Dev Builds » 20190620-1439

Use this dev build

NCM plays each Stockfish dev build 20,000 times against Stockfish 7. This yields an approximate Elo difference and establishes confidence in the strength of the dev builds.

Summary

Host Duration Avg Base NPS Games Wins Losses Draws Elo
ncm-et-3 08:31:50 1970490 3340 1981 69 1290 +226.24 +/- 9.38
ncm-et-4 08:32:03 1956560 3308 1966 77 1265 +225.51 +/- 9.49
ncm-et-9 08:31:46 1977085 3317 1941 78 1298 +220.71 +/- 9.36
ncm-et-10 08:31:47 1962289 3352 2009 77 1266 +228.27 +/- 9.49
ncm-et-13 08:32:11 1976254 3346 1983 72 1291 +225.55 +/- 9.38
ncm-et-15 08:31:36 1970628 3337 1996 72 1269 +228.37 +/- 9.47
20000 11876 445 7679 +225.77 +/- 3.85

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
86102 ncm-et-4 2019-06-21 16:42 00:48:47 1917428 308 187 8 113 +230.78 +/- 32.02
86101 ncm-et-9 2019-06-21 16:42 00:48:55 1974082 317 191 3 123 +237.08 +/- 30.25
86100 ncm-et-15 2019-06-21 16:39 00:51:56 1972504 337 194 9 134 +214.33 +/- 29.26
86099 ncm-et-3 2019-06-21 16:37 00:53:25 1966153 340 193 8 139 +211.93 +/- 28.62
86098 ncm-et-13 2019-06-21 16:36 00:54:20 1980986 346 208 6 132 +232.17 +/- 29.41
86097 ncm-et-10 2019-06-21 16:36 00:54:45 1956506 352 202 13 137 +208.4 +/- 29.09
86096 ncm-et-9 2019-06-21 15:24 01:16:13 1970807 500 283 15 202 +207.95 +/- 23.81
86095 ncm-et-4 2019-06-21 15:23 01:17:39 1963532 500 296 11 193 +224.97 +/- 24.33
86094 ncm-et-15 2019-06-21 15:21 01:16:30 1974047 500 315 11 174 +245.2 +/- 25.77
86093 ncm-et-3 2019-06-21 15:19 01:16:47 1974389 500 310 10 180 +240.82 +/- 25.27
86092 ncm-et-10 2019-06-21 15:19 01:15:36 1962339 500 300 15 185 +224.97 +/- 25.0
86091 ncm-et-13 2019-06-21 15:19 01:16:17 1984244 500 301 10 189 +231.21 +/- 24.59
86090 ncm-et-4 2019-06-21 14:06 01:15:41 1981935 500 282 7 211 +214.85 +/- 22.98
86089 ncm-et-9 2019-06-21 14:06 01:17:34 1979040 500 292 12 196 +219.87 +/- 24.15
86088 ncm-et-15 2019-06-21 14:03 01:17:05 1968175 500 294 9 197 +224.97 +/- 23.99
86087 ncm-et-13 2019-06-21 14:02 01:15:30 1966305 500 285 15 200 +209.91 +/- 23.94
86086 ncm-et-10 2019-06-21 14:01 01:16:54 1961236 500 300 9 191 +231.21 +/- 24.42
86085 ncm-et-3 2019-06-21 14:01 01:17:24 1972777 500 279 8 213 +210.89 +/- 22.89
86084 ncm-et-4 2019-06-21 12:48 01:17:08 1963839 500 296 15 189 +220.88 +/- 24.71
86083 ncm-et-9 2019-06-21 12:47 01:17:52 1976552 500 294 9 197 +224.97 +/- 23.99
86082 ncm-et-13 2019-06-21 12:46 01:15:13 1985049 500 301 11 188 +230.16 +/- 24.69
86081 ncm-et-15 2019-06-21 12:45 01:16:53 1965381 500 310 6 184 +245.2 +/- 24.83
86080 ncm-et-10 2019-06-21 12:45 01:15:35 1962612 500 304 9 187 +235.45 +/- 24.71
86079 ncm-et-3 2019-06-21 12:42 01:17:19 1963230 500 293 11 196 +221.9 +/- 24.12
86078 ncm-et-4 2019-06-21 11:29 01:18:24 1911860 500 306 10 184 +236.51 +/- 24.97
86077 ncm-et-9 2019-06-21 11:29 01:17:10 1983147 500 292 10 198 +221.9 +/- 23.95
86076 ncm-et-13 2019-06-21 11:28 01:17:01 1977474 500 294 8 198 +226.0 +/- 23.89
86075 ncm-et-15 2019-06-21 11:28 01:16:03 1970801 500 294 12 194 +221.9 +/- 24.29
86074 ncm-et-10 2019-06-21 11:28 01:15:46 1964151 500 301 13 186 +228.08 +/- 24.89
86073 ncm-et-3 2019-06-21 11:25 01:15:53 1963840 500 305 12 183 +233.32 +/- 25.09
86072 ncm-et-4 2019-06-21 10:10 01:17:31 1965234 500 308 14 178 +234.38 +/- 25.51
86071 ncm-et-9 2019-06-21 10:10 01:17:27 1969878 500 308 16 176 +232.26 +/- 25.7
86070 ncm-et-15 2019-06-21 10:10 01:17:12 1969730 500 299 9 192 +230.16 +/- 24.35
86069 ncm-et-10 2019-06-21 10:10 01:17:10 1961715 500 297 11 192 +226.0 +/- 24.4
86068 ncm-et-13 2019-06-21 10:09 01:17:45 1971409 500 293 12 195 +220.88 +/- 24.22
86067 ncm-et-3 2019-06-21 10:09 01:15:12 1980337 500 290 10 200 +219.87 +/- 23.82
86066 ncm-et-15 2019-06-21 08:53 01:15:57 1973760 500 290 16 194 +213.85 +/- 24.38
86065 ncm-et-10 2019-06-21 08:52 01:16:01 1967464 500 305 7 188 +238.66 +/- 24.57
86064 ncm-et-9 2019-06-21 08:52 01:16:35 1986095 500 281 13 206 +207.95 +/- 23.5
86063 ncm-et-3 2019-06-21 08:52 01:15:50 1972707 500 311 10 179 +241.91 +/- 25.35
86062 ncm-et-13 2019-06-21 08:52 01:16:05 1968313 500 301 10 189 +231.21 +/- 24.59
86061 ncm-et-4 2019-06-21 08:52 01:16:53 1992095 500 291 12 197 +218.86 +/- 24.08

Commit

Commit ID 8bf21a723e77399382f05c75e45201882772684a
Author Miguel Lahoz
Date 2019-06-20 14:39:22 UTC
Change multi-cut pruning condition Use comparison of eval with beta to predict potential cutNodes. This allows multi-cut pruning to also prune possibly mislabeled Pv and NonPv nodes. STC: LLR: 2.95 (-2.94,2.94) [0.50,4.50] Total: 54305 W: 12302 L: 11867 D: 30136 http://tests.stockfishchess.org/tests/view/5d048ba50ebc5925cf0a15e8 LTC: LLR: 2.95 (-2.94,2.94) [0.00,3.50] Total: 189512 W: 32620 L: 31904 D: 124988 http://tests.stockfishchess.org/tests/view/5d04bf740ebc5925cf0a17f0 Normally I would think such changes are risky, specially for PvNodes, but after trying a few other versions, it seems this version is more sound than I initially thought. Aside from this, a small funtional change is made to return singularBeta instead of beta to be more consistent with the fail-soft logic used in other parts of search. ============================= How to continue from there ? We could try to audit other parts of the search where the "cutNode" variable is used, and try to use dynamic info based on heuristic eval rather than on this variable, to check if the idea behind this patch could also be applied successfuly. Bench: 3503788
Copyright 2011–2024 Next Chess Move LLC