Dev Builds » 20230701-1059

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 06:55:46 582576 4000 1383 689 1928 +60.9 ± 5.07 0 183 948 861 8 +124.2 ± 11.03
ncm-dbt-02 06:55:31 586316 4000 1342 683 1975 +57.77 ± 5.0 1 175 998 816 10 +117.16 ± 10.71
ncm-dbt-03 06:52:51 585276 4000 1353 724 1923 +55.09 ± 5.05 2 185 1007 794 12 +111.18 ± 10.66
ncm-dbt-04 06:55:42 570363 4016 1396 684 1936 +62.25 ± 5.09 4 171 952 871 10 +127.6 ± 11.0
ncm-dbt-05 06:54:39 584731 3984 1330 693 1961 +56.03 ± 5.11 2 197 960 828 5 +114.56 ± 10.96
20000 6804 3473 9723 +58.41 ± 2.27 9 911 4865 4170 45 +118.92 ± 4.86

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
408333 ncm-dbt-04 570909 16 8 4 4 +88.68 ± 65.24 0 0 4 4 0 +190.76 ± 189.19 ↓
408332 ncm-dbt-05 585042 484 166 77 241 +64.62 ± 14.74 1 20 110 111 0 +135.72 ± 32.55 ↓
408331 ncm-dbt-02 583028 500 173 82 245 +63.94 ± 13.8 0 17 127 104 2 +129.35 ± 29.91 ↓
408330 ncm-dbt-01 580199 500 182 94 224 +61.79 ± 13.99 0 21 120 109 0 +127.76 ± 31.05 ↓
408329 ncm-dbt-03 586223 500 169 84 247 +59.64 ± 13.47 0 18 129 103 0 +123.02 ± 29.65 ↓
408328 ncm-dbt-04 572114 500 174 90 236 +58.93 ± 15.35 2 26 108 114 0 +124.6 ± 32.89 ↓
408327 ncm-dbt-05 585632 500 163 91 246 +50.38 ± 14.94 0 31 117 101 1 +101.46 ± 31.58 ↓
408326 ncm-dbt-02 586774 500 159 96 245 +44.01 ± 14.6 0 30 129 89 2 +86.52 ± 29.92 ↓
408325 ncm-dbt-01 582527 500 181 74 245 +75.52 ± 14.08 0 16 113 119 2 +155.54 ± 32.01 ↓
408324 ncm-dbt-03 586181 500 170 94 236 +53.22 ± 14.15 0 22 133 92 3 +104.49 ± 29.21 ↓
408323 ncm-dbt-04 569669 500 173 88 239 +59.64 ± 14.86 1 22 121 103 3 +119.89 ± 30.94 ↓
408322 ncm-dbt-05 584958 500 152 91 257 +42.6 ± 13.98 0 27 136 86 1 +85.04 ± 28.92 ↓
408321 ncm-dbt-01 582819 500 172 90 238 +57.5 ± 15.67 0 32 107 108 3 +113.68 ± 33.0 ↓
408320 ncm-dbt-02 586689 500 167 89 244 +54.65 ± 14.22 1 22 125 102 0 +113.68 ± 30.37 ↓
408319 ncm-dbt-03 585506 500 167 84 249 +58.21 ± 13.69 1 16 133 99 1 +119.89 ± 29.04 ↓
408318 ncm-dbt-04 569390 500 177 76 247 +71.16 ± 13.64 0 15 120 114 1 +147.19 ± 30.88 ↓
408317 ncm-dbt-05 583489 500 159 89 252 +48.96 ± 14.07 0 26 128 96 0 +99.95 ± 30.0 ↓
408316 ncm-dbt-02 587113 500 170 87 243 +58.21 ± 13.4 0 18 131 101 0 +119.89 ± 29.37 ↓
408315 ncm-dbt-01 583196 500 177 96 227 +56.78 ± 14.46 0 26 117 107 0 +116.78 ± 31.56 ↓
408314 ncm-dbt-03 585211 500 165 84 251 +56.78 ± 14.6 0 25 121 102 2 +113.68 ± 30.97 ↓
408313 ncm-dbt-04 570909 500 172 80 248 +64.66 ± 14.26 0 20 120 108 2 +130.94 ± 31.03 ↓
408312 ncm-dbt-05 586731 500 173 90 237 +58.21 ± 14.39 0 24 120 105 1 +118.33 ± 31.1 ↓
408311 ncm-dbt-02 586985 500 163 78 259 +59.64 ± 13.76 0 20 125 105 0 +123.02 ± 30.3 ↓
408310 ncm-dbt-01 583196 500 165 81 254 +58.93 ± 13.87 0 21 124 105 0 +121.46 ± 30.47 ↓
408309 ncm-dbt-03 582987 500 174 94 232 +56.07 ± 14.15 0 23 125 101 1 +113.68 ± 30.37 ↓
408308 ncm-dbt-04 570308 500 181 86 233 +66.82 ± 14.75 0 24 108 117 1 +137.37 ± 32.9 ↓
408307 ncm-dbt-05 578918 500 174 92 234 +57.5 ± 14.22 1 21 123 105 0 +119.89 ± 30.63 ↓
408306 ncm-dbt-02 587325 500 170 85 245 +59.64 ± 14.18 0 22 122 105 1 +121.46 ± 30.78 ↓
408305 ncm-dbt-03 586774 500 166 92 242 +51.8 ± 14.88 0 31 114 105 0 +106.01 ± 32.0 ↓
408304 ncm-dbt-01 582027 500 173 83 244 +63.23 ± 13.63 0 17 127 105 1 +129.35 ± 29.91 ↓
408303 ncm-dbt-04 566651 500 169 85 246 +58.93 ± 14.69 0 25 118 105 2 +118.33 ± 31.4 ↓
408302 ncm-dbt-05 587070 500 171 75 254 +67.54 ± 14.51 0 21 114 113 2 +137.37 ± 31.96 ↓
408301 ncm-dbt-03 585801 500 172 114 214 +40.48 ± 15.42 1 34 124 88 3 +79.17 ± 30.64 ↓
408300 ncm-dbt-02 586646 500 169 85 246 +58.93 ± 15.22 0 27 116 103 4 +115.23 ± 31.71 ↓
408299 ncm-dbt-01 583908 500 164 87 249 +53.93 ± 15.12 0 32 109 109 0 +110.6 ± 32.71 ↓
408298 ncm-dbt-04 569749 500 164 78 258 +60.36 ± 13.93 1 18 125 106 0 +126.18 ± 30.27 ↓
408297 ncm-dbt-05 586012 500 172 88 240 +58.93 ± 14.69 0 27 112 111 0 +121.46 ± 32.29 ↓
408296 ncm-dbt-03 583531 500 170 78 252 +64.66 ± 13.69 0 16 128 104 2 +130.94 ± 29.72 ↓
408295 ncm-dbt-02 585970 500 171 81 248 +63.23 ± 13.91 0 19 123 107 1 +129.35 ± 30.56 ↓
408294 ncm-dbt-04 573568 500 178 97 225 +56.78 ± 13.91 0 21 128 100 1 +115.23 ± 29.89 ↓
408293 ncm-dbt-01 582736 500 169 84 247 +59.64 ± 13.76 0 18 131 99 2 +119.89 ± 29.37 ↓

Commit

Commit ID ef94f77f8c827a2395f1c40f53311a3b1f20bc5b
Author Daniel Monroe
Date 2023-07-01 10:59:28 UTC
Update default net to nn-a3d1bfca1672.nnue faster permutation of master net weights Activation data taken from https://drive.google.com/drive/folders/1Ec9YuuRx4N03GPnVPoQOW70eucOKngQe?usp=sharing Permutation found using https://github.com/Ergodice/nnue-pytorch/blob/836387a0e5e690431d404158c46648710f13904d/ftperm.py See also https://github.com/glinscott/nnue-pytorch/pull/254 The algorithm greedily selects 2- and 3-cycles that can be permuted to increase the number of runs of zeroes. The percent of zero runs from the master net increased from 68.46 to 70.11 from 2-cycles and only increased to 70.32 when considering 3-cycles. Interestingly, allowing both halves of L1 to intermix when creating zero runs can give another 0.5% zero-run density increase with this method. Measured speedup: ``` CPU: 16 x AMD Ryzen 9 3950X 16-Core Processor Result of 50 runs base (./stockfish.master ) = 1561556 +/- 5439 test (./stockfish.patch ) = 1575788 +/- 5427 diff = +14231 +/- 2636 speedup = +0.0091 P(speedup > 0) = 1.0000 ``` closes https://github.com/official-stockfish/Stockfish/pull/4640 No functional change
Copyright 2011–2025 Next Chess Move LLC
Feedback & Support Terms of Service Privacy Policy Desktop Version