Dev Builds » 20210622-0951

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 06:56:42 583688 3996 475 1561 1960 -96.86 ± 4.94 25 1124 761 88 0 -205.54 ± 12.33
ncm-dbt-02 06:54:27 586648 4000 509 1567 1924 -94.13 ± 5.07 26 1115 750 109 0 -198.34 ± 12.44
ncm-dbt-03 06:55:48 586780 4000 447 1544 2009 -97.79 ± 5.13 52 1081 779 88 0 -201.43 ± 12.18
ncm-dbt-04 06:54:41 567589 4000 457 1576 1967 -99.86 ± 5.0 35 1133 749 82 1 -211.13 ± 12.43
ncm-dbt-05 06:55:40 583225 4004 478 1565 1961 -96.75 ± 4.96 28 1119 767 88 0 -204.54 ± 12.28
20000 2366 7813 9821 -97.07 ± 2.25 166 5572 3806 455 1 -204.17 ± 5.51

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
453305 ncm-dbt-05 581277 4 0 2 2 -190.27 ± 27.79 0 2 0 0 0 -1199.83 ± 312.71
453304 ncm-dbt-01 584034 496 53 195 248 -102.33 ± 14.09 5 141 93 9 0 -216.06 ± 35.43
453303 ncm-dbt-04 566020 500 71 197 232 -89.48 ± 13.38 2 131 108 9 0 -189.0 ± 32.58
453302 ncm-dbt-03 585042 500 56 194 250 -98.44 ± 15.46 13 123 103 11 0 -190.85 ± 33.57
453301 ncm-dbt-05 581111 500 63 198 239 -96.19 ± 13.48 2 140 99 9 0 -206.01 ± 34.24
453300 ncm-dbt-02 587452 500 63 190 247 -90.22 ± 14.16 3 134 100 13 0 -189.0 ± 34.18
453299 ncm-dbt-04 569949 500 61 191 248 -92.45 ± 14.05 3 136 99 12 0 -194.57 ± 34.34
453298 ncm-dbt-01 583782 500 62 200 238 -98.44 ± 14.44 5 140 93 12 0 -206.01 ± 35.51
453297 ncm-dbt-03 588558 500 63 202 235 -99.2 ± 15.04 8 136 93 13 0 -202.15 ± 35.52
453296 ncm-dbt-05 587155 500 53 199 248 -104.49 ± 14.96 9 139 91 11 0 -213.85 ± 35.9
453295 ncm-dbt-02 585253 500 68 190 242 -86.52 ± 14.38 2 134 98 16 0 -181.7 ± 34.58
453294 ncm-dbt-04 567997 500 58 205 237 -105.25 ± 14.66 7 144 88 11 0 -219.87 ± 36.54
453293 ncm-dbt-01 583321 500 61 199 240 -98.44 ± 13.67 3 141 97 9 0 -209.91 ± 34.63
453292 ncm-dbt-03 587664 500 45 192 263 -105.25 ± 13.72 5 144 94 7 0 -223.94 ± 35.14
453291 ncm-dbt-05 575313 500 51 200 249 -106.77 ± 12.71 2 149 95 4 0 -234.38 ± 34.71
453290 ncm-dbt-02 585548 500 68 195 237 -90.22 ± 14.01 0 142 93 15 0 -194.57 ± 35.52
453289 ncm-dbt-01 582987 500 61 198 241 -97.69 ± 14.13 2 146 89 13 0 -209.91 ± 36.34
453288 ncm-dbt-04 565351 500 58 197 245 -99.2 ± 13.99 2 146 92 9 1 -215.85 ± 35.68
453287 ncm-dbt-03 586435 500 63 192 245 -91.71 ± 14.49 4 135 97 14 0 -190.85 ± 34.75
453286 ncm-dbt-02 587664 500 58 200 242 -101.46 ± 13.53 3 144 95 8 0 -217.85 ± 34.98
453285 ncm-dbt-05 586477 500 60 200 240 -99.95 ± 14.0 4 142 94 10 0 -211.87 ± 35.26
453284 ncm-dbt-01 583405 500 59 190 251 -93.2 ± 14.21 5 132 102 11 0 -192.71 ± 33.75
453283 ncm-dbt-04 569230 500 51 200 249 -106.77 ± 13.89 5 147 90 8 0 -228.08 ± 36.04
453282 ncm-dbt-03 587494 500 60 196 244 -96.94 ± 14.42 6 135 98 11 0 -200.24 ± 34.5
453281 ncm-dbt-05 583824 500 59 179 262 -85.04 ± 13.9 2 129 106 13 0 -178.11 ± 33.1
453280 ncm-dbt-02 586223 500 66 199 235 -94.69 ± 13.93 2 141 95 12 0 -202.15 ± 35.11
453279 ncm-dbt-04 567640 500 54 200 246 -104.49 ± 14.81 7 144 87 12 0 -217.85 ± 36.77
453278 ncm-dbt-03 586520 500 50 198 252 -106.01 ± 14.67 6 148 84 12 0 -223.94 ± 37.43
453277 ncm-dbt-01 583321 500 60 185 255 -88.74 ± 13.83 1 136 100 13 0 -189.0 ± 34.18
453276 ncm-dbt-05 585632 500 68 199 233 -93.2 ± 14.06 2 140 95 13 0 -198.34 ± 35.12
453275 ncm-dbt-02 585928 500 72 194 234 -86.52 ± 14.95 4 132 96 18 0 -178.11 ± 34.94
453274 ncm-dbt-03 586943 500 63 182 255 -84.3 ± 13.73 6 115 121 8 0 -169.27 ± 30.34
453273 ncm-dbt-05 584916 500 65 190 245 -88.74 ± 14.28 2 136 97 15 0 -187.16 ± 34.76
453272 ncm-dbt-04 567284 500 48 191 261 -102.22 ± 13.37 2 147 93 8 0 -221.9 ± 35.4
453271 ncm-dbt-01 583698 500 59 193 248 -95.44 ± 13.79 1 144 93 12 0 -206.01 ± 35.51
453270 ncm-dbt-02 587494 500 58 203 239 -103.73 ± 14.95 6 147 83 14 0 -217.85 ± 37.63
453269 ncm-dbt-04 567244 500 56 195 249 -99.2 ± 14.9 7 138 92 13 0 -204.07 ± 35.72
453268 ncm-dbt-03 585590 500 47 188 265 -100.7 ± 14.32 4 145 89 12 0 -213.85 ± 36.34
453267 ncm-dbt-05 583321 500 59 198 243 -99.2 ± 14.6 5 142 90 13 0 -207.95 ± 36.13
453266 ncm-dbt-02 587622 500 56 196 248 -99.95 ± 14.76 6 141 90 13 0 -207.95 ± 36.13
453265 ncm-dbt-01 584958 500 60 201 239 -100.7 ± 13.69 3 144 94 9 0 -215.85 ± 35.23

Commit

Commit ID 0470bcef0e1962b4f8da15108170b991d3f90d0e
Author Stéphane Nicolet
Date 2021-06-22 09:51:03 UTC
Detect fortresses a little bit quicker In the so-called "hybrid" method of evaluation of current master, we use the classical eval (because of its speed) instead of the NNUE eval when the classical material balance approximation hints that the position is "winning enough" to rely on the classical eval. This trade-off idea between speed and accuracy works well in general, but in some fortress positions the classical eval is just bad. So in shuffling branches of the search tree, we (slowly) increase the thresehold so that eventually we don't trust classical anymore and switch to NNUE evaluation. This patch increases that threshold faster, so that we switch to NNUE quicker in shuffling branches. Idea is to incite Stockfish to spend less time in fortresses lines in the search tree, and spend more time searching the critical lines. passed STC: LLR: 2.96 (-2.94,2.94) <-0.50,2.50> Total: 47872 W: 3908 L: 3720 D: 40244 Ptnml(0-2): 122, 3053, 17419, 3199, 143 https://tests.stockfishchess.org/tests/view/60cef34b457376eb8bcab79d passed LTC: LLR: 2.93 (-2.94,2.94) <0.50,3.50> Total: 73616 W: 2326 L: 2143 D: 69147 Ptnml(0-2): 21, 1940, 32705, 2119, 23 https://tests.stockfishchess.org/tests/view/60cf6d842114332881e73528 Retested at LTC against lastest master: LLR: 2.93 (-2.94,2.94) <0.50,3.50> Total: 18264 W: 642 L: 532 D: 17090 Ptnml(0-2): 6, 479, 8055, 583, 9 https://tests.stockfishchess.org/tests/view/60d18cd540925195e7a6c351 closes https://github.com/official-stockfish/Stockfish/pull/3578 Bench: 5139233
Copyright 2011–2026 Next Chess Move LLC