Dev Builds » 20160607-1306

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 05:48:45 1990507 2384 415 188 1781 +33.18 +/- 6.95
ncm-et-4 05:50:07 1972422 2386 430 187 1769 +35.51 +/- 7.02
ncm-et-5 01:44:09 2023531 712 127 57 528 +34.27 +/- 12.86
ncm-et-6 01:44:01 2026312 719 118 52 549 +31.98 +/- 12.23
ncm-et-7 01:45:32 2012872 715 126 61 528 +31.67 +/- 12.93
ncm-et-8 01:43:26 1974370 706 152 55 499 +48.04 +/- 13.69
ncm-et-9 05:51:40 1988566 2374 448 174 1752 +40.28 +/- 7.06
ncm-et-10 05:48:35 1982265 2380 413 180 1787 +34.12 +/- 6.9
ncm-et-11 01:44:43 1988762 709 132 57 520 +36.89 +/- 13.08
ncm-et-12 01:44:05 2020103 718 139 55 524 +40.83 +/- 13.06
ncm-et-13 05:49:58 1993039 2380 443 175 1762 +39.29 +/- 7.03
ncm-et-14 01:42:16 2020754 711 142 49 520 +45.71 +/- 13.04
ncm-et-15 05:51:20 1989021 2394 433 192 1769 +35.09 +/- 7.04
ncm-et-16 01:44:27 2021977 712 122 55 535 +32.79 +/- 12.61
20000 3640 1537 14823 +36.67 +/- 2.42

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
69140 ncm-et-9 2019-02-07 10:35 00:22:42 1965858 158 38 14 106 +53.19 +/- 30.76
69139 ncm-et-3 2019-02-07 10:32 00:25:21 1974773 167 30 16 121 +29.19 +/- 27.55
69138 ncm-et-13 2019-02-07 10:32 00:25:42 1980600 172 36 14 122 +44.68 +/- 27.71
69137 ncm-et-15 2019-02-07 10:32 00:25:35 1975614 170 34 12 124 +45.22 +/- 26.82
69136 ncm-et-10 2019-02-07 10:32 00:25:52 1969397 166 28 13 125 +31.48 +/- 26.09
69135 ncm-et-4 2019-02-07 10:31 00:26:14 1954573 167 30 15 122 +31.29 +/- 27.22
69134 ncm-et-9 2019-02-07 09:19 01:14:49 1971584 500 98 39 363 +41.19 +/- 15.76
69133 ncm-et-3 2019-02-07 09:18 01:13:20 1977451 500 79 44 377 +24.36 +/- 15.04
69132 ncm-et-13 2019-02-07 09:18 01:13:02 1973119 500 92 30 378 +43.3 +/- 14.8
69131 ncm-et-15 2019-02-07 09:17 01:13:37 1972816 500 89 47 364 +29.25 +/- 15.8
69130 ncm-et-4 2019-02-07 09:17 01:13:25 1971883 500 102 44 354 +40.48 +/- 16.3
69129 ncm-et-10 2019-02-07 09:17 01:13:55 1964629 500 83 35 382 +33.46 +/- 14.65
69128 ncm-et-9 2019-02-07 08:05 01:13:15 1975137 500 93 37 370 +39.08 +/- 15.35
69127 ncm-et-4 2019-02-07 08:03 01:13:08 1966461 500 98 30 372 +47.55 +/- 15.13
69126 ncm-et-13 2019-02-07 08:03 01:14:08 1982223 500 85 40 375 +31.35 +/- 15.11
69125 ncm-et-10 2019-02-07 08:03 01:13:07 1961237 500 90 39 371 +35.56 +/- 15.32
69124 ncm-et-3 2019-02-07 08:02 01:14:57 1966153 500 85 37 378 +33.46 +/- 14.9
69123 ncm-et-15 2019-02-07 08:02 01:14:32 1975293 500 83 46 371 +25.76 +/- 15.4
69122 ncm-et-10 2019-02-07 06:48 01:13:05 1961542 500 87 44 369 +29.95 +/- 15.49
69121 ncm-et-3 2019-02-07 06:48 01:12:23 1973027 500 84 42 374 +29.25 +/- 15.19
69120 ncm-et-13 2019-02-07 06:48 01:13:16 1973278 500 103 44 353 +41.19 +/- 16.36
69119 ncm-et-9 2019-02-07 06:48 01:15:28 1968657 500 91 39 370 +36.26 +/- 15.38
69118 ncm-et-4 2019-02-07 06:48 01:13:32 1922498 500 85 43 372 +29.25 +/- 15.31
69117 ncm-et-15 2019-02-07 06:48 01:12:31 1973125 500 90 42 368 +33.46 +/- 15.53
6768 ncm-et-13 2018-03-09 23:22 00:30:54 2025899 208 38 12 158 +43.66 +/- 22.77
6767 ncm-et-14 2018-03-09 23:22 00:30:56 2020347 211 44 17 150 +44.7 +/- 24.93
6766 ncm-et-8 2018-03-09 23:22 00:31:06 1936438 206 47 16 143 +52.68 +/- 25.88
6765 ncm-et-5 2018-03-09 23:22 00:31:16 2020836 212 42 16 154 +42.83 +/- 24.18
6764 ncm-et-3 2018-03-09 23:22 00:31:13 2024263 217 40 14 163 +41.83 +/- 22.74
6763 ncm-et-6 2018-03-09 23:22 00:31:09 2024101 219 40 14 165 +41.44 +/- 22.54
6762 ncm-et-16 2018-03-09 23:22 00:31:13 2021488 212 41 12 159 +47.83 +/- 22.96
6761 ncm-et-7 2018-03-09 23:22 00:31:09 2025408 215 36 22 157 +22.66 +/- 24.07
6760 ncm-et-11 2018-03-09 23:22 00:31:21 1991046 209 32 19 158 +21.64 +/- 23.21
6759 ncm-et-9 2018-03-09 23:22 00:31:24 2025736 216 36 17 163 +30.64 +/- 22.8
6758 ncm-et-15 2018-03-09 23:22 00:31:24 2019696 224 46 11 167 +54.74 +/- 22.4
6757 ncm-et-12 2018-03-09 23:22 00:31:23 2019045 218 38 18 162 +31.96 +/- 23.22
6756 ncm-et-4 2018-03-09 23:22 00:31:08 1995019 219 34 17 168 +27.02 +/- 22.08
6755 ncm-et-10 2018-03-09 23:22 00:31:17 2019208 214 41 9 164 +52.35 +/- 21.92
6726 ncm-et-4 2018-03-09 20:17 01:12:40 2024100 500 81 38 381 +29.95 +/- 14.74
6725 ncm-et-6 2018-03-09 20:16 01:12:52 2028523 500 78 38 384 +27.85 +/- 14.57
6724 ncm-et-10 2018-03-09 20:16 01:11:19 2017582 500 84 40 376 +30.65 +/- 15.05
6723 ncm-et-16 2018-03-09 20:16 01:13:14 2022466 500 81 43 376 +26.46 +/- 15.08
6722 ncm-et-11 2018-03-09 20:16 01:13:22 1986478 500 100 38 362 +43.3 +/- 15.8
6721 ncm-et-3 2018-03-09 20:16 01:11:31 2027375 500 97 35 368 +43.3 +/- 15.43
6720 ncm-et-13 2018-03-09 20:16 01:12:56 2023119 500 89 35 376 +37.67 +/- 14.99
6719 ncm-et-14 2018-03-09 20:16 01:11:20 2021162 500 98 32 370 +46.13 +/- 15.28
6718 ncm-et-12 2018-03-09 20:16 01:12:42 2021162 500 101 37 362 +44.72 +/- 15.79
6717 ncm-et-8 2018-03-09 20:16 01:12:20 2012303 500 105 39 356 +46.13 +/- 16.13
6716 ncm-et-5 2018-03-09 20:16 01:12:53 2026227 500 85 41 374 +30.65 +/- 15.18
6715 ncm-et-15 2018-03-09 20:16 01:13:41 2017583 500 91 34 375 +39.78 +/- 15.03
6714 ncm-et-7 2018-03-09 20:16 01:14:23 2000337 500 90 39 371 +35.56 +/- 15.32
6713 ncm-et-9 2018-03-09 20:16 01:14:02 2024427 500 92 28 380 +44.72 +/- 14.65

Commit

Commit ID 6e2ca97d93812b2f630f8b4ec1042105557abaec
Author Marco Costalba
Date 2016-06-07 13:06:26 UTC
Fix syzygy DTZ bug In this position: 3K4/8/3k4/8/4p3/4B3/5P2/8 w - - 0 5 Current DTZ probe returns 1 instead of 15 What happens is that the double push f4 is erroneously detected as a win move. After the push we have: [D]3K4/8/3k4/8/4pP2/4B3/8/8 b - f3 0 5 And here the code misses the possible ep capture exf3. The bug is in probe_dtz_no_ep() where is used probe_ab() that is blind to ep captures so it returns v == 2 (win) for position 3K4/8/3k4/8/4pP2/4B3/8/8 b - f3 0 5 Note that at the caller site the original position did not have any possible ep capture, so probe_dtz() returns immediately after calling probe_dtz_no_ep(). The fix is to call the ep-aware probe_wdl() instead of probe_ab() I have verified that DTZ is correct now and also there are no more mistmatches compared to the new 'syzygy' branch. Tested on a set of more than 600 endgame positions, included some tricky ones. For people interested to redo the test or doing additional tests please pull branch tb_dbg from https://github.com/mcostalba/Stockfish repo. bench: 8450534 (bench unaffected because syzygy is not exercized during bench)
Copyright 2011–2024 Next Chess Move LLC