Dev Builds » 20220911-1219

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 06:53:28 584037 3990 1104 911 1975 +16.82 ± 5.12 5 342 1106 539 3 +34.07 ± 10.17
ncm-dbt-02 06:54:03 586257 4000 1131 928 1941 +17.65 ± 5.21 8 347 1082 560 3 +36.26 ± 10.31
ncm-dbt-03 06:53:19 586315 4000 1134 953 1913 +15.73 ± 5.22 10 352 1089 545 4 +32.58 ± 10.27
ncm-dbt-04 06:53:15 571567 4010 1152 899 1959 +21.95 ± 5.32 11 335 1057 594 8 +44.6 ± 10.45
ncm-dbt-05 06:53:33 584413 4000 1079 953 1968 +10.95 ± 5.32 16 377 1075 529 3 +24.19 ± 10.35
20000 5600 4644 9756 +16.62 ± 2.34 50 1753 5409 2767 21 +34.33 ± 4.61

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
379255 ncm-dbt-04 572558 10 4 4 2 -0.0 ± 99.06 0 1 3 1 0 -0.0 ± 217.03 ↓
379254 ncm-dbt-01 582652 490 129 113 248 +11.35 ± 13.56 0 40 149 56 0 +22.72 ± 27.26 ↓
379253 ncm-dbt-02 584790 500 146 116 238 +20.87 ± 14.88 1 43 131 75 0 +43.3 ± 29.76 ↓
379252 ncm-dbt-03 587579 500 144 112 244 +22.27 ± 15.35 2 43 126 79 0 +47.55 ± 30.4 ↓
379251 ncm-dbt-05 587113 500 145 117 238 +19.48 ± 15.76 3 45 123 79 0 +43.3 ± 30.78 ↓
379250 ncm-dbt-04 570228 500 132 108 260 +16.69 ± 14.94 2 42 137 68 1 +34.86 ± 28.99 ↓
379249 ncm-dbt-01 584034 500 131 113 256 +12.51 ± 13.69 0 42 148 60 0 +25.06 ± 27.53 ↓
379248 ncm-dbt-02 585801 500 141 128 231 +9.04 ± 14.71 1 49 136 64 0 +19.48 ± 29.14 ↓
379247 ncm-dbt-03 585590 500 144 128 228 +11.12 ± 15.0 2 47 134 67 0 +25.06 ± 29.4 ↓
379246 ncm-dbt-05 581194 500 137 118 245 +13.21 ± 14.15 0 44 144 61 1 +25.06 ± 28.08 ↓
379245 ncm-dbt-04 572034 500 141 123 236 +12.51 ± 15.24 1 50 130 68 1 +25.06 ± 29.91 ↓
379244 ncm-dbt-01 586477 500 150 123 227 +18.78 ± 14.47 0 44 135 71 0 +37.67 ± 29.25 ↓
379243 ncm-dbt-03 586901 500 144 120 236 +16.69 ± 14.44 1 42 139 68 0 +34.86 ± 28.73 ↓
379242 ncm-dbt-02 586520 500 138 125 237 +9.03 ± 14.45 0 49 140 60 1 +16.69 ± 28.62 ↓
379241 ncm-dbt-04 570669 500 151 120 229 +21.57 ± 15.06 1 43 131 74 1 +43.3 ± 29.76 ↓
379240 ncm-dbt-05 583531 500 133 120 247 +9.04 ± 15.08 2 49 133 66 0 +20.87 ± 29.53 ↓
379239 ncm-dbt-03 585843 500 143 116 241 +18.78 ± 14.6 2 39 139 70 0 +40.48 ± 28.71 ↓
379238 ncm-dbt-01 584495 500 130 101 269 +20.17 ± 16.04 1 52 115 81 1 +40.48 ± 31.77 ↓
379237 ncm-dbt-02 584117 500 133 115 252 +12.51 ± 14.99 2 46 134 68 0 +27.85 ± 29.39 ↓
379236 ncm-dbt-04 573002 500 135 103 262 +22.27 ± 14.6 2 36 141 70 1 +46.13 ± 28.42 ↓
379235 ncm-dbt-05 587197 500 140 118 242 +15.3 ± 15.09 3 42 135 70 0 +34.86 ± 29.25 ↓
379234 ncm-dbt-03 584706 500 151 113 236 +26.46 ± 15.03 1 40 130 78 1 +53.22 ± 29.88 ↓
379233 ncm-dbt-01 581777 500 145 124 231 +14.6 ± 14.91 1 45 138 64 2 +27.85 ± 28.87 ↓
379232 ncm-dbt-04 569869 500 153 111 236 +29.25 ± 14.59 0 38 133 78 1 +57.5 ± 29.47 ↓
379231 ncm-dbt-02 586181 500 147 111 242 +25.06 ± 14.16 0 38 138 74 0 +50.38 ± 28.82 ↓
379230 ncm-dbt-05 582152 500 129 114 257 +10.43 ± 15.56 3 49 128 70 0 +25.06 ± 30.16 ↓
379229 ncm-dbt-01 583196 500 136 104 260 +22.27 ± 13.54 1 32 151 66 0 +46.13 ± 27.03 ↓
379228 ncm-dbt-03 588345 500 132 116 252 +11.12 ± 14.88 2 46 136 66 0 +25.06 ± 29.14 ↓
379227 ncm-dbt-04 570709 500 145 109 246 +25.06 ± 15.06 2 38 133 76 1 +51.8 ± 29.48 ↓
379226 ncm-dbt-02 589539 500 135 107 258 +19.48 ± 14.78 2 40 136 72 0 +41.89 ± 29.11 ↓
379225 ncm-dbt-05 583321 500 131 122 247 +6.25 ± 14.47 0 52 137 61 0 +12.51 ± 29.02 ↓
379224 ncm-dbt-01 583531 500 137 122 241 +10.43 ± 14.05 0 46 143 61 0 +20.87 ± 28.22 ↓
379223 ncm-dbt-03 585801 500 128 124 248 +2.78 ± 14.03 0 51 144 55 0 +5.56 ± 28.1 ↓
379222 ncm-dbt-04 572074 500 145 108 247 +25.76 ± 15.35 2 40 128 79 1 +53.22 ± 30.14 ↓
379221 ncm-dbt-05 585464 500 130 121 249 +6.25 ± 15.7 4 49 132 64 1 +16.69 ± 29.66 ↓
379220 ncm-dbt-02 583447 500 144 113 243 +21.57 ± 15.06 1 44 128 77 0 +44.72 ± 30.15 ↓
379219 ncm-dbt-01 586139 500 146 111 243 +24.36 ± 15.26 2 41 127 80 0 +51.8 ± 30.27 ↓
379218 ncm-dbt-03 585759 500 148 124 228 +16.69 ± 14.69 0 44 141 62 3 +29.25 ± 28.47 ↓
379217 ncm-dbt-04 572961 500 146 113 241 +22.96 ± 15.65 1 47 121 80 1 +46.13 ± 31.04 ↓
379216 ncm-dbt-02 589668 500 147 113 240 +23.66 ± 14.71 1 38 139 70 2 +46.13 ± 28.7 ↓
379215 ncm-dbt-05 585337 500 134 123 243 +7.64 ± 14.46 1 47 143 58 1 +15.3 ± 28.23 ↓

Commit

Commit ID 82bb21dc7a198609589ef0cc78d185f00f619a90
Author mstembera
Date 2022-09-11 12:19:57 UTC
Optimize AVX2 path in NNUE evaluation always selecting AffineTransform specialization for small inputs. A related patch was tested as Initially tested as a simplification STC https://tests.stockfishchess.org/tests/view/6317c3f437f41b13973d6dff LLR: 2.95 (-2.94,2.94) <-1.75,0.25> Total: 58072 W: 15619 L: 15425 D: 27028 Ptnml(0-2): 241, 6191, 15992, 6357, 255 Elo gain speedup test STC https://tests.stockfishchess.org/tests/view/63181c1b37f41b13973d79dc LLR: 2.94 (-2.94,2.94) <0.00,2.00> Total: 184496 W: 49922 L: 49401 D: 85173 Ptnml(0-2): 851, 19397, 51208, 19964, 828 and this patch gained in testing speedup = +0.0071 P(speedup > 0) = 1.0000 on CPU: 16 x AMD Ryzen 9 3950X closes https://github.com/official-stockfish/Stockfish/pull/4158 No functional change
Copyright 2011–2024 Next Chess Move LLC
Feedback & Support Terms of Service Privacy Policy