Dev Builds » 20211207-1708

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 09:25:12 1145126 3164 1084 461 1619 +69.32 ± 5.91 2 137 696 730 17 +140.75 ± 12.91
ncm-dbt-02 10:09:41 1329761 3394 1137 524 1733 +63.45 ± 5.85 2 172 758 741 24 +126.28 ± 12.37
ncm-dbt-03 10:14:30 1327547 3428 1180 523 1725 +67.42 ± 5.68 2 144 789 753 26 +134.67 ± 12.1
ncm-dbt-04 10:14:53 1320849 3438 1206 554 1678 +66.7 ± 5.6 0 146 798 752 23 +133.31 ± 12.03
ncm-dbt-05 10:10:04 1323025 3414 1201 548 1665 +67.28 ± 5.68 3 146 772 767 19 +136.23 ± 12.25
ncm-dbt-06 09:23:23 1225379 3162 1064 484 1614 +64.46 ± 5.91 1 150 713 702 15 +130.15 ± 12.75
20000 6872 3094 10034 +66.43 ± 2.36 10 895 4526 4445 124 +133.51 ± 5.06

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
188010 ncm-dbt-02 1248466 144 48 22 74 +63.42 ± 26.2 0 6 34 32 0 +131.39 ± 58.95
188009 ncm-dbt-05 1247180 164 50 28 86 +46.88 ± 26.15 0 10 41 30 1 +91.0 ± 53.55
188008 ncm-dbt-06 1201988 162 61 25 76 +78.51 ± 25.07 0 5 36 39 1 +160.68 ± 57.3
188007 ncm-dbt-01 1147171 164 62 24 78 +81.99 ± 25.02 0 5 35 41 1 +168.93 ± 58.27
188006 ncm-dbt-03 1206988 178 64 28 86 +71.25 ± 26.35 0 8 40 38 3 +135.27 ± 54.41
188005 ncm-dbt-04 1236056 188 65 35 88 +55.92 ± 23.08 0 9 46 39 0 +114.9 ± 50.46
188004 ncm-dbt-02 1235101 500 154 73 273 +56.78 ± 15.25 0 30 111 107 2 +113.68 ± 32.44
188003 ncm-dbt-05 1233520 500 169 81 250 +61.79 ± 15.73 0 30 106 110 4 +121.46 ± 33.17
188002 ncm-dbt-06 1203203 500 177 71 252 +74.79 ± 14.49 1 18 105 126 0 +158.93 ± 33.37
188001 ncm-dbt-01 1144251 500 175 73 252 +71.88 ± 13.81 0 14 123 110 3 +145.54 ± 30.38
188000 ncm-dbt-03 1240520 500 174 77 249 +68.27 ± 15.21 1 20 115 109 5 +135.76 ± 31.8
187999 ncm-dbt-04 1250206 500 172 78 250 +66.1 ± 14.59 0 23 111 115 1 +135.76 ± 32.43
187998 ncm-dbt-02 1231120 500 168 79 253 +62.51 ± 15.37 0 28 108 111 3 +124.6 ± 32.89
187997 ncm-dbt-05 1224138 500 180 81 239 +69.71 ± 13.74 1 14 120 115 0 +147.19 ± 30.88
187996 ncm-dbt-06 1218888 500 145 77 278 +47.54 ± 14.92 0 31 122 95 2 +93.95 ± 30.89
187995 ncm-dbt-01 1142415 500 166 84 250 +57.5 ± 15.28 1 27 113 107 2 +116.78 ± 32.15
187994 ncm-dbt-03 1230228 500 177 83 240 +66.1 ± 15.39 0 26 108 112 4 +130.94 ± 32.9
187993 ncm-dbt-04 1241745 500 180 86 234 +66.1 ± 15.39 0 25 111 109 5 +129.35 ± 32.44
187992 ncm-dbt-02 1233230 500 182 88 230 +66.1 ± 15.13 0 22 118 104 6 +127.76 ± 31.37
187991 ncm-dbt-05 1206474 500 187 74 239 +79.9 ± 15.33 0 23 94 130 3 +164.07 ± 35.23
187990 ncm-dbt-01 1150641 500 162 70 268 +64.66 ± 14.53 0 24 110 116 0 +134.15 ± 32.59
187989 ncm-dbt-06 1250199 500 164 76 260 +61.79 ± 15.6 0 28 111 106 5 +119.89 ± 32.44
187988 ncm-dbt-03 1256593 500 176 74 250 +71.88 ± 13.67 0 15 119 115 1 +148.85 ± 31.04
187987 ncm-dbt-04 1233584 500 175 82 243 +65.38 ± 15.36 0 24 115 105 6 +126.18 ± 31.84
187986 ncm-dbt-02 1245394 500 172 68 260 +73.33 ± 14.85 0 21 107 119 3 +148.85 ± 33.05
187985 ncm-dbt-05 1218208 500 191 84 225 +75.52 ± 15.6 1 19 109 114 7 +148.85 ± 32.73
187984 ncm-dbt-06 1211102 500 174 79 247 +66.82 ± 14.2 0 19 119 110 2 +135.76 ± 31.16
187983 ncm-dbt-01 1159150 500 175 81 244 +66.1 ± 14.73 0 24 109 116 1 +135.76 ± 32.75
187982 ncm-dbt-03 1218726 500 165 65 270 +70.43 ± 14.62 0 19 116 111 4 +140.62 ± 31.61
187981 ncm-dbt-04 1234777 500 166 84 250 +57.5 ± 14.08 0 21 128 99 2 +115.23 ± 29.89
187980 ncm-dbt-06 1234976 500 169 85 246 +58.93 ± 14.83 0 27 113 109 1 +119.89 ± 32.15
187979 ncm-dbt-02 1239475 500 156 93 251 +44.01 ± 16.57 2 38 109 97 4 +86.52 ± 32.64
187978 ncm-dbt-05 1233633 500 168 73 259 +66.82 ± 14.06 0 18 121 109 2 +135.76 ± 30.83
187977 ncm-dbt-03 1232803 500 166 81 253 +59.64 ± 16.61 1 34 99 111 5 +116.78 ± 34.14
187976 ncm-dbt-04 1222715 500 182 66 252 +82.1 ± 14.13 0 14 109 124 3 +169.27 ± 32.62
187975 ncm-dbt-01 1145716 500 177 54 269 +87.26 ± 14.83 0 17 97 132 4 +179.9 ± 34.76
187974 ncm-dbt-06 1257303 500 174 71 255 +72.61 ± 15.1 0 22 107 117 4 +145.54 ± 33.06
187973 ncm-dbt-02 1216515 500 180 68 252 +79.17 ± 14.48 0 16 110 120 4 +160.64 ± 32.5
187972 ncm-dbt-05 1229826 500 171 90 239 +56.78 ± 14.6 0 25 121 102 2 +113.68 ± 30.97
187971 ncm-dbt-01 1126542 500 167 75 258 +64.66 ± 15.97 1 26 109 108 6 +126.18 ± 32.74
187970 ncm-dbt-04 1216185 500 179 77 244 +71.88 ± 15.21 0 21 112 111 6 +140.62 ± 32.26
187969 ncm-dbt-03 1256340 500 173 68 259 +74.06 ± 13.59 0 11 127 108 4 +148.85 ± 29.62
174226 ncm-dbt-02 1463331 50 16 8 26 +56.05 ± 47.6 0 2 14 8 1 +99.95 ± 91.41
174225 ncm-dbt-04 1445464 50 19 7 24 +85.04 ± 41.74 0 1 11 13 0 +181.7 ± 107.11
174224 ncm-dbt-03 1456188 50 16 8 26 +56.06 ± 32.73 0 0 17 8 0 +115.21 ± 72.12
174223 ncm-dbt-05 1456184 50 17 6 27 +77.7 ± 45.94 0 2 10 13 0 +164.07 ± 113.84
174222 ncm-dbt-02 1461836 50 14 5 31 +63.22 ± 39.41 0 1 14 10 0 +130.94 ± 89.9
174221 ncm-dbt-03 1460496 50 18 7 25 +77.7 ± 41.12 0 1 12 12 0 +164.07 ± 100.82
174220 ncm-dbt-04 1437756 50 14 9 27 +34.86 ± 43.75 0 3 14 8 0 +70.44 ± 92.08
174219 ncm-dbt-05 1457813 50 18 10 22 +56.08 ± 51.65 1 1 12 11 0 +130.94 ± 101.68
174218 ncm-dbt-02 1463028 50 17 7 26 +70.43 ± 49.53 0 3 9 13 0 +147.19 ± 118.91
174217 ncm-dbt-03 1459747 50 17 12 21 +34.86 ± 47.98 0 4 12 9 0 +70.44 ± 101.42
174216 ncm-dbt-04 1442971 50 18 13 19 +34.86 ± 47.98 0 4 12 9 0 +70.44 ± 101.42
174215 ncm-dbt-05 1450586 50 16 7 27 +63.22 ± 44.25 0 2 12 11 0 +130.94 ± 101.68
174214 ncm-dbt-02 1460046 50 17 4 29 +92.44 ± 47.1 0 1 11 12 1 +181.7 ± 107.11
174213 ncm-dbt-03 1454131 50 14 9 27 +34.86 ± 47.98 0 4 12 9 0 +70.44 ± 101.42
174211 ncm-dbt-05 1458556 50 18 6 26 +85.03 ± 36.28 0 0 13 12 0 +181.68 ± 91.94
174210 ncm-dbt-04 1444426 50 18 9 23 +63.22 ± 39.41 0 1 14 10 0 +130.94 ± 89.9
174204 ncm-dbt-02 1459601 50 13 9 28 +27.85 ± 46.49 0 4 13 8 0 +56.07 ± 96.84
174203 ncm-dbt-03 1457814 50 20 11 19 +63.22 ± 44.25 0 2 12 11 0 +130.94 ± 101.68
174202 ncm-dbt-05 1460193 50 16 8 26 +56.07 ± 43.2 0 2 13 10 0 +115.23 ± 96.38
174201 ncm-dbt-04 1444304 50 18 8 24 +70.42 ± 34.91 0 0 15 10 0 +147.17 ± 81.61

Commit

Commit ID 4766dfc3956f78d853c5e0c4636d6f90fd93df9a
Author Tomasz Sobczyk
Date 2021-12-07 17:08:54 UTC
Optimize FT activation and affine transform for NEON. This patch optimizes the NEON implementation in two ways. The activation layer after the feature transformer is rewritten to make it easier for the compiler to see through dependencies and unroll. This in itself is a minimal, but a positive improvement. Other architectures could benefit from this too in the future. This is not an algorithmic change. The affine transform for large matrices (first layer after FT) on NEON now utilizes the same optimized code path as >=SSSE3, which makes the memory accesses more sequential and makes better use of the available registers, which allows for code that has longer dependency chains. Benchmarks from Redshift#161, profile-build with apple clang george@Georges-MacBook-Air nets % ./stockfish-b82d93 bench 2>&1 | tail -4 (current master) =========================== Total time (ms) : 2167 Nodes searched : 4667742 Nodes/second : 2154011 george@Georges-MacBook-Air nets % ./stockfish-7377b8 bench 2>&1 | tail -4 (this patch) =========================== Total time (ms) : 1842 Nodes searched : 4667742 Nodes/second : 2534061 This is a solid 18% improvement overall, larger in a bench with NNUE-only, not mixed. Improvement is also observed on armv7-neon (Raspberry Pi, and older phones), around 5% speedup. No changes for architectures other than NEON. closes https://github.com/official-stockfish/Stockfish/pull/3837 No functional changes.
Copyright 2011–2024 Next Chess Move LLC