Dev Builds » 20190927-1046

Use this dev build

NCM plays each Stockfish dev build 20,000 times against Stockfish 15. This yields an approximate Elo difference and establishes confidence in the strength of the dev builds.

Summary

Host Duration Avg Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo
ncm-dbt-01 06:26:47 585353 4016 58 2682 1276 -271.41 ± 7.17 695 1235 77 1 0 -679.02 ± 39.43
ncm-dbt-02 06:25:08 586416 4000 67 2654 1279 -267.42 ± 7.1 673 1242 84 1 0 -663.25 ± 37.69
ncm-dbt-03 06:25:30 585570 4002 49 2628 1325 -266.03 ± 6.86 649 1282 69 1 0 -697.3 ± 41.73
ncm-dbt-04 06:26:55 568448 3988 54 2616 1318 -264.85 ± 6.81 638 1287 68 1 0 -699.19 ± 42.05
ncm-dbt-05 06:26:21 583270 3994 47 2596 1351 -262.36 ± 6.68 622 1305 70 0 0 -699.45 ± 41.42
20000 275 13176 6549 -266.4 ± 3.1 3277 6351 368 4 0 -687.04 ± 17.81

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
471918 ncm-dbt-03 585253 2 0 1 1 -189.7 ± 55.98 0 1 0 0 0 -1129.65 ± 376.02
471917 ncm-dbt-01 585169 16 0 10 6 -254.49 ± 90.13 2 6 0 0 0 -1199.83 ± 190.04
471916 ncm-dbt-05 579744 494 6 323 165 -264.41 ± 19.11 78 161 8 0 0 -713.38 ± 146.51
471915 ncm-dbt-04 569390 488 5 325 158 -272.83 ± 19.68 82 156 6 0 0 -761.9 ± 187.95
471914 ncm-dbt-03 585421 500 8 330 162 -265.77 ± 19.69 82 158 10 0 0 -676.04 ± 124.55
471913 ncm-dbt-02 584958 500 10 330 160 -263.41 ± 19.68 81 158 11 0 0 -659.13 ± 116.77
471912 ncm-dbt-01 585801 500 6 341 153 -281.67 ± 20.75 92 151 7 0 0 -739.05 ± 163.27
471911 ncm-dbt-05 584958 500 6 336 158 -275.44 ± 20.59 89 152 9 0 0 -694.7 ± 134.18
471910 ncm-dbt-04 569549 500 9 323 168 -256.44 ± 18.8 75 164 11 0 0 -659.13 ± 116.77
471909 ncm-dbt-03 587367 500 6 328 166 -265.77 ± 19.69 82 158 10 0 0 -676.04 ± 124.55
471908 ncm-dbt-01 584706 500 7 333 160 -270.57 ± 20.28 86 154 10 0 0 -676.04 ± 124.55
471907 ncm-dbt-02 585632 500 5 320 175 -257.59 ± 19.22 76 164 9 1 0 -659.17 ± 131.36
471906 ncm-dbt-05 583866 500 5 312 183 -248.52 ± 17.52 67 173 10 0 0 -676.04 ± 124.55
471905 ncm-dbt-04 568077 500 5 320 175 -257.59 ± 17.54 71 173 6 0 0 -766.17 ± 188.02
471904 ncm-dbt-03 584622 500 6 334 160 -272.99 ± 20.29 87 154 9 0 0 -694.7 ± 134.18
471903 ncm-dbt-01 584790 500 10 337 153 -271.78 ± 21.25 90 147 13 0 0 -629.41 ± 104.8
471902 ncm-dbt-02 587197 500 9 338 153 -274.22 ± 20.72 89 151 10 0 0 -676.04 ± 124.55
471901 ncm-dbt-05 586054 500 10 319 171 -250.76 ± 18.88 73 163 14 0 0 -616.18 ± 100.06
471900 ncm-dbt-04 564919 500 9 336 155 -271.78 ± 20.71 88 151 11 0 0 -659.13 ± 116.77
471899 ncm-dbt-03 587579 500 9 323 168 -256.44 ± 18.25 73 168 9 0 0 -694.7 ± 134.18
471898 ncm-dbt-01 582569 500 4 333 163 -274.22 ± 19.57 85 159 6 0 0 -766.17 ± 188.02
471897 ncm-dbt-02 588132 500 8 328 164 -263.41 ± 19.12 79 162 9 0 0 -694.7 ± 134.18
471896 ncm-dbt-05 584706 500 7 334 159 -271.78 ± 20.71 88 151 11 0 0 -659.13 ± 116.77
471895 ncm-dbt-04 569549 500 7 322 171 -257.59 ± 18.95 76 163 11 0 0 -659.13 ± 116.77
471894 ncm-dbt-03 587155 500 4 321 175 -259.9 ± 19.24 77 164 8 1 0 -676.08 ± 142.43
471893 ncm-dbt-01 585379 500 9 331 160 -265.77 ± 19.97 83 156 11 0 0 -659.13 ± 116.77
471892 ncm-dbt-02 587877 500 12 330 158 -261.07 ± 19.92 81 156 13 0 0 -629.41 ± 104.8
471891 ncm-dbt-05 581527 500 4 320 176 -258.74 ± 17.98 73 170 7 0 0 -739.05 ± 163.27
471890 ncm-dbt-03 583321 500 4 335 161 -276.68 ± 19.56 86 159 5 0 0 -798.18 ± 231.48
471889 ncm-dbt-02 586097 500 12 340 148 -272.99 ± 21.67 92 144 14 0 0 -616.18 ± 100.06
471888 ncm-dbt-01 586562 500 6 335 159 -274.22 ± 21.0 90 149 11 0 0 -659.13 ± 116.77
471887 ncm-dbt-04 568355 500 9 330 161 -264.59 ± 20.09 83 155 12 0 0 -643.67 ± 110.29
471886 ncm-dbt-05 582235 500 6 322 172 -258.74 ± 17.68 72 172 6 0 0 -766.17 ± 188.02
471885 ncm-dbt-03 585295 500 8 330 162 -265.77 ± 19.97 83 156 11 0 0 -659.13 ± 116.77
471884 ncm-dbt-04 570348 500 5 326 169 -264.59 ± 18.7 77 168 4 1 0 -766.23 ± 273.11
471883 ncm-dbt-02 585169 500 5 325 170 -263.41 ± 18.84 78 164 8 0 0 -715.51 ± 146.53
471882 ncm-dbt-01 586350 500 9 324 167 -257.59 ± 18.68 75 165 10 0 0 -676.04 ± 124.55
471881 ncm-dbt-03 584117 500 4 326 170 -265.77 ± 18.84 79 164 7 0 0 -739.05 ± 163.27
471880 ncm-dbt-05 583070 500 3 330 167 -271.78 ± 18.98 82 163 5 0 0 -798.18 ± 231.48
471879 ncm-dbt-01 586858 500 7 338 155 -276.68 ± 21.58 92 148 9 1 0 -659.17 ± 131.36
471878 ncm-dbt-02 586266 500 6 343 151 -284.21 ± 21.93 97 143 10 0 0 -676.04 ± 124.55
471877 ncm-dbt-04 567403 500 5 334 161 -274.22 ± 19.86 86 157 7 0 0 -739.05 ± 163.27

Commit

Commit ID 3a3ca6af0390d74427c218f29cb5fe1a913efb42
Author 31m059
Date 2019-09-27 10:46:47 UTC
Extend castling independently of singular extension A curious feature of Stockfish's current extension code is its repeated use of "else if." In most cases, this makes no functional difference, because no more than one extension is applied; once one extension has been applied, the remaining ones can be safely ignored. However, if most singular extension search conditions are true, except "value < singularBeta", no non-singular extensions (e.g., castling) can be performed! Three tests were submitted, for three of Stockfish's four non-singular extensions. I excluded the shuffle extension, because historically there have been concerns about the fragility of its conditions, and I did not want to risk causing any serious search problems. - Modifying the passed pawn extension appeared roughly neutral at STC. At best, it appeared to be an improvement of less than 1 Elo. - Modifying check extension performed very poorly at STC - Modifying castling extension (this patch) produced a long "yellow" run at STC (insufficient to pass, but positive score) and a strong LTC. In simple terms, prior to this patch castling extension was occasionally not applied during search--on castling moves. The effect of this patch is to perform castling extension on more castling moves. It does so without adding any code complexity, simply by replacing an "else if" with "if" and reordering some existing code. STC: LLR: -2.96 (-2.94,2.94) [0.00,4.00] Total: 108114 W: 23877 L: 23615 D: 60622 http://tests.stockfishchess.org/tests/view/5d8d86bd0ebc590f3beb0c88 LTC: LLR: 2.96 (-2.94,2.94) [0.00,4.00] Total: 20862 W: 3517 L: 3298 D: 14047 http://tests.stockfishchess.org/tests/view/5d8d99cd0ebc590f3beb1899 Bench: 3728191 -------- Where do we go from here? - It seems strange to me that check extension performed so poorly -- clearly some of the singular extension conditions are also very important for check extension. I am not an expert in search, and I do not have any intuition about which of the eight conditions is/are the culprit. I will try a succession of eight STC tests to identify the relevant conditions, then try to replicate this PR for check extension. - Recent tests interacting with the castle extension may deserve retesting. I will shortly resubmit a few of my recent castling extension tweaks, rebased on this PR/commit. My deepest thanks to @noobpwnftw for the extraordinary CPU donation, and to all our other fishtest volunteers, who made it possible for a speculative LTC to pass in 70 minutes! Closes https://github.com/official-stockfish/Stockfish/pull/2331
Copyright 2011–2026 Next Chess Move LLC