Dev Builds » 20170810-1746

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 06:02:48 2001834 2473 888 80 1505 +117.84 +/- 8.18
ncm-et-4 06:01:57 1996707 2419 877 95 1447 +116.49 +/- 8.43
ncm-et-5 01:53:39 2021490 786 265 29 492 +107.64 +/- 14.23
ncm-et-6 01:53:22 2026390 778 280 24 474 +118.74 +/- 14.56
ncm-et-7 01:53:10 2013193 764 276 31 457 +115.49 +/- 15.03
ncm-et-8 01:53:35 1969971 752 292 31 429 +125.81 +/- 15.71
ncm-et-9 06:02:36 2000464 2451 855 104 1492 +109.99 +/- 8.28
ncm-et-10 06:02:36 1989922 2429 888 92 1449 +118.22 +/- 8.42
ncm-et-11 01:53:33 2019047 763 284 22 457 +124.35 +/- 14.88
ncm-et-12 01:53:41 2018722 767 280 32 455 +116.52 +/- 15.11
ncm-et-13 06:02:49 1982222 2425 846 80 1499 +113.63 +/- 8.16
ncm-et-14 01:53:27 2020022 763 269 33 461 +111.1 +/- 14.95
ncm-et-15 06:02:51 1989027 2430 843 89 1498 +111.48 +/- 8.19
20000 7143 742 12115 +115.25 +/- 2.9

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
56343 ncm-et-15 2018-10-04 05:33 00:22:32 1981068 151 63 3 85 +146.1 +/- 34.94
56342 ncm-et-4 2018-10-04 05:31 00:24:26 1990032 160 58 3 99 +124.5 +/- 31.43
56341 ncm-et-13 2018-10-04 05:31 00:24:39 1987513 162 49 4 109 +99.11 +/- 29.07
56340 ncm-et-10 2018-10-04 05:30 00:25:26 1982008 162 63 6 93 +127.7 +/- 33.71
56339 ncm-et-9 2018-10-04 05:28 00:27:38 1987666 176 64 5 107 +121.15 +/- 30.69
56338 ncm-et-3 2018-10-04 05:27 00:28:07 1990665 189 66 9 114 +108.14 +/- 30.23
56337 ncm-et-15 2018-10-04 04:16 01:15:08 1981697 500 166 23 311 +102.22 +/- 18.07
56336 ncm-et-10 2018-10-04 04:15 01:13:41 1981852 500 182 21 297 +116.0 +/- 18.7
56335 ncm-et-4 2018-10-04 04:14 01:15:25 1988769 500 186 12 302 +126.18 +/- 18.19
56334 ncm-et-9 2018-10-04 04:14 01:12:59 1989400 500 183 23 294 +115.23 +/- 18.9
56333 ncm-et-13 2018-10-04 04:13 01:16:00 1886177 500 183 15 302 +121.46 +/- 18.29
56332 ncm-et-3 2018-10-04 04:13 01:13:33 1990664 500 173 16 311 +112.91 +/- 17.88
56331 ncm-et-15 2018-10-04 03:00 01:15:15 1982166 500 178 20 302 +113.68 +/- 18.43
56330 ncm-et-10 2018-10-04 03:00 01:13:44 1982009 500 176 17 307 +114.45 +/- 18.1
56329 ncm-et-9 2018-10-04 02:59 01:13:40 1986877 500 155 26 319 +91.71 +/- 17.75
56328 ncm-et-4 2018-10-04 02:58 01:15:21 1989245 500 174 15 311 +114.45 +/- 17.85
56327 ncm-et-3 2018-10-04 02:58 01:13:37 1988769 500 179 16 305 +117.55 +/- 18.17
56326 ncm-et-13 2018-10-04 02:58 01:14:43 1988296 500 171 18 311 +109.83 +/- 17.94
56325 ncm-et-4 2018-10-04 01:43 01:13:59 1989875 500 184 16 300 +121.46 +/- 18.42
56324 ncm-et-9 2018-10-04 01:43 01:15:05 1986879 500 176 19 305 +112.91 +/- 18.26
56323 ncm-et-3 2018-10-04 01:43 01:14:01 1988454 500 178 17 305 +116.0 +/- 18.2
56322 ncm-et-10 2018-10-04 01:43 01:16:08 1948404 500 173 20 307 +109.83 +/- 18.19
56321 ncm-et-15 2018-10-04 01:43 01:16:22 1947237 500 166 17 317 +106.78 +/- 17.62
56320 ncm-et-13 2018-10-04 01:43 01:13:49 1979876 500 155 12 333 +102.22 +/- 16.67
8906 ncm-et-8 2018-03-16 09:31 00:36:59 1975962 252 89 9 154 +114.24 +/- 25.67
8905 ncm-et-4 2018-03-16 09:29 00:37:35 2023446 259 97 19 143 +107.98 +/- 27.89
8904 ncm-et-13 2018-03-16 09:29 00:38:33 2026882 263 93 13 157 +109.14 +/- 25.84
8903 ncm-et-11 2018-03-16 09:29 00:38:51 2020023 263 92 10 161 +112.06 +/- 25.13
8902 ncm-et-12 2018-03-16 09:29 00:39:01 2020184 267 93 12 162 +108.83 +/- 25.25
8901 ncm-et-14 2018-03-16 09:28 00:39:05 2020999 263 92 9 162 +113.52 +/- 24.93
8900 ncm-et-7 2018-03-16 09:28 00:39:29 2025735 264 84 15 165 +92.96 +/- 24.96
8899 ncm-et-10 2018-03-16 09:28 00:39:54 2022303 267 99 9 159 +121.88 +/- 25.42
8898 ncm-et-6 2018-03-16 09:27 00:39:48 2026390 278 88 10 180 +100.17 +/- 23.24
8897 ncm-et-3 2018-03-16 09:27 00:40:14 2025737 284 94 8 182 +108.61 +/- 23.05
8896 ncm-et-15 2018-03-16 09:27 00:40:30 2021488 279 93 10 176 +106.58 +/- 23.74
8895 ncm-et-9 2018-03-16 09:27 00:40:22 2027374 275 98 13 164 +111.02 +/- 25.26
8894 ncm-et-5 2018-03-16 09:26 00:41:38 2019371 286 100 7 179 +117.23 +/- 23.37
8893 ncm-et-4 2018-03-16 08:13 01:15:11 1998877 500 178 30 292 +106.01 +/- 19.15
8892 ncm-et-7 2018-03-16 08:13 01:13:41 2000652 500 192 16 292 +127.76 +/- 18.81
8891 ncm-et-14 2018-03-16 08:13 01:14:22 2019045 500 177 24 299 +109.83 +/- 18.68
8890 ncm-et-3 2018-03-16 08:12 01:13:16 2026720 500 198 14 288 +134.15 +/- 18.95
8889 ncm-et-9 2018-03-16 08:12 01:12:52 2024592 500 179 18 303 +116.0 +/- 18.33
8888 ncm-et-10 2018-03-16 08:12 01:13:43 2022958 500 195 19 286 +127.76 +/- 19.2
8887 ncm-et-15 2018-03-16 08:12 01:13:04 2020510 500 177 16 307 +116.0 +/- 18.07
8886 ncm-et-13 2018-03-16 08:12 01:15:05 2024590 500 195 18 287 +128.55 +/- 19.12
8885 ncm-et-6 2018-03-16 08:12 01:13:34 2026391 500 192 14 294 +129.35 +/- 18.65
8884 ncm-et-12 2018-03-16 08:12 01:14:40 2017260 500 187 20 293 +120.67 +/- 18.88
8883 ncm-et-11 2018-03-16 08:12 01:14:42 2018071 500 192 12 296 +130.94 +/- 18.49
8882 ncm-et-5 2018-03-16 08:12 01:12:01 2023609 500 165 22 313 +102.22 +/- 17.95
8881 ncm-et-8 2018-03-16 08:12 01:16:36 1963980 500 203 22 275 +131.74 +/- 19.82

Commit

Commit ID 5410424e3d036b43715c7989aa99e449cdcde18e
Author Joost VandeVondele
Date 2017-08-10 17:46:46 UTC
Fix a race on Limits::ponder Limits::ponder was used as a signal between uci and search threads, but is not an atomic variable, leading to the following race as flagged by a sanitized binary. Expect input: ``` spawn ./stockfish send "uci\n" expect "uciok" send "setoption name Ponder value true\n" send "go wtime 4000 btime 4000\n" expect "bestmove" send "position startpos e2e4 d7d5\n" send "go wtime 4000 btime 4000 ponder\n" sleep 0.01 send "ponderhit\n" expect "bestmove" send "quit\n" expect eof ``` Race: ``` WARNING: ThreadSanitizer: data race (pid=7191) Read of size 4 at 0x0000005c2260 by thread T1: Previous write of size 4 at 0x0000005c2260 by main thread: Location is global 'Search::Limits' of size 88 at 0x0000005c2220 (stockfish+0x0000005c2260) ``` The reason of teh race is that ponder is not just set in UCI go() assignment but also is signaled by an async ponderhit in uci.cpp: else if (token == "ponderhit") Search::Limits.ponder = 0; // Switch to normal search The fix is to add an atomic bool to the threads structure to signal the ponder status, letting Search::Limits to reflect just what was passed to 'go'. No functional change.
Copyright 2011–2024 Next Chess Move LLC