Dev Builds » 20210319-1058

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 06:43:53 583741 4012 321 1794 1897 -133.81 ± 4.93 75 1368 518 45 0 -299.23 ± 14.98
ncm-dbt-02 06:42:34 586448 3988 304 1837 1847 -140.79 ± 5.09 105 1364 484 41 0 -312.59 ± 15.5
ncm-dbt-03 06:42:02 584941 4000 300 1817 1883 -138.69 ± 5.0 108 1326 541 25 0 -304.42 ± 14.62
ncm-dbt-04 06:43:41 565976 4000 298 1829 1873 -140.11 ± 5.08 105 1361 494 40 0 -310.35 ± 15.34
ncm-dbt-05 06:43:57 581814 4000 312 1792 1896 -134.95 ± 4.75 69 1371 531 29 0 -305.11 ± 14.77
20000 1535 9069 9396 -137.66 ± 2.22 462 6790 2568 180 0 -306.29 ± 6.72

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
456389 ncm-dbt-01 584243 12 2 6 4 -120.31 ± 75.55 0 4 2 0 0 -279.41 ± 548.26
456388 ncm-dbt-02 586731 488 37 238 213 -152.13 ± 15.55 22 160 59 3 0 -325.37 ± 44.92
456387 ncm-dbt-05 579578 500 36 221 243 -134.95 ± 12.44 6 174 69 1 0 -312.48 ± 41.11
456386 ncm-dbt-04 561835 500 39 235 226 -143.89 ± 14.07 14 171 62 3 0 -321.19 ± 43.77
456385 ncm-dbt-03 584117 500 35 232 233 -144.71 ± 14.41 15 171 60 4 0 -321.19 ± 44.57
456384 ncm-dbt-01 584243 500 39 228 233 -138.18 ± 14.06 12 169 65 4 0 -306.84 ± 42.73
456383 ncm-dbt-02 585590 500 29 235 236 -152.18 ± 13.16 11 187 49 3 0 -363.2 ± 49.57
456382 ncm-dbt-05 581943 500 35 231 234 -143.89 ± 13.31 10 179 58 3 0 -333.32 ± 45.35
456381 ncm-dbt-03 584285 500 38 242 220 -150.51 ± 14.59 19 168 61 2 0 -330.23 ± 44.08
456380 ncm-dbt-04 565548 500 39 227 234 -137.37 ± 14.78 11 175 55 9 0 -306.84 ± 46.27
456379 ncm-dbt-01 583614 500 33 219 248 -135.76 ± 13.2 9 170 69 2 0 -306.84 ± 41.23
456378 ncm-dbt-02 584874 500 36 238 226 -148.85 ± 14.46 12 186 44 8 0 -346.12 ± 51.63
456377 ncm-dbt-05 580157 500 43 225 232 -132.54 ± 13.65 9 168 69 4 0 -295.94 ± 41.39
456376 ncm-dbt-04 569749 500 30 230 240 -147.19 ± 12.98 12 176 62 0 0 -339.63 ± 43.41
456375 ncm-dbt-03 582402 500 28 222 250 -142.25 ± 13.56 11 175 61 3 0 -324.17 ± 44.15
456374 ncm-dbt-01 584832 500 49 225 226 -127.76 ± 14.44 8 169 64 9 0 -282.94 ± 42.98
456373 ncm-dbt-02 585843 500 42 223 235 -131.74 ± 14.2 10 167 67 6 0 -290.66 ± 42.08
456372 ncm-dbt-04 568236 500 27 237 236 -155.54 ± 14.0 14 187 44 5 0 -366.78 ± 52.23
456371 ncm-dbt-03 586689 500 43 228 229 -134.95 ± 13.41 10 167 71 2 0 -301.33 ± 40.6
456370 ncm-dbt-05 582360 500 38 217 245 -130.14 ± 14.74 13 159 72 6 0 -277.93 ± 40.53
456369 ncm-dbt-01 582485 500 37 226 237 -138.18 ± 12.93 7 178 62 3 0 -321.19 ± 43.77
456368 ncm-dbt-02 587579 500 36 228 236 -140.62 ± 15.38 18 162 64 6 0 -298.62 ± 43.1
456367 ncm-dbt-04 568156 500 41 213 246 -124.6 ± 14.47 10 159 74 7 0 -268.17 ± 39.97
456366 ncm-dbt-05 583321 500 39 222 239 -133.34 ± 13.26 6 176 63 5 0 -306.84 ± 43.45
456365 ncm-dbt-03 586943 500 42 214 244 -124.6 ± 14.14 9 160 75 6 0 -270.57 ± 39.67
456364 ncm-dbt-01 585126 500 29 215 256 -135.76 ± 13.76 10 170 66 4 0 -304.07 ± 42.38
456363 ncm-dbt-02 584958 500 39 225 236 -135.76 ± 13.57 8 175 62 5 0 -309.64 ± 43.82
456362 ncm-dbt-04 564409 500 34 228 238 -142.25 ± 14.48 16 165 66 3 0 -309.64 ± 42.33
456361 ncm-dbt-03 583237 500 40 232 228 -140.62 ± 15.05 19 157 71 3 0 -295.94 ± 40.69
456360 ncm-dbt-05 579001 500 42 231 227 -138.18 ± 13.87 11 171 64 4 0 -309.64 ± 43.08
456359 ncm-dbt-01 583656 500 41 231 228 -138.99 ± 15.25 15 168 59 8 0 -301.33 ± 44.81
456358 ncm-dbt-02 588217 500 43 228 229 -134.95 ± 14.31 12 166 67 5 0 -295.94 ± 42.08
456357 ncm-dbt-04 564448 500 38 225 237 -136.56 ± 14.62 13 167 64 6 0 -298.62 ± 43.1
456356 ncm-dbt-05 583279 500 46 223 231 -128.55 ± 13.38 7 167 72 4 0 -288.06 ± 40.46
456355 ncm-dbt-03 585717 500 36 236 228 -147.19 ± 14.51 16 172 58 4 0 -327.18 ± 45.37
456354 ncm-dbt-01 582402 500 47 224 229 -128.55 ± 14.26 8 169 65 8 0 -285.49 ± 42.7
456353 ncm-dbt-02 587792 500 42 222 236 -130.94 ± 14.39 12 161 72 5 0 -282.94 ± 40.5
456352 ncm-dbt-03 586139 500 38 211 251 -125.38 ± 13.25 9 156 84 1 0 -273.0 ± 36.86
456351 ncm-dbt-04 565430 500 50 234 216 -134.15 ± 15.17 15 161 67 7 0 -285.49 ± 42.08
456350 ncm-dbt-05 584874 500 33 222 245 -138.18 ± 12.73 7 177 64 2 0 -321.19 ± 42.95
456349 ncm-dbt-01 583070 500 44 220 236 -127.76 ± 13.75 6 171 66 7 0 -288.06 ± 42.4

Commit

Commit ID ace9632c6713e346ccca73b3e2edc046c1f5527c
Author Vizvezdenec
Date 2021-03-19 10:58:17 UTC
Add a specific FRC correction from classical to NNUE our net currently is not trained on FRC games, and so doesn't know about the important pattern of a bishop that is cornered in FRC. This patch introduces a term we have in the classical evaluation for this case, and adds it to the NNUE eval. Since fishtest doesn't support FRC right now, the patch was tested locally at STC conditions, starting from the book of FRC starting positions. Score of master vs patch: 993 - 2226 - 6781 [0.438] 10000 Which corresponds to approximately 40 Elo The patch passes non-regression testing for traditional chess (where it adds one branch). passed STC: https://tests.stockfishchess.org/tests/view/604fa2532433018de7a38b67 LLR: 2.95 (-2.94,2.94) {-1.25,0.25} Total: 30560 W: 2701 L: 2636 D: 25223 Ptnml(0-2): 88, 2056, 10921, 2133, 82 passed STC also in an earlier version: https://tests.stockfishchess.org/tests/view/604f61282433018de7a38b4d closes https://github.com/official-stockfish/Stockfish/pull/3398 No functional change
Copyright 2011–2026 Next Chess Move LLC