Dev Builds » 20230223-1225

You are viewing an old NCM Stockfish dev build test. You may find the most recent dev build tests using Stockfish 15 as the baseline here.

Use this dev build

NCM plays each Stockfish dev build 20,000 times against Stockfish 14. 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 05:01:49 1201370 1664 687 157 820 +114.65 ± 7.92 0 33 258 519 22 +246.62 ± 21.25
ncm-dbt-02 04:57:42 1236231 1656 700 176 780 +113.84 ± 7.68 0 28 265 518 17 +247.59 ± 20.96
ncm-dbt-03 05:01:38 1233766 1670 681 175 814 +108.68 ± 7.72 0 29 290 497 19 +231.86 ± 20.01
ncm-dbt-04 05:01:03 1224536 1670 702 175 793 +113.51 ± 7.95 0 29 278 500 28 +239.53 ± 20.46
ncm-dbt-05 04:59:54 1234533 1670 671 160 839 +109.83 ± 7.67 0 27 289 500 19 +235.03 ± 20.04
ncm-dbt-06 05:01:44 1221221 1670 677 156 837 +112.13 ± 7.75 1 30 267 521 16 +244.09 ± 20.89
ncm-et-3 06:24:09 1299568 1660 679 162 819 +111.92 ± 7.83 1 31 265 516 17 +242.79 ± 20.97
ncm-et-4 06:24:18 1309780 1672 683 168 821 +110.61 ± 7.97 0 37 268 510 21 +235.93 ± 20.85
ncm-et-9 06:24:35 1298404 1656 674 176 806 +107.81 ± 7.73 0 32 281 500 15 +231.91 ± 20.35
ncm-et-10 06:23:46 1288025 1670 704 165 801 +116.29 ± 7.95 1 28 263 517 26 +249.4 ± 21.05
ncm-et-13 06:24:16 1303349 1678 695 191 792 +107.67 ± 7.77 1 28 296 494 20 +229.15 ± 19.8
ncm-et-15 06:24:17 1302029 1664 693 168 803 +113.49 ± 7.95 0 31 270 506 25 +241.35 ± 20.77
20000 8246 2029 9725 +111.7 ± 2.26 4 363 3290 6098 245 +239.52 ± 5.94

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
184502 ncm-dbt-02 1246589 156 71 15 70 +130.53 ± 24.5 0 2 20 54 2 +296.15 ± 79.37
184501 ncm-dbt-05 1245552 170 65 12 93 +112.04 ± 27.41 0 4 29 47 5 +222.26 ± 64.66
184500 ncm-dbt-01 1216067 164 68 11 85 +126.0 ± 25.66 0 2 25 51 4 +267.17 ± 70.11
184499 ncm-dbt-06 1236242 170 66 15 89 +107.53 ± 27.35 0 7 22 54 2 +228.32 ± 72.81
184498 ncm-dbt-04 1217251 170 72 24 74 +100.83 ± 25.77 0 4 32 46 3 +204.75 ± 61.28
184497 ncm-dbt-03 1238476 170 68 18 84 +105.29 ± 28.7 0 8 22 52 3 +216.32 ± 72.15
184496 ncm-dbt-02 1216358 500 201 51 248 +107.54 ± 13.89 0 8 89 148 5 +230.16 ± 36.26
184495 ncm-dbt-04 1217245 500 207 54 239 +109.83 ± 14.53 0 13 75 158 4 +238.66 ± 39.57
184494 ncm-dbt-03 1230075 500 194 50 256 +102.97 ± 13.87 0 8 95 142 5 +217.85 ± 34.98
184493 ncm-dbt-05 1227942 500 201 54 245 +105.25 ± 13.22 0 5 97 144 4 +226.0 ± 34.38
184492 ncm-dbt-06 1224885 500 211 51 238 +115.22 ± 14.21 0 8 81 154 7 +247.41 ± 38.13
184491 ncm-dbt-01 1215645 500 210 52 238 +113.68 ± 15.3 0 12 78 150 10 +236.51 ± 38.85
184490 ncm-dbt-02 1239117 500 219 52 229 +120.67 ± 14.01 0 8 73 163 6 +265.78 ± 40.25
184489 ncm-dbt-04 1213562 500 216 45 239 +123.81 ± 13.98 0 5 78 158 9 +268.17 ± 38.8
184488 ncm-dbt-03 1237992 500 209 56 235 +109.83 ± 14.37 0 7 92 142 9 +228.08 ± 35.56
184487 ncm-dbt-05 1226969 500 206 52 242 +110.6 ± 14.69 0 12 78 154 6 +236.51 ± 38.85
184486 ncm-dbt-06 1206103 500 202 42 256 +115.23 ± 14.21 1 6 81 156 6 +251.89 ± 38.11
184485 ncm-dbt-01 1182890 500 204 42 254 +116.77 ± 14.04 0 10 72 164 4 +258.75 ± 40.49
184484 ncm-dbt-02 1242862 500 209 58 233 +108.3 ± 14.05 0 10 83 153 4 +234.38 ± 37.67
184483 ncm-dbt-05 1237670 500 199 42 259 +112.91 ± 13.39 0 6 85 155 4 +247.41 ± 37.09
184482 ncm-dbt-04 1250087 500 207 52 241 +111.37 ± 14.84 0 7 93 138 12 +226.0 ± 35.35
184481 ncm-dbt-03 1228522 500 210 51 239 +114.45 ± 13.03 0 6 81 161 2 +256.44 ± 38.07
184480 ncm-dbt-01 1190878 500 205 52 243 +109.83 ± 13.89 0 9 83 154 4 +238.66 ± 37.66
184479 ncm-dbt-06 1217655 500 198 48 254 +107.54 ± 13.4 0 9 83 157 1 +238.66 ± 37.66
167018 ncm-et-3 1300347 160 62 20 78 +93.39 ± 28.23 1 5 26 47 1 +202.63 ± 67.98
167017 ncm-et-9 1285314 156 64 21 71 +98.3 ± 26.49 0 5 26 46 1 +209.15 ± 68.26
167016 ncm-et-15 1310583 164 72 21 71 +111.74 ± 23.16 0 1 31 48 2 +239.5 ± 61.5
167015 ncm-et-10 1285055 170 68 16 86 +109.78 ± 26.65 0 5 26 51 3 +228.32 ± 68.26
167014 ncm-et-4 1322321 172 70 17 85 +110.65 ± 24.15 0 3 29 52 2 +237.06 ± 64.68
167013 ncm-et-13 1312634 178 82 14 82 +139.81 ± 22.91 0 0 26 58 5 +306.73 ± 67.89
167012 ncm-et-3 1291676 500 195 48 257 +105.25 ± 14.35 0 9 92 142 7 +219.87 ± 35.65
167011 ncm-et-9 1297763 500 201 54 245 +105.25 ± 14.2 0 10 88 147 5 +223.94 ± 36.53
167010 ncm-et-15 1300566 500 200 44 256 +112.14 ± 14.22 0 9 82 153 6 +240.82 ± 37.9
167009 ncm-et-10 1286213 500 214 47 239 +120.67 ± 15.3 0 11 72 156 11 +254.16 ± 40.45
167008 ncm-et-4 1305982 500 208 52 240 +112.14 ± 13.89 0 7 86 151 6 +240.82 ± 36.9
167007 ncm-et-13 1275072 500 210 65 225 +103.73 ± 15.1 0 11 93 136 10 +209.91 ± 35.49
167006 ncm-et-3 1295653 500 215 55 230 +115.22 ± 14.21 0 12 69 166 3 +256.44 ± 41.25
167005 ncm-et-10 1296919 500 208 55 237 +109.83 ± 13.23 0 6 88 153 3 +240.82 ± 36.39
167004 ncm-et-15 1285638 500 200 47 253 +109.83 ± 14.84 0 12 80 151 7 +232.26 ± 38.36
167003 ncm-et-9 1308305 500 203 58 239 +103.73 ± 13.87 0 9 91 146 4 +221.9 ± 35.86
167002 ncm-et-4 1304549 500 206 52 242 +110.6 ± 14.99 0 12 80 150 8 +232.26 ± 38.36
167001 ncm-et-13 1309775 500 195 53 252 +101.46 ± 13.53 0 8 95 144 3 +217.85 ± 34.98
167000 ncm-et-10 1283914 500 214 47 239 +120.67 ± 14.67 1 6 77 157 9 +261.07 ± 39.14
166999 ncm-et-15 1311330 500 221 56 223 +119.11 ± 14.84 0 9 77 154 10 +251.89 ± 39.16
166998 ncm-et-4 1306271 500 199 47 254 +109.07 ± 14.99 0 15 73 157 5 +234.38 ± 39.98
166997 ncm-et-13 1315916 500 208 59 233 +106.78 ± 14.05 1 9 82 156 2 +236.51 ± 37.9
166996 ncm-et-3 1310599 500 207 39 254 +121.45 ± 13.49 0 5 78 161 6 +268.17 ± 38.8
166995 ncm-et-9 1302237 500 206 43 251 +117.55 ± 13.87 0 8 76 161 5 +258.75 ± 39.42

Commit

Commit ID 08385527dd470ece814ac85013802995a0e7f6ca
Author Joost VandeVondele
Date 2023-02-23 12:25:35 UTC
Introduce a function to compute NNUE accumulator This patch introduces `hint_common_parent_position()` to signal that potentially several child nodes will require an NNUE eval. By populating explicitly the accumulator, these subsequent evaluations can be performed more efficiently. This was based on the observation that calculating the evaluation in an excluded move position yielded a significant Elo gain, even though the evaluation itself was already available (work by pb00067). Sopel wrote the code to perform just the accumulator update. This PR is based on cleaned up code that passed STC: https://tests.stockfishchess.org/tests/view/63f62f9be74a12625bcd4aa0 LLR: 2.94 (-2.94,2.94) <0.50,2.50> Total: 110368 W: 29607 L: 29167 D: 51594 Ptnml(0-2): 41, 10551, 33572, 10967, 53 and in an the earlier (equivalent) version passed STC: https://tests.stockfishchess.org/tests/view/63f3c3fee74a12625bcce2a6 LLR: 2.95 (-2.94,2.94) <0.00,2.00> Total: 47552 W: 12786 L: 12467 D: 22299 Ptnml(0-2): 120, 5107, 12997, 5438, 114 passed LTC: https://tests.stockfishchess.org/tests/view/63f45cc2e74a12625bccfa63 LLR: 2.94 (-2.94,2.94) <0.50,2.50> Total: 110368 W: 29607 L: 29167 D: 51594 Ptnml(0-2): 41, 10551, 33572, 10967, 53 closes https://github.com/official-stockfish/Stockfish/pull/4402 Bench: 3726250
Copyright 2011–2024 Next Chess Move LLC