Dev Builds » 20211105-2122

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 06:52:29 585097 4000 736 1286 1978 -48.08 ± 5.2 9 771 983 235 2 -96.75 ± 10.84
ncm-dbt-02 06:53:24 586348 4000 732 1295 1973 -49.23 ± 5.05 2 777 1003 218 0 -100.14 ± 10.71
ncm-dbt-03 06:52:45 587524 4000 739 1304 1957 -49.41 ± 5.19 7 783 983 222 5 -100.52 ± 10.84
ncm-dbt-04 06:53:09 571208 4000 738 1307 1955 -49.76 ± 5.2 10 780 982 225 3 -100.33 ± 10.85
ncm-dbt-05 06:54:52 584538 4000 711 1312 1977 -52.6 ± 5.12 12 785 997 204 2 -105.82 ± 10.74
20000 3656 6504 9840 -49.81 ± 2.3 40 3896 4948 1104 12 -100.71 ± 4.83

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
405742 ncm-dbt-03 586054 500 100 167 233 -46.84 ± 15.14 1 97 122 28 2 -96.94 ± 30.88
405741 ncm-dbt-04 572800 500 85 168 247 -58.21 ± 14.39 2 102 123 23 0 -116.78 ± 30.65
405740 ncm-dbt-05 587282 500 89 168 243 -55.36 ± 14.79 2 102 119 27 0 -110.6 ± 31.28
405739 ncm-dbt-02 589240 500 104 180 216 -53.22 ± 14.56 0 104 118 28 0 -109.07 ± 31.43
405738 ncm-dbt-01 585928 500 99 180 221 -56.78 ± 14.19 1 102 124 23 0 -115.23 ± 30.51
405132 ncm-dbt-05 586393 500 84 160 256 -53.22 ± 14.15 1 98 127 24 0 -107.54 ± 30.1
405131 ncm-dbt-04 573527 500 97 158 245 -42.6 ± 14.38 1 89 130 30 0 -85.04 ± 29.79
405130 ncm-dbt-01 586435 500 92 158 250 -46.13 ± 14.85 2 92 127 28 1 -92.46 ± 30.18
405129 ncm-dbt-03 588643 500 92 154 254 -43.3 ± 14.69 2 88 131 28 1 -86.52 ± 29.64
405128 ncm-dbt-02 587155 500 82 161 257 -55.36 ± 13.98 0 102 125 23 0 -113.68 ± 30.37
405127 ncm-dbt-05 577438 500 107 171 222 -44.72 ± 15.53 4 91 120 35 0 -85.04 ± 31.17
405126 ncm-dbt-04 568434 500 92 153 255 -42.6 ± 15.16 0 98 115 37 0 -86.52 ± 31.85
405125 ncm-dbt-01 583447 500 97 166 237 -48.25 ± 14.43 3 89 132 26 0 -93.95 ± 29.45
405124 ncm-dbt-02 589027 500 90 165 245 -52.51 ± 14.11 1 97 128 24 0 -106.01 ± 29.96
405123 ncm-dbt-03 586604 500 103 164 233 -42.6 ± 14.12 0 90 131 29 0 -86.52 ± 29.64
405122 ncm-dbt-04 572074 500 99 167 234 -47.54 ± 14.66 1 96 123 30 0 -95.44 ± 30.74
405121 ncm-dbt-05 583028 500 88 167 245 -55.36 ± 14.66 0 107 115 28 0 -113.68 ± 31.86
405120 ncm-dbt-01 584958 500 85 161 254 -53.22 ± 14.42 0 103 120 27 0 -109.07 ± 31.14
405119 ncm-dbt-02 588728 500 90 158 252 -47.54 ± 14.66 1 96 123 30 0 -95.44 ± 30.74
405118 ncm-dbt-03 586816 500 96 167 237 -49.67 ± 13.83 1 91 137 20 1 -101.46 ± 28.63
405117 ncm-dbt-04 570749 500 100 179 221 -55.36 ± 13.7 1 97 132 20 0 -112.14 ± 29.29
405116 ncm-dbt-01 586139 500 84 154 262 -48.96 ± 15.0 1 100 117 32 0 -98.44 ± 31.58
405115 ncm-dbt-05 587028 500 79 151 270 -50.38 ± 14.41 0 99 125 25 1 -104.49 ± 30.42
405114 ncm-dbt-02 586266 500 91 159 250 -47.55 ± 14.4 0 97 124 29 0 -96.94 ± 30.59
405113 ncm-dbt-03 588174 500 85 153 262 -47.54 ± 14.66 1 96 123 30 0 -95.44 ± 30.74
405112 ncm-dbt-04 570428 500 97 153 250 -39.08 ± 15.21 0 95 116 39 0 -79.17 ± 31.71
405111 ncm-dbt-01 583321 500 87 160 253 -51.09 ± 14.58 1 99 122 28 0 -102.97 ± 30.86
405110 ncm-dbt-05 586689 500 94 158 248 -44.72 ± 15.41 1 99 113 37 0 -89.48 ± 32.12
405109 ncm-dbt-02 583028 500 101 172 227 -49.67 ± 13.83 0 95 131 24 0 -101.46 ± 29.54
405108 ncm-dbt-03 586054 500 83 155 262 -50.38 ± 14.94 0 104 114 32 0 -102.97 ± 32.0
405107 ncm-dbt-05 582736 500 82 170 248 -61.79 ± 13.56 4 94 138 14 0 -121.46 ± 28.19
405106 ncm-dbt-03 587240 500 90 183 227 -65.38 ± 14.0 1 110 120 19 0 -134.15 ± 31.01
405105 ncm-dbt-04 571029 500 79 165 256 -60.36 ± 14.62 2 106 118 24 0 -121.46 ± 31.39
405104 ncm-dbt-01 584369 500 97 149 254 -36.26 ± 15.29 1 90 119 40 0 -71.89 ± 31.31
405103 ncm-dbt-02 580613 500 93 151 256 -40.48 ± 14.26 0 89 130 31 0 -82.1 ± 29.8
405102 ncm-dbt-01 586181 500 95 158 247 -44.01 ± 14.86 0 96 122 31 1 -90.97 ± 30.89
405101 ncm-dbt-04 570629 500 89 164 247 -52.51 ± 15.3 3 97 125 22 3 -107.54 ± 30.4
405100 ncm-dbt-03 590610 500 90 161 249 -49.67 ± 15.91 1 107 105 36 1 -101.46 ± 33.23
405099 ncm-dbt-05 585717 500 88 167 245 -55.36 ± 13.11 0 95 140 14 1 -115.23 ± 27.96
405098 ncm-dbt-02 586731 500 81 149 270 -47.55 ± 14.4 0 97 124 29 0 -96.94 ± 30.59

Commit

Commit ID a0259d8ab9661b7f625474d2cbe18481ef69bbf2
Author Stefan Geschwentner
Date 2021-11-05 21:22:30 UTC
Tweak initial aspiration window. Maintain for each root move an exponential average of the search value with a weight ratio of 2:1 (new value vs old values). Then the average score is used as the center of the initial aspiration window instead of the previous score. Stats indicate (see PR) that the deviation for previous score is in general greater than using average score, so later seems a better estimation of the next search value. This is probably the reason this patch succeded besides smoothing the sometimes wild swings in search score. An additional observation is that at higher depth previous score is above but average score below zero. So for average score more/less fail/low highs should be occur than previous score. STC: LLR: 2.97 (-2.94,2.94) <0.00,2.50> Total: 59792 W: 15106 L: 14792 D: 29894 Ptnml(0-2): 144, 6718, 15869, 7010, 155 https://tests.stockfishchess.org/tests/view/61841612d7a085ad008eef06 LTC: LLR: 2.94 (-2.94,2.94) <0.50,3.00> Total: 46448 W: 11835 L: 11537 D: 23076 Ptnml(0-2): 21, 4756, 13374, 5050, 23 https://tests.stockfishchess.org/tests/view/618463abd7a085ad008eef3e closes https://github.com/official-stockfish/Stockfish/pull/3776 Bench: 6719976
Copyright 2011–2025 Next Chess Move LLC