Dev Builds » 20140503-1940

Use this dev build

NCM plays each Stockfish dev build 20,000 times against Stockfish 7. This yields an approximate Elo difference and establishes confidence in the strength of the dev builds.

Summary

Host Duration Avg Base NPS Games Wins Losses Draws Elo
ncm-et-3 06:03:22 1992075 2398 158 885 1355 -108.75 +/- 8.97
ncm-et-4 06:02:57 1948376 2367 139 879 1349 -112.38 +/- 8.95
ncm-et-5 01:47:36 2023610 723 35 253 435 -108.12 +/- 15.46
ncm-et-6 01:47:48 2026636 720 39 262 419 -111.26 +/- 15.95
ncm-et-7 01:47:48 2009712 702 47 270 385 -114.32 +/- 16.94
ncm-et-8 01:47:19 1976038 711 34 281 396 -125.94 +/- 16.51
ncm-et-9 06:03:16 1991348 2383 154 921 1308 -115.95 +/- 9.18
ncm-et-10 06:02:42 1983298 2372 140 935 1297 -121.13 +/- 9.2
ncm-et-11 01:47:42 2016853 716 38 270 408 -116.79 +/- 16.23
ncm-et-12 01:47:37 2021326 713 47 260 406 -107.06 +/- 16.37
ncm-et-13 06:03:09 1989295 2398 143 919 1336 -116.62 +/- 9.03
ncm-et-14 01:47:32 2018964 714 47 257 410 -105.3 +/- 16.26
ncm-et-15 06:03:08 1988504 2364 143 867 1354 -109.93 +/- 8.93
ncm-et-16 01:47:40 2011656 719 34 293 392 -131.03 +/- 16.66
20000 1198 7552 11250 -114.34 +/- 3.11

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
91382 ncm-et-15 2019-08-04 04:42 00:22:54 1971414 147 11 57 79 -112.49 +/- 37.79
91381 ncm-et-10 2019-08-04 04:39 00:24:48 1970014 164 8 58 98 -109.4 +/- 32.74
91380 ncm-et-4 2019-08-04 04:39 00:25:09 1909565 152 14 51 87 -86.31 +/- 35.78
91379 ncm-et-9 2019-08-04 04:38 00:26:17 1983759 173 12 75 86 -132.61 +/- 36.51
91378 ncm-et-13 2019-08-04 04:37 00:27:45 1968320 176 10 61 105 -103.65 +/- 31.78
91377 ncm-et-3 2019-08-04 04:37 00:28:01 1978100 187 12 60 115 -91.22 +/- 30.21
91376 ncm-et-4 2019-08-04 03:23 01:15:40 1975556 500 23 196 281 -125.39 +/- 19.55
91375 ncm-et-9 2019-08-04 03:22 01:14:59 1975804 500 30 197 273 -120.67 +/- 20.09
91374 ncm-et-10 2019-08-04 03:22 01:16:18 1965547 500 29 192 279 -117.55 +/- 19.77
91373 ncm-et-15 2019-08-04 03:22 01:18:48 1981335 500 32 173 295 -100.71 +/- 19.04
91372 ncm-et-13 2019-08-04 03:21 01:14:54 1976694 500 29 201 270 -124.6 +/- 20.22
91371 ncm-et-3 2019-08-04 03:21 01:15:11 1980442 500 37 189 274 -109.07 +/- 20.15
91370 ncm-et-4 2019-08-04 02:07 01:15:00 1963840 500 35 178 287 -102.22 +/- 19.48
91369 ncm-et-10 2019-08-04 02:05 01:15:37 1965863 500 33 203 264 -123.02 +/- 20.59
91368 ncm-et-13 2019-08-04 02:05 01:15:15 1972812 500 43 185 272 -101.46 +/- 20.32
91367 ncm-et-9 2019-08-04 02:04 01:17:21 1962193 500 33 207 260 -126.18 +/- 20.79
91366 ncm-et-15 2019-08-04 02:04 01:17:16 1971736 500 34 167 299 -94.7 +/- 18.88
91365 ncm-et-3 2019-08-04 02:03 01:16:06 1968323 500 40 190 270 -107.54 +/- 20.39
91364 ncm-et-4 2019-08-04 00:46 01:19:20 1906918 500 25 189 286 -118.33 +/- 19.34
91363 ncm-et-13 2019-08-04 00:46 01:17:26 1969094 500 26 193 281 -120.67 +/- 19.61
91362 ncm-et-3 2019-08-04 00:46 01:16:18 1973133 500 24 180 296 -112.14 +/- 18.83
91361 ncm-et-10 2019-08-04 00:46 01:18:20 1955718 500 23 196 281 -125.39 +/- 19.55
91360 ncm-et-9 2019-08-04 00:46 01:16:58 1974208 500 27 185 288 -113.68 +/- 19.29
91359 ncm-et-15 2019-08-04 00:46 01:16:42 1969099 500 28 201 271 -125.39 +/- 20.15
23720 ncm-et-7 2018-04-27 18:23 00:30:54 2024263 202 17 74 111 -100.77 +/- 31.81
23719 ncm-et-3 2018-04-27 18:22 00:31:23 2027046 211 12 69 130 -96.25 +/- 28.27
23718 ncm-et-9 2018-04-27 18:22 00:31:32 2025899 210 17 84 109 -114.86 +/- 32.34
23717 ncm-et-10 2018-04-27 18:22 00:31:33 2021977 208 21 78 109 -97.71 +/- 32.44
23716 ncm-et-4 2018-04-27 18:22 00:31:59 1972266 215 12 79 124 -111.99 +/- 29.47
23715 ncm-et-11 2018-04-27 18:21 00:32:13 2015798 216 21 70 125 -80.21 +/- 29.79
23714 ncm-et-12 2018-04-27 18:21 00:32:14 2022630 213 13 81 119 -114.94 +/- 30.37
23713 ncm-et-16 2018-04-27 18:21 00:32:18 2022630 219 12 86 121 -122.2 +/- 30.09
23712 ncm-et-14 2018-04-27 18:21 00:32:31 2018070 214 11 83 120 -121.63 +/- 30.07
23711 ncm-et-8 2018-04-27 18:21 00:32:17 1953560 211 9 86 116 -132.92 +/- 30.56
23710 ncm-et-15 2018-04-27 18:21 00:32:50 2018887 217 14 74 129 -98.63 +/- 28.79
23709 ncm-et-13 2018-04-27 18:21 00:33:01 2024427 222 10 85 127 -122.18 +/- 28.99
23708 ncm-et-6 2018-04-27 18:20 00:33:14 2027374 220 12 76 132 -104.08 +/- 28.24
23707 ncm-et-5 2018-04-27 18:20 00:33:21 2022630 223 9 60 154 -80.89 +/- 24.45
23706 ncm-et-15 2018-04-27 17:05 01:14:38 2018558 500 24 195 281 -123.81 +/- 19.57
23705 ncm-et-14 2018-04-27 17:05 01:15:01 2019859 500 36 174 290 -98.44 +/- 19.35
23704 ncm-et-8 2018-04-27 17:05 01:15:02 1998517 500 25 195 280 -123.02 +/- 19.64
23703 ncm-et-5 2018-04-27 17:05 01:14:15 2024590 500 26 193 281 -120.67 +/- 19.61
23702 ncm-et-12 2018-04-27 17:05 01:15:23 2020022 500 34 179 287 -103.73 +/- 19.47
23701 ncm-et-9 2018-04-27 17:05 01:16:09 2026227 500 35 173 292 -98.44 +/- 19.24
23700 ncm-et-16 2018-04-27 17:05 01:15:22 2000683 500 22 207 271 -134.95 +/- 20.03
23699 ncm-et-11 2018-04-27 17:05 01:15:29 2017908 500 17 200 283 -133.34 +/- 19.29
23698 ncm-et-3 2018-04-27 17:05 01:16:23 2025409 500 33 197 270 -118.33 +/- 20.29
23697 ncm-et-6 2018-04-27 17:05 01:14:34 2025899 500 27 186 287 -114.45 +/- 19.34
23696 ncm-et-13 2018-04-27 17:05 01:14:48 2024427 500 25 194 281 -122.24 +/- 19.59
23695 ncm-et-4 2018-04-27 17:05 01:15:49 1962116 500 30 186 284 -112.14 +/- 19.54
23694 ncm-et-7 2018-04-27 17:05 01:16:54 1995161 500 30 196 274 -119.89 +/- 20.04
23693 ncm-et-10 2018-04-27 17:05 01:16:06 2020674 500 26 208 266 -132.54 +/- 20.37

Commit

Commit ID c0d3010438e9cec2d40dec9e92695b58514497bb
Author Ronald de Man
Date 2014-05-03 19:40:09 UTC
Fix KXK endgame Position is win also if strong side has a bishop and a knight (plus other material, otherwise KBNK would be triggered instead of KXK). This fixes a subtle bug where a search on position k7/8/8/8/8/P7/PB6/K7 b - - 6 1 Instead of returning a draw score, suddendly returns a big score. This happens because at one point in search we reach this position: 8/Pk6/8/8/8/4B3/P7/K7 w - - 3 8 Where white can promote. In case of rook promotion (and also in case of queen promotion) the resutling position gets a huge static eval that is above VALUE_KNOWN_WIN (from the point of view of white). So for rook promotion it is && futilityBase > -VALUE_KNOWN_WIN that prevents futility pruning in qsearch. (Removing this condition indeed lets the problem occur). Raising the static eval for K+B+N+X v K to a value higher than VALUE_KNOWN_WIN fixes this particular problem without having to introduce an extra futility pruning condition in qsearch. I just checked and it seems K+R v K, K+2B v K and even K+B+N v K already get a huge static eval. Why not K+B+N+P v K? I think this fix corrects an oversight. There is special code for KBNK, but KBNXK is handled by KXK, so the test for sufficient material should also test for B+N. bench: 8678654
Copyright 2011–2024 Next Chess Move LLC