Dev Builds » 20210425-1116

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 06:41:19 584393 4000 320 1801 1879 -135.05 ± 5.06 101 1315 548 36 0 -294.61 ± 14.55
ncm-dbt-02 06:41:50 585936 4000 313 1818 1869 -137.47 ± 5.04 98 1347 517 38 0 -303.73 ± 14.99
ncm-dbt-03 06:42:00 583488 4012 319 1808 1885 -135.41 ± 4.99 83 1369 509 44 1 -302.28 ± 15.12
ncm-dbt-04 06:40:26 567594 4000 308 1822 1870 -138.39 ± 5.04 96 1363 501 39 1 -307.89 ± 15.24
ncm-dbt-05 06:41:26 580920 3988 310 1835 1843 -139.97 ± 4.99 98 1363 499 34 0 -312.23 ± 15.26
20000 1570 9084 9346 -137.25 ± 2.25 476 6757 2574 191 2 -304.07 ± 6.71

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
456090 ncm-dbt-03 584453 12 2 6 4 -120.31 ± 75.55 0 4 2 0 0 -279.41 ± 548.26
456089 ncm-dbt-05 582987 488 48 227 213 -133.66 ± 14.22 10 164 65 5 0 -296.36 ± 42.74
456088 ncm-dbt-01 583237 500 41 233 226 -140.62 ± 14.35 13 171 61 5 0 -312.48 ± 44.19
456087 ncm-dbt-02 587494 500 35 225 240 -138.99 ± 14.03 13 167 67 3 0 -306.84 ± 41.99
456086 ncm-dbt-04 569749 500 42 231 227 -138.18 ± 14.41 13 168 64 5 0 -304.07 ± 43.1
456085 ncm-dbt-03 583572 500 44 224 232 -130.94 ± 13.87 10 164 72 4 0 -288.06 ± 40.46
456084 ncm-dbt-05 583070 500 49 230 221 -131.74 ± 14.88 14 159 71 6 0 -280.42 ± 40.83
456083 ncm-dbt-01 584495 500 39 224 237 -134.95 ± 14.13 13 162 72 3 0 -293.29 ± 40.39
456082 ncm-dbt-04 567997 500 40 220 240 -130.94 ± 14.89 13 161 69 7 0 -280.42 ± 41.44
456081 ncm-dbt-02 584076 500 41 226 233 -134.95 ± 14.65 13 165 66 6 0 -293.29 ± 42.41
456080 ncm-dbt-03 584076 500 49 226 225 -128.55 ± 15.08 8 173 58 10 1 -288.06 ± 44.88
456079 ncm-dbt-05 579702 500 30 232 238 -148.85 ± 14.09 13 181 51 5 0 -342.85 ± 48.46
456078 ncm-dbt-04 566691 500 38 225 237 -136.56 ± 14.79 13 168 62 7 0 -298.62 ± 43.77
456077 ncm-dbt-01 585000 500 40 215 245 -126.97 ± 14.78 14 152 79 5 0 -265.78 ± 38.54
456076 ncm-dbt-02 587367 500 41 225 234 -134.15 ± 14.33 15 156 77 2 0 -285.49 ± 38.83
456075 ncm-dbt-03 584201 500 40 220 240 -130.94 ± 13.32 7 170 69 4 0 -295.94 ± 41.39
456074 ncm-dbt-05 580489 500 28 234 238 -152.18 ± 14.34 15 181 49 5 0 -349.43 ± 49.46
456073 ncm-dbt-04 569549 500 37 232 231 -143.07 ± 13.91 12 175 59 4 0 -324.17 ± 44.97
456072 ncm-dbt-01 582569 500 43 224 233 -131.74 ± 14.2 13 158 76 3 0 -282.94 ± 39.21
456071 ncm-dbt-02 583154 500 44 224 232 -130.94 ± 14.22 10 166 68 6 0 -288.06 ± 41.76
456070 ncm-dbt-03 585000 500 35 228 237 -141.44 ± 13.96 11 176 58 5 0 -321.19 ± 45.36
456069 ncm-dbt-05 583321 500 41 225 234 -134.15 ± 14.33 12 165 68 5 0 -293.29 ± 41.75
456068 ncm-dbt-01 584117 500 39 232 229 -141.44 ± 14.33 15 166 66 3 0 -309.64 ± 42.33
456067 ncm-dbt-02 585970 500 40 233 227 -141.44 ± 13.59 10 177 59 4 0 -324.17 ± 44.97
456066 ncm-dbt-04 565469 500 35 221 244 -135.76 ± 14.29 12 167 66 5 0 -298.62 ± 42.41
456065 ncm-dbt-03 584201 500 41 223 236 -132.54 ± 14.87 12 166 64 8 0 -288.06 ± 43.03
456064 ncm-dbt-05 580489 500 35 225 240 -138.99 ± 13.29 10 172 66 2 0 -315.35 ± 42.24
456063 ncm-dbt-01 586943 500 33 222 245 -138.18 ± 14.06 12 169 65 4 0 -306.84 ± 42.73
456062 ncm-dbt-04 566928 500 32 232 236 -147.19 ± 12.98 10 182 56 2 0 -346.12 ± 46.15
456061 ncm-dbt-02 586139 500 35 224 241 -138.18 ± 14.59 14 166 65 5 0 -301.33 ± 42.75
456060 ncm-dbt-03 580406 500 36 226 238 -138.99 ± 14.03 10 176 58 6 0 -315.35 ± 45.33
456059 ncm-dbt-04 567958 500 45 229 226 -134.15 ± 14.84 11 170 62 6 1 -298.62 ± 43.77
456058 ncm-dbt-05 577643 500 31 233 236 -148.85 ± 13.51 13 178 57 2 0 -342.85 ± 45.71
456057 ncm-dbt-01 583363 500 47 220 233 -125.38 ± 14.79 10 162 69 9 0 -270.57 ± 41.4
456056 ncm-dbt-02 586308 500 39 232 229 -141.44 ± 13.96 11 176 58 5 0 -321.19 ± 45.36
456055 ncm-dbt-03 580447 500 35 233 232 -145.54 ± 13.25 12 175 62 1 0 -333.32 ± 43.57
456054 ncm-dbt-05 579661 500 48 229 223 -131.74 ± 14.03 11 163 72 4 0 -288.06 ± 40.46
456053 ncm-dbt-01 585421 500 38 231 231 -141.44 ± 13.78 11 175 60 4 0 -321.19 ± 44.57
456052 ncm-dbt-03 585042 500 37 222 241 -134.95 ± 14.65 13 165 66 6 0 -293.29 ± 42.41
456051 ncm-dbt-02 586985 500 38 229 233 -139.81 ± 14.55 12 174 57 7 0 -312.48 ± 45.66
456050 ncm-dbt-04 566414 500 39 232 229 -141.44 ± 13.78 12 172 63 3 0 -318.25 ± 43.39

Commit

Commit ID b748b46714d5f8e0acca0a042ede1fc95e4f5190
Author Tomasz Sobczyk
Date 2021-04-25 11:16:30 UTC
Cleanup and simplify NNUE code. A lot of optimizations happend since the NNUE was introduced and since then some parts of the code were left unused. This got to the point where asserts were have to be made just to let people know that modifying something will not have any effects or may even break everything due to the assumptions being made. Removing these parts removes those inexisting "false dependencies". Additionally: * append_changed_indices now takes the king pos and stateinfo explicitly, no more misleading pos parameter * IndexList is removed in favor of a generic ValueList. Feature transformer just instantiates the type it needs. * The update cost and refresh requirement is deferred to the feature set once again, but now doesn't go through the whole FeatureSet machinery and just calls HalfKP directly. * accumulator no longer has a singular dimension. * The PS constants and the PieceSquareIndex array are made local to the HalfKP feature set because they are specific to it and DO differ for other feature sets. * A few names are changed to more descriptive Passed STC non-regression: https://tests.stockfishchess.org/tests/view/608421dd95e7f1852abd2790 LLR: 2.95 (-2.94,2.94) <-2.50,0.50> Total: 180008 W: 16186 L: 16258 D: 147564 Ptnml(0-2): 587, 12593, 63725, 12503, 596 closes https://github.com/official-stockfish/Stockfish/pull/3441 No functional change
Copyright 2011–2026 Next Chess Move LLC