Dev Builds » 20210604-1207

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 06:45:18 582777 4000 376 1774 1850 -126.77 ± 5.21 89 1280 572 58 1 -272.39 ± 14.25
ncm-dbt-02 06:43:54 585401 3994 375 1761 1858 -125.79 ± 5.1 91 1247 616 43 0 -268.45 ± 13.71
ncm-dbt-03 06:45:34 585233 3996 372 1730 1894 -122.96 ± 5.21 89 1239 611 59 0 -260.57 ± 13.79
ncm-dbt-04 06:44:15 566695 4010 383 1776 1851 -125.93 ± 5.06 78 1290 584 53 0 -272.92 ± 14.1
ncm-dbt-05 06:45:09 580172 4000 351 1772 1877 -129.05 ± 5.1 92 1282 581 45 0 -278.24 ± 14.13
20000 1857 8813 9330 -126.1 ± 2.3 439 6338 2964 258 1 -270.45 ± 6.25

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
438773 ncm-dbt-04 566336 10 0 4 6 -147.05 ± 74.68 0 4 1 0 0 -381.35 ± 516.06
438772 ncm-dbt-03 585169 496 40 225 231 -136.15 ± 16.2 21 150 70 7 0 -276.25 ± 41.14
438771 ncm-dbt-01 583531 500 55 224 221 -122.24 ± 13.83 9 155 82 4 0 -263.42 ± 37.69
438770 ncm-dbt-05 580282 500 52 211 237 -114.45 ± 14.53 8 152 81 9 0 -243.0 ± 38.14
438769 ncm-dbt-02 586646 494 42 213 239 -125.45 ± 14.59 11 155 75 6 0 -268.03 ± 39.66
438768 ncm-dbt-04 567205 500 49 217 234 -121.45 ± 14.17 5 168 67 10 0 -270.57 ± 41.96
438767 ncm-dbt-01 585421 500 42 208 250 -119.89 ± 15.3 10 158 70 12 0 -254.16 ± 40.96
438766 ncm-dbt-03 583196 500 47 211 242 -118.33 ± 15.31 14 144 84 8 0 -240.82 ± 37.4
438765 ncm-dbt-05 578300 500 36 219 245 -133.34 ± 13.63 10 166 71 3 0 -295.94 ± 40.69
438764 ncm-dbt-02 586266 500 51 228 221 -128.55 ± 14.92 14 155 75 6 0 -270.57 ± 39.67
438763 ncm-dbt-04 567363 500 46 219 235 -125.38 ± 13.96 8 163 73 6 0 -275.45 ± 40.24
438762 ncm-dbt-01 581485 500 50 219 231 -122.24 ± 15.29 13 152 76 9 0 -254.16 ± 39.42
438761 ncm-dbt-02 584076 500 46 218 236 -124.6 ± 15.12 14 151 78 7 0 -258.75 ± 38.88
438760 ncm-dbt-05 582235 500 44 233 223 -138.18 ± 14.06 10 175 59 6 0 -312.48 ± 44.93
438759 ncm-dbt-03 585717 500 47 213 240 -119.89 ± 14.35 8 158 76 8 0 -258.75 ± 39.42
438758 ncm-dbt-04 565154 500 59 223 218 -118.33 ± 14.03 8 154 82 6 0 -254.16 ± 37.82
438757 ncm-dbt-02 586097 500 52 219 229 -120.67 ± 14.99 11 154 76 9 0 -254.16 ± 39.42
438756 ncm-dbt-01 580157 500 55 234 211 -130.14 ± 15.39 15 157 70 8 0 -273.0 ± 41.13
438755 ncm-dbt-03 585548 500 43 204 253 -116.0 ± 14.53 7 157 76 10 0 -249.64 ± 39.41
438754 ncm-dbt-05 580033 500 44 215 241 -123.81 ± 15.28 12 157 71 10 0 -261.07 ± 40.78
438753 ncm-dbt-04 566178 500 38 220 242 -132.54 ± 14.7 13 162 69 6 0 -285.49 ± 41.44
438752 ncm-dbt-02 585928 500 48 225 227 -128.55 ± 13.91 10 161 75 4 0 -280.42 ± 39.57
438751 ncm-dbt-03 586181 500 51 218 231 -120.67 ± 14.99 12 151 79 8 0 -251.89 ± 38.63
438750 ncm-dbt-01 583070 500 44 216 240 -124.6 ± 14.64 8 165 69 7 1 -275.45 ± 41.43
438749 ncm-dbt-05 579289 500 44 226 230 -132.54 ± 14.01 12 161 74 3 0 -288.06 ± 39.79
438748 ncm-dbt-04 569589 500 45 235 220 -138.99 ± 14.21 14 165 68 3 0 -304.07 ± 41.65
438747 ncm-dbt-02 584874 500 47 210 243 -117.55 ± 14.53 8 156 77 9 0 -251.89 ± 39.16
438746 ncm-dbt-04 568991 500 47 218 235 -123.81 ± 14.65 9 162 70 9 0 -268.17 ± 41.1
438745 ncm-dbt-01 583698 500 55 232 213 -128.55 ± 15.56 16 153 73 8 0 -265.78 ± 40.25
438744 ncm-dbt-05 574013 500 40 224 236 -134.15 ± 14.33 12 165 68 5 0 -293.29 ± 41.75
438743 ncm-dbt-03 585169 500 48 212 240 -118.33 ± 14.2 9 152 83 6 0 -251.89 ± 37.57
438742 ncm-dbt-02 584874 500 49 222 229 -125.38 ± 13.61 12 149 89 0 0 -265.78 ± 35.51
438741 ncm-dbt-04 564801 500 57 222 221 -119.11 ± 15.15 12 150 79 9 0 -247.41 ± 38.64
438740 ncm-dbt-01 580655 500 33 217 250 -134.15 ± 13.97 7 178 57 8 0 -306.84 ± 45.58
438739 ncm-dbt-05 584117 500 45 215 240 -123.02 ± 14.65 14 146 86 4 0 -254.16 ± 36.71
438738 ncm-dbt-03 585843 500 48 223 229 -126.97 ± 13.94 8 165 71 6 0 -280.42 ± 40.83
438737 ncm-dbt-02 584453 500 40 226 234 -135.76 ± 13.57 11 166 71 2 0 -301.33 ± 40.6
438736 ncm-dbt-04 564644 500 42 218 240 -127.76 ± 13.75 9 162 75 4 0 -280.42 ± 39.57
438735 ncm-dbt-05 583112 500 46 229 225 -133.34 ± 14.68 14 160 71 5 0 -285.49 ± 40.81
438734 ncm-dbt-03 585042 500 48 224 228 -127.76 ± 14.27 10 162 72 6 0 -277.93 ± 40.53
438733 ncm-dbt-01 584201 500 42 224 234 -132.54 ± 13.65 11 162 75 2 0 -290.66 ± 39.4

Commit

Commit ID 8f081c86f7f8827ea35fc687e6f6591950cc8f90
Author Stéphane Nicolet
Date 2021-06-04 12:07:46 UTC
Clean SIMD code a bit Cleaner vector code structure in feature transformer. This patch just regroups the parts of the inner loop for each SIMD instruction set. Tested for non-regression: LLR: 2.96 (-2.94,2.94) <-2.50,0.50> Total: 115760 W: 9835 L: 9831 D: 96094 Ptnml(0-2): 326, 7776, 41715, 7694, 369 https://tests.stockfishchess.org/tests/view/60b96b39457376eb8bcaa26e It would be nice if a future patch could use some of the macros at the top of the file to unify the code between the distincts SIMD instruction sets (of course, unifying the Relu will be the challenge). closes https://github.com/official-stockfish/Stockfish/pull/3506 No functional change
Copyright 2011–2025 Next Chess Move LLC