Dev Builds » 20210415-1045

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 06:42:10 581972 4000 312 1830 1858 -138.79 ± 4.98 107 1328 541 24 0 -305.11 ± 14.62
ncm-dbt-02 06:41:37 585630 4000 295 1815 1890 -138.99 ± 4.91 100 1342 536 22 0 -308.24 ± 14.68
ncm-dbt-03 06:40:42 586172 4000 296 1808 1896 -138.18 ± 5.01 87 1385 481 47 0 -310.0 ± 15.55
ncm-dbt-04 06:41:58 566950 4000 314 1804 1882 -135.96 ± 5.01 86 1362 509 42 1 -303.04 ± 15.11
ncm-dbt-05 06:42:34 580915 4000 296 1825 1879 -139.91 ± 4.91 84 1400 477 39 0 -317.16 ± 15.62
20000 1513 9082 9405 -138.37 ± 2.22 464 6817 2544 174 1 -308.66 ± 6.75

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
437692 ncm-dbt-03 586435 500 41 233 226 -140.62 ± 14.53 11 178 53 8 0 -318.25 ± 47.23
437691 ncm-dbt-05 581319 500 38 236 226 -145.54 ± 14.02 13 176 57 4 0 -330.23 ± 45.79
437690 ncm-dbt-02 586646 500 28 231 241 -149.68 ± 14.24 16 174 57 3 0 -336.46 ± 45.77
437689 ncm-dbt-04 565390 500 46 219 235 -125.38 ± 13.79 6 168 69 7 0 -280.42 ± 41.44
437688 ncm-dbt-01 583656 500 33 224 243 -139.81 ± 14.01 15 162 72 1 0 -304.07 ± 40.16
437687 ncm-dbt-03 587834 500 28 221 251 -141.44 ± 13.21 7 184 54 5 0 -333.32 ± 47.06
437686 ncm-dbt-02 584201 500 43 224 233 -131.74 ± 14.37 14 156 77 3 0 -280.42 ± 38.93
437685 ncm-dbt-05 580406 500 43 230 227 -136.56 ± 14.45 12 169 63 6 0 -301.33 ± 43.45
437684 ncm-dbt-04 567205 500 36 228 236 -140.62 ± 14.7 12 176 54 8 0 -315.35 ± 46.8
437683 ncm-dbt-01 581652 500 32 230 238 -145.54 ± 14.2 14 174 58 4 0 -327.18 ± 45.37
437682 ncm-dbt-03 586689 500 40 228 232 -137.37 ± 15.93 18 161 62 9 0 -288.06 ± 43.65
437681 ncm-dbt-05 580572 500 39 229 232 -138.99 ± 14.39 14 166 66 4 0 -304.07 ± 42.38
437680 ncm-dbt-02 589069 500 40 234 226 -142.25 ± 14.66 16 166 64 4 0 -309.64 ± 43.08
437679 ncm-dbt-04 568315 500 41 224 235 -133.34 ± 14.51 13 162 70 5 0 -288.06 ± 41.11
437678 ncm-dbt-01 581277 500 43 231 226 -137.37 ± 13.34 8 176 62 4 0 -315.35 ± 43.81
437677 ncm-dbt-03 586097 500 31 231 238 -147.19 ± 13.96 12 181 52 5 0 -339.63 ± 47.98
437676 ncm-dbt-02 584874 500 47 236 217 -138.18 ± 14.23 15 161 72 2 0 -298.62 ± 40.29
437675 ncm-dbt-04 566849 500 46 232 222 -135.76 ± 13.94 11 168 67 4 0 -301.33 ± 42.04
437674 ncm-dbt-05 578794 500 34 224 242 -138.99 ± 13.85 8 181 54 7 0 -321.19 ± 46.91
437673 ncm-dbt-01 581693 500 40 225 235 -134.95 ± 13.78 11 166 70 3 0 -298.62 ± 41.01
437672 ncm-dbt-03 585759 500 36 230 234 -142.25 ± 13.75 11 176 59 4 0 -324.17 ± 44.97
437671 ncm-dbt-04 568037 500 31 218 251 -136.56 ± 14.79 12 171 59 8 0 -301.33 ± 44.81
437670 ncm-dbt-02 584453 500 34 218 248 -134.15 ± 14.15 11 167 67 5 0 -295.94 ± 42.08
437669 ncm-dbt-05 581194 500 35 224 241 -138.18 ± 14.59 13 169 62 6 0 -304.07 ± 43.81
437668 ncm-dbt-01 580199 500 39 235 226 -143.89 ± 14.97 16 170 58 6 0 -315.35 ± 45.33
437667 ncm-dbt-03 584537 500 48 211 241 -117.55 ± 14.53 10 150 83 7 0 -247.41 ± 37.61
437666 ncm-dbt-04 568196 500 37 215 248 -129.35 ± 13.72 8 167 70 5 0 -288.06 ± 41.11
437665 ncm-dbt-02 585717 500 45 225 230 -130.94 ± 13.87 11 161 75 3 0 -285.49 ± 39.5
437664 ncm-dbt-01 583866 500 32 225 243 -141.44 ± 13.59 13 168 68 1 0 -315.35 ± 41.44
437663 ncm-dbt-05 582318 500 30 225 245 -143.07 ± 12.95 6 188 51 5 0 -342.85 ± 48.46
437662 ncm-dbt-03 586985 500 39 226 235 -136.56 ± 13.36 7 178 60 5 0 -315.35 ± 44.57
437661 ncm-dbt-01 580282 500 51 231 218 -130.94 ± 14.72 15 154 77 4 0 -275.45 ± 39.01
437660 ncm-dbt-02 585506 500 23 226 251 -149.68 ± 12.23 8 188 53 1 0 -363.2 ± 47.43
437659 ncm-dbt-04 566454 500 34 235 231 -148.02 ± 12.53 9 184 56 1 0 -352.8 ± 46.04
437658 ncm-dbt-05 581153 500 36 236 228 -147.19 ± 13.38 11 181 55 3 0 -342.85 ± 46.64
437657 ncm-dbt-03 585042 500 33 228 239 -143.07 ± 13.73 11 177 58 4 0 -327.18 ± 45.37
437656 ncm-dbt-02 584579 500 35 221 244 -135.76 ± 13.01 9 169 71 1 0 -306.84 ± 40.47
437655 ncm-dbt-04 565154 500 43 233 224 -138.99 ± 15.08 15 166 64 4 1 -304.07 ± 43.1
437654 ncm-dbt-05 581569 500 41 221 238 -130.94 ± 13.32 7 170 69 4 0 -295.94 ± 41.39
437653 ncm-dbt-01 583154 500 42 229 229 -136.56 ± 14.1 15 158 76 1 0 -293.29 ± 38.98

Commit

Commit ID a7ab92ec25c91e8413630c52cfc2db6b4ecacf0c
Author dsmsgms
Date 2021-04-15 10:45:39 UTC
Use classical eval for Bishop vs Pawns NNUE evaluation is incapable of recognizing trivially drawn bishop endgames (the wrong-colored rook pawn), which are in fact ubiquitous and stock standard in chess analysis. Switching off NNUE evaluation in KBPs vs KPs endgames is a measure that stops Stockfish from trading down to a drawn version of these endings when we presumably have advantage. The patch is able to edge over master in endgame positions. Patch tested for Elo gain with the "endgame.epd" book, and verified for non-regression with our usual book (see the pull request for details). STC: LLR: 2.93 (-2.94,2.94) {-0.20,1.10} Total: 33232 W: 6655 L: 6497 D: 20080 Ptnml(0-2): 4, 2342, 11769, 2494, 7 https://tests.stockfishchess.org/tests/view/6074a52981417533789605b8 LTC: LLR: 2.93 (-2.94,2.94) {0.20,0.90} Total: 159056 W: 29799 L: 29378 D: 99879 Ptnml(0-2): 7, 9004, 61085, 9425, 7 https://tests.stockfishchess.org/tests/view/6074c39a81417533789605ca Closes https://github.com/official-stockfish/Stockfish/pull/3427 Bench: 4503918 blah
Copyright 2011–2025 Next Chess Move LLC