#hashtable size in mb
ht 1024

#number of processors
mt 6

#split depths
smp_depth           4
cluster_depth       8

#nodes to be searched before message polling
message_poll_nodes  200 


#resign value in centipawns
resign              99999

But despite this, Nebiyu is still only using one core... Daniel Shawul (or anybody else), any suggestions for how to make it work?

[Daniel Shawul]

Probably it is not compiled for MP so I will send you a link to the new one right away.
Edit: Here is a link. Please keep the log files on so that I can debug problems later if anything comes up.

Thanks.