Dev Builds » 20211203-0751

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 06:50:33 582606 4000 836 1225 1939 -33.9 ± 5.26 16 650 1043 289 2 -65.92 ± 10.51
ncm-dbt-02 06:52:49 585850 3996 822 1192 1982 -32.26 ± 5.25 19 627 1058 293 1 -61.86 ± 10.43
ncm-dbt-03 06:52:37 585015 3994 828 1251 1915 -36.94 ± 5.18 11 672 1046 265 3 -73.27 ± 10.48
ncm-dbt-04 06:51:53 567268 4010 821 1201 1988 -33.02 ± 5.29 13 662 1023 306 1 -64.5 ± 10.63
ncm-dbt-05 06:51:22 581419 4000 820 1240 1940 -36.62 ± 5.3 14 682 1018 282 4 -72.25 ± 10.65
20000 4127 6109 9764 -34.54 ± 2.35 73 3293 5188 1435 11 -67.55 ± 4.71

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
435188 ncm-dbt-04 566809 10 1 3 6 -70.36 ± 79.51 0 2 3 0 0 -147.07 ± 211.27 ↓
435187 ncm-dbt-03 582778 494 102 141 251 -27.49 ± 15.63 2 78 126 39 2 -55.32 ± 30.38 ↓
435186 ncm-dbt-02 584537 496 106 163 227 -40.1 ± 14.73 5 76 138 29 0 -73.95 ± 28.66 ↓
435185 ncm-dbt-01 584369 500 102 150 248 -33.46 ± 14.74 1 83 129 37 0 -66.1 ± 29.99 ↓
435184 ncm-dbt-05 583363 500 106 159 235 -36.97 ± 13.9 2 77 143 28 0 -71.89 ± 28.0 ↓
435183 ncm-dbt-04 568832 500 98 148 254 -34.86 ± 14.57 1 83 131 35 0 -68.99 ± 29.71 ↓
435182 ncm-dbt-03 584874 500 107 164 229 -39.78 ± 14.48 1 87 130 32 0 -79.17 ± 29.81 ↓
435181 ncm-dbt-02 586858 500 109 150 241 -28.55 ± 15.05 0 84 123 43 0 -57.5 ± 30.78 ↓
435180 ncm-dbt-01 582318 500 108 154 238 -32.05 ± 15.03 0 87 122 41 0 -64.66 ± 30.91 ↓
435179 ncm-dbt-05 581652 500 115 159 226 -30.65 ± 14.81 0 83 129 37 1 -63.23 ± 29.99 ↓
435178 ncm-dbt-04 567799 500 102 151 247 -34.16 ± 15.89 2 89 116 42 1 -67.55 ± 31.69 ↓
435177 ncm-dbt-03 584664 500 120 159 221 -27.15 ± 14.04 0 75 139 36 0 -54.65 ± 28.67 ↓
435176 ncm-dbt-02 584117 500 101 140 259 -27.15 ± 15.08 2 77 129 42 0 -51.8 ± 30.01 ↓
435175 ncm-dbt-01 580530 500 96 147 257 -35.56 ± 15.0 2 83 130 34 1 -70.44 ± 29.84 ↓
435174 ncm-dbt-04 565705 500 91 159 250 -47.54 ± 13.99 1 91 133 25 0 -95.44 ± 29.29 ↓
435173 ncm-dbt-05 576660 500 88 153 259 -45.42 ± 15.07 2 94 121 33 0 -89.48 ± 31.03 ↓
435172 ncm-dbt-03 582945 500 109 161 230 -36.26 ± 15.42 4 81 129 35 1 -68.99 ± 29.98 ↓
435171 ncm-dbt-04 567324 500 103 146 251 -29.95 ± 15.14 1 83 124 42 0 -58.93 ± 30.66 ↓
435170 ncm-dbt-01 582611 500 106 141 253 -24.36 ± 13.97 1 69 144 36 0 -47.55 ± 28.01 ↓
435169 ncm-dbt-02 586731 500 101 145 254 -30.65 ± 14.94 4 73 136 37 0 -56.07 ± 29.07 ↓
435168 ncm-dbt-05 577971 500 100 152 248 -36.26 ± 14.4 1 83 133 33 0 -71.89 ± 29.42 ↓
435167 ncm-dbt-03 586350 500 93 149 258 -39.08 ± 14.57 0 90 126 34 0 -79.17 ± 30.36 ↓
435166 ncm-dbt-04 566651 500 111 139 250 -19.48 ± 14.78 1 71 133 45 0 -37.67 ± 29.51 ↓
435165 ncm-dbt-01 585928 500 114 166 220 -36.26 ± 14.66 4 76 138 32 0 -67.55 ± 28.74 ↓
435164 ncm-dbt-02 587155 500 112 142 246 -20.87 ± 15.26 1 76 125 48 0 -40.48 ± 30.53 ↓
435163 ncm-dbt-05 584243 500 110 153 237 -29.95 ± 15.02 3 76 132 39 0 -56.07 ± 29.61 ↓
435162 ncm-dbt-03 585253 500 93 161 246 -47.54 ± 14.92 3 92 125 30 0 -92.46 ± 30.47 ↓
435161 ncm-dbt-04 568236 500 110 157 233 -32.75 ± 14.95 3 78 132 37 0 -61.79 ± 29.59 ↓
435160 ncm-dbt-01 583154 500 106 149 245 -29.95 ± 15.51 4 76 130 39 1 -56.07 ± 29.87 ↓
435159 ncm-dbt-05 583405 500 114 173 213 -41.19 ± 15.08 2 89 126 32 1 -82.1 ± 30.36 ↓
435158 ncm-dbt-02 586181 500 101 155 244 -37.67 ± 15.13 5 78 133 34 0 -68.99 ± 29.43 ↓
435157 ncm-dbt-04 565744 500 104 151 245 -32.75 ± 14.95 2 81 129 38 0 -63.23 ± 29.99 ↓
435156 ncm-dbt-03 587877 500 99 157 244 -40.48 ± 13.85 0 86 136 28 0 -82.1 ± 28.94 ↓
435155 ncm-dbt-01 579620 500 106 159 235 -36.97 ± 14.96 1 88 124 37 0 -73.34 ± 30.64 ↓
435154 ncm-dbt-05 580572 500 93 137 270 -30.65 ± 15.68 1 87 118 43 1 -61.79 ± 31.43 ↓
435153 ncm-dbt-02 585674 500 99 151 250 -36.26 ± 14.79 2 82 133 32 1 -71.89 ± 29.42 ↓
435152 ncm-dbt-03 585379 500 105 159 236 -37.67 ± 14.22 1 83 135 31 0 -74.79 ± 29.13 ↓
435151 ncm-dbt-01 582318 500 98 159 243 -42.6 ± 15.04 3 88 126 33 0 -82.1 ± 30.36 ↓
435150 ncm-dbt-04 568315 500 101 147 252 -32.05 ± 15.4 2 84 122 42 0 -61.79 ± 30.91 ↓
435149 ncm-dbt-02 585548 500 93 146 261 -36.97 ± 13.63 0 81 141 28 0 -74.79 ± 28.27 ↓
435148 ncm-dbt-05 583489 500 94 154 252 -41.89 ± 15.86 3 93 116 37 1 -82.1 ± 31.71 ↓

Commit

Commit ID c9977aa0a89c83bf21651bffd3b6f10c344ccc46
Author Gian-Carlo Pascutto
Date 2021-12-03 07:51:06 UTC
Add AVX-VNNI support for Alder Lake and later. In their infinite wisdom, Intel axed AVX512 from Alder Lake chips (well, not entirely, but we kind of want to use the Gracemont cores for chess!) but still added VNNI support. Confusingly enough, this is not the same as VNNI256 support. This adds a specific AVX-VNNI target that will use this AVX-VNNI mode, by prefixing the VNNI instructions with the appropriate VEX prefix, and avoiding AVX512 usage. This is about 1% faster on P cores: Result of 20 runs ================== base (./clang-bmi2 ) = 3306337 +/- 7519 test (./clang-vnni ) = 3344226 +/- 7388 diff = +37889 +/- 4153 speedup = +0.0115 P(speedup > 0) = 1.0000 But a nice 3% faster on E cores: Result of 20 runs ================== base (./clang-bmi2 ) = 1938054 +/- 28257 test (./clang-vnni ) = 1994606 +/- 31756 diff = +56552 +/- 3735 speedup = +0.0292 P(speedup > 0) = 1.0000 This was measured on Clang 13. GCC 11.2 appears to generate worse code for Alder Lake, though the speedup on the E cores is similar. It is possible to run the engine specifically on the P or E using binding, for example in linux it is possible to use (for an 8 P + 8 E setup like i9-12900K): taskset -c 0-15 ./stockfish taskset -c 16-23 ./stockfish where the first call binds to the P-cores and the second to the E-cores. closes https://github.com/official-stockfish/Stockfish/pull/3824 No functional change
Copyright 2011–2025 Next Chess Move LLC
Feedback & Support Terms of Service Privacy Policy Desktop Version