Dev Builds » 20210615-2008

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 06:40:18 581585 4004 466 1602 1936 -101.35 ± 5.12 54 1110 757 80 1 -210.38 ± 12.36
ncm-dbt-02 06:40:30 586213 4000 476 1601 1923 -100.42 ± 5.07 43 1126 744 87 0 -210.4 ± 12.48
ncm-dbt-03 06:40:34 585373 4014 492 1606 1916 -99.02 ± 5.08 47 1107 766 87 0 -205.83 ± 12.29
ncm-dbt-04 06:39:36 566790 4000 457 1599 1944 -102.03 ± 5.09 54 1112 756 78 0 -211.87 ± 12.36
ncm-dbt-05 06:39:27 580457 3982 472 1595 1915 -100.71 ± 5.15 51 1109 743 88 0 -209.13 ± 12.48
20000 2363 8003 9634 -100.71 ± 2.28 249 5564 3766 420 1 -209.51 ± 5.54

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
438730 ncm-dbt-01 578013 4 1 0 3 +88.51 ± 137.63 0 0 1 1 0 +190.5 ± 634.21
438729 ncm-dbt-03 584159 14 3 5 6 -49.93 ± 60.15 0 2 5 0 0 -102.04 ± 134.61
438728 ncm-dbt-05 582027 482 55 178 249 -90.66 ± 13.54 1 130 101 9 0 -193.74 ± 33.79
438727 ncm-dbt-02 585928 500 62 199 239 -97.69 ± 14.28 3 144 90 13 0 -207.95 ± 36.13
438726 ncm-dbt-04 559823 500 47 192 261 -103.73 ± 14.19 4 148 87 11 0 -221.9 ± 36.76
438725 ncm-dbt-01 578177 500 64 199 237 -96.19 ± 14.56 7 132 100 11 0 -196.45 ± 34.12
438724 ncm-dbt-03 583740 500 64 199 237 -96.19 ± 14.26 4 139 95 12 0 -202.15 ± 35.11
438723 ncm-dbt-05 578095 500 64 194 242 -92.45 ± 15.22 8 129 98 15 0 -185.33 ± 34.57
438722 ncm-dbt-02 584117 500 59 205 236 -104.49 ± 15.1 6 149 80 15 0 -219.87 ± 38.28
438721 ncm-dbt-04 568315 500 66 202 232 -96.94 ± 14.42 6 135 98 11 0 -200.24 ± 34.5
438720 ncm-dbt-01 581735 500 51 210 239 -114.45 ± 13.55 4 158 81 7 0 -251.89 ± 38.11
438719 ncm-dbt-03 588941 500 58 208 234 -107.54 ± 14.83 7 148 83 12 0 -226.0 ± 37.66
438718 ncm-dbt-05 576251 500 53 202 245 -106.77 ± 15.27 9 144 84 13 0 -219.87 ± 37.42
438717 ncm-dbt-02 585801 500 54 186 260 -93.95 ± 14.38 6 131 102 11 0 -192.71 ± 33.75
438716 ncm-dbt-04 567126 500 43 193 264 -107.54 ± 14.37 8 142 92 8 0 -223.94 ± 35.61
438715 ncm-dbt-01 582193 500 58 188 254 -92.45 ± 13.9 5 129 107 9 0 -190.85 ± 32.75
438714 ncm-dbt-03 585674 500 49 205 246 -112.14 ± 14.85 10 145 86 9 0 -232.26 ± 36.96
438713 ncm-dbt-05 578671 500 69 200 231 -93.2 ± 14.36 6 130 103 11 0 -190.85 ± 33.57
438712 ncm-dbt-02 584579 500 61 198 241 -97.69 ± 13.82 4 138 99 9 0 -206.01 ± 34.24
438711 ncm-dbt-04 567047 500 61 192 247 -93.2 ± 13.91 6 127 109 8 0 -190.85 ± 32.34
438710 ncm-dbt-01 582318 500 55 199 246 -102.97 ± 14.64 7 140 94 8 1 -215.85 ± 35.23
438709 ncm-dbt-03 584664 500 68 192 240 -88.0 ± 15.27 8 124 102 16 0 -174.55 ± 33.86
438708 ncm-dbt-05 584285 500 57 209 234 -109.07 ± 14.05 7 145 91 7 0 -230.16 ± 35.78
438707 ncm-dbt-02 584537 500 58 204 238 -104.49 ± 14.35 6 144 90 10 0 -219.87 ± 36.1
438706 ncm-dbt-04 566414 500 67 195 238 -90.96 ± 14.62 6 129 102 13 0 -185.33 ± 33.81
438705 ncm-dbt-03 584622 500 63 199 238 -96.94 ± 13.96 4 138 98 10 0 -204.07 ± 34.47
438704 ncm-dbt-01 583992 500 61 201 238 -99.95 ± 14.76 8 135 96 11 0 -204.07 ± 34.89
438703 ncm-dbt-05 582152 500 57 212 231 -111.37 ± 15.15 10 146 83 11 0 -230.16 ± 37.67
438702 ncm-dbt-02 589924 500 71 213 216 -101.46 ± 14.48 7 138 95 10 0 -209.91 ± 35.06
438701 ncm-dbt-04 567047 500 56 199 245 -102.22 ± 15.08 9 137 92 12 0 -207.95 ± 35.71
438700 ncm-dbt-03 585759 500 64 197 239 -94.69 ± 14.24 5 134 100 11 0 -196.45 ± 34.12
438699 ncm-dbt-01 583992 500 60 202 238 -101.46 ± 14.78 9 134 97 10 0 -206.01 ± 34.66
438698 ncm-dbt-05 581277 500 54 201 245 -105.25 ± 13.72 4 147 91 8 0 -226.0 ± 35.82
438697 ncm-dbt-02 587197 500 56 205 239 -106.77 ± 13.56 2 154 85 9 0 -234.38 ± 37.19
438696 ncm-dbt-04 567442 500 63 219 218 -112.14 ± 14.38 7 151 83 9 0 -238.66 ± 37.66
438695 ncm-dbt-03 585084 500 62 204 234 -101.46 ± 13.85 3 146 91 10 0 -217.85 ± 35.89
438694 ncm-dbt-01 582193 500 61 191 248 -92.45 ± 14.94 5 136 93 16 0 -190.85 ± 35.52
438693 ncm-dbt-05 580904 500 63 199 238 -96.94 ± 14.87 6 138 92 14 0 -200.24 ± 35.72
438692 ncm-dbt-01 581652 500 55 212 233 -112.91 ± 14.38 9 146 88 7 0 -236.51 ± 36.44
438691 ncm-dbt-04 571110 500 54 207 239 -109.83 ± 14.06 8 143 93 6 0 -230.16 ± 35.29
438690 ncm-dbt-03 585717 500 61 197 242 -96.94 ± 13.81 6 131 106 7 0 -200.24 ± 32.81
438689 ncm-dbt-02 587622 500 55 191 254 -96.94 ± 14.72 9 128 103 10 0 -194.57 ± 33.53

Commit

Commit ID 8ec9e108664ce38fa98ccfb69f048d7d804f99f9
Author Joost VandeVondele
Date 2021-06-15 20:08:40 UTC
New default net nn-33c9d39e5eb6.nnue As the previous net, this net is trained on Leela games as provided by borg. See also https://lczero.org/blog/2021/06/the-importance-of-open-data/ The particular data set, which is a mix of T60 and T74 data, is now available as a single binpack: https://drive.google.com/file/d/1RFkQES3DpsiJqsOtUshENtzPfFgUmEff/view?usp=sharing The training command was: python train.py ../../training_data_pylon.binpack ../../training_data_pylon.binpack --gpus 1 --threads 2 --num-workers 2 --batch-size 16384 --progress_bar_refresh_rate 300 --smart-fen-skipping --random-fen-skipping 10 --features=HalfKAv2^ --lambda=1.0 --max_epochs=440 --seed $RANDOM --default_root_dir exp/run_2 passed STC: https://tests.stockfishchess.org/tests/view/60c887cb457376eb8bcab054 LLR: 2.94 (-2.94,2.94) <-0.50,2.50> Total: 12792 W: 1483 L: 1311 D: 9998 Ptnml(0-2): 62, 989, 4131, 1143, 71 passed LTC: https://tests.stockfishchess.org/tests/view/60c8e5c4457376eb8bcab0f0 LLR: 2.95 (-2.94,2.94) <0.50,3.50> Total: 11272 W: 601 L: 477 D: 10194 Ptnml(0-2): 9, 421, 4657, 535, 14 also had strong LTC performance against another strong net of the series: https://tests.stockfishchess.org/tests/view/60c8c40d457376eb8bcab0c6 closes https://github.com/official-stockfish/Stockfish/pull/3557 Bench: 5032320
Copyright 2011–2025 Next Chess Move LLC