Dev Builds » 20180313-2221

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 06:11:43 2001340 2458 1103 71 1284 +155.48 +/- 9.16
ncm-et-4 06:11:49 2000721 2440 1103 83 1254 +154.72 +/- 9.32
ncm-et-5 01:56:16 2019457 763 359 13 391 +169.92 +/- 16.5
ncm-et-6 01:56:01 2024590 770 343 21 406 +154.78 +/- 16.25
ncm-et-7 01:55:42 2003908 771 352 26 393 +156.74 +/- 16.69
ncm-et-8 01:56:04 1962061 746 357 21 368 +168.58 +/- 17.26
ncm-et-9 06:11:38 2001885 2453 1076 70 1307 +151.39 +/- 9.04
ncm-et-10 06:11:16 1993190 2427 1106 68 1253 +158.8 +/- 9.28
ncm-et-11 01:56:11 2016448 775 374 19 382 +171.93 +/- 16.9
ncm-et-12 01:55:50 2001874 767 381 19 367 +178.1 +/- 17.32
ncm-et-13 06:11:33 1977360 2424 1128 82 1214 +160.43 +/- 9.51
ncm-et-14 01:56:05 2019373 777 337 26 414 +147.3 +/- 16.13
ncm-et-15 06:11:33 1988138 2429 1161 56 1212 +170.55 +/- 9.45
20000 9180 575 10245 +159.89 +/- 3.25

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
52602 ncm-et-13 2018-09-06 00:52 00:22:16 1968547 142 75 2 65 +197.44 +/- 41.3
52601 ncm-et-15 2018-09-06 00:50 00:24:07 1982165 153 79 2 72 +192.37 +/- 39.03
52600 ncm-et-10 2018-09-06 00:50 00:24:40 1982477 160 77 3 80 +173.89 +/- 36.86
52599 ncm-et-4 2018-09-06 00:47 00:27:46 1991614 179 77 5 97 +148.12 +/- 33.2
52598 ncm-et-9 2018-09-06 00:47 00:27:46 1988927 183 74 1 108 +146.74 +/- 30.09
52597 ncm-et-3 2018-09-06 00:46 00:27:58 1989401 183 87 7 89 +162.85 +/- 35.54
52596 ncm-et-13 2018-09-05 23:34 01:17:07 1930395 500 231 13 256 +162.35 +/- 20.56
52595 ncm-et-15 2018-09-05 23:31 01:18:21 1944854 500 257 15 228 +183.51 +/- 22.19
52594 ncm-et-10 2018-09-05 23:31 01:17:53 1981694 500 234 12 254 +165.8 +/- 20.64
52593 ncm-et-3 2018-09-05 23:30 01:15:15 1989401 500 232 15 253 +161.49 +/- 20.78
52592 ncm-et-9 2018-09-05 23:30 01:16:06 1989558 500 241 17 242 +167.53 +/- 21.44
52591 ncm-et-4 2018-09-05 23:28 01:17:27 1987193 500 222 20 258 +148.85 +/- 20.65
52590 ncm-et-10 2018-09-05 22:14 01:15:30 1981538 500 223 16 261 +153.02 +/- 20.39
52589 ncm-et-13 2018-09-05 22:13 01:19:29 1956704 500 252 22 226 +172.78 +/- 22.45
52588 ncm-et-3 2018-09-05 22:13 01:15:57 1989716 500 235 12 253 +166.66 +/- 20.69
52587 ncm-et-15 2018-09-05 22:12 01:17:23 1979194 500 231 12 257 +163.21 +/- 20.47
52586 ncm-et-9 2018-09-05 22:12 01:16:17 1988927 500 226 16 258 +155.54 +/- 20.54
52585 ncm-et-4 2018-09-05 22:11 01:16:06 1989400 500 233 18 249 +159.78 +/- 21.08
52584 ncm-et-10 2018-09-05 20:56 01:17:14 1974852 500 228 18 254 +155.54 +/- 20.81
52583 ncm-et-15 2018-09-05 20:56 01:15:38 1984361 500 239 15 246 +167.53 +/- 21.16
52582 ncm-et-3 2018-09-05 20:56 01:16:28 1990507 500 198 17 285 +131.74 +/- 19.19
52581 ncm-et-13 2018-09-05 20:56 01:16:42 1958848 500 221 20 259 +148.02 +/- 20.6
52580 ncm-et-9 2018-09-05 20:56 01:15:35 1991772 500 216 17 267 +146.36 +/- 20.1
52579 ncm-et-4 2018-09-05 20:56 01:14:28 1990032 500 222 14 264 +153.86 +/- 20.17
8217 ncm-et-8 2018-03-14 06:13 00:38:25 1976905 246 114 9 123 +158.44 +/- 30.04
8216 ncm-et-4 2018-03-14 06:11 00:39:44 2020683 261 118 7 136 +157.78 +/- 28.19
8215 ncm-et-5 2018-03-14 06:11 00:40:01 2020673 263 130 5 128 +179.58 +/- 29.2
8214 ncm-et-9 2018-03-14 06:11 00:40:06 2027376 270 113 4 153 +148.73 +/- 25.81
8213 ncm-et-6 2018-03-14 06:11 00:40:19 2024100 270 117 6 147 +151.81 +/- 26.78
8212 ncm-et-10 2018-03-14 06:10 00:40:43 2018720 267 118 10 139 +149.05 +/- 28.09
8211 ncm-et-7 2018-03-14 06:10 00:41:00 1981099 271 123 10 138 +154.27 +/- 28.28
8210 ncm-et-15 2018-03-14 06:09 00:41:30 2019534 276 136 4 136 +180.92 +/- 28.15
8209 ncm-et-12 2018-03-14 06:09 00:41:28 1983727 267 137 9 121 +181.43 +/- 30.6
8208 ncm-et-11 2018-03-14 06:09 00:41:50 2015312 275 132 6 137 +171.98 +/- 28.18
8207 ncm-et-3 2018-03-14 06:09 00:42:06 2025408 275 140 6 129 +185.0 +/- 29.29
8206 ncm-et-13 2018-03-14 06:09 00:42:09 2024427 282 120 9 153 +144.56 +/- 26.47
8205 ncm-et-14 2018-03-14 06:08 00:42:30 2021814 277 114 10 153 +137.15 +/- 26.45
8204 ncm-et-9 2018-03-14 04:54 01:15:48 2024753 500 206 15 279 +139.81 +/- 19.43
8203 ncm-et-6 2018-03-14 04:54 01:15:42 2025081 500 226 15 259 +156.39 +/- 20.46
8202 ncm-et-10 2018-03-14 04:54 01:15:16 2019859 500 226 9 265 +161.49 +/- 19.95
8201 ncm-et-7 2018-03-14 04:54 01:14:42 2026718 500 229 16 255 +158.08 +/- 20.7
8200 ncm-et-8 2018-03-14 04:54 01:17:39 1947217 500 243 12 245 +173.67 +/- 21.13
8199 ncm-et-12 2018-03-14 04:54 01:14:22 2020021 500 244 10 246 +176.33 +/- 21.0
8198 ncm-et-14 2018-03-14 04:54 01:13:35 2016933 500 223 16 261 +153.02 +/- 20.39
8197 ncm-et-4 2018-03-14 04:54 01:16:18 2025408 500 231 19 250 +157.24 +/- 21.05
8196 ncm-et-11 2018-03-14 04:54 01:14:21 2017584 500 242 13 245 +171.9 +/- 21.16
8195 ncm-et-15 2018-03-14 04:54 01:14:34 2018720 500 219 8 273 +156.39 +/- 19.49
8194 ncm-et-3 2018-03-14 04:54 01:13:59 2023610 500 211 14 275 +144.72 +/- 19.6
8193 ncm-et-13 2018-03-14 04:54 01:13:50 2025244 500 229 16 255 +158.08 +/- 20.7
8192 ncm-et-5 2018-03-14 04:54 01:16:15 2018241 500 229 8 263 +164.93 +/- 20.01

Commit

Commit ID c5f6bd517c68e16c3ead7892e1d83a6b1bb89b69
Author Jerry Donald Watson
Date 2018-03-13 22:21:46 UTC
Tweak probcut margin with 'improving' flag Adjust ProbCut rBeta by whether the score is improving, and also set improving to false when in check. More precisely, this patch has two parts: 1) the increased beta threshold for ProbCut is now adjusted based on whether the score is improving 2) when in check, improving is always set to false. Co-authored by Joost VandeVondele (@vondele) and Bill Henry (@VoyagerOne). STC: LLR: 2.96 (-2.94,2.94) [0.00,5.00] Total: 13480 W: 2840 L: 2648 D: 7992 http://tests.stockfishchess.org/tests/view/5aa693fe0ebc59029781004c LTC: LLR: 2.97 (-2.94,2.94) [0.00,5.00] Total: 25895 W: 4099 L: 3880 D: 17916 http://tests.stockfishchess.org/tests/view/5aa6ac940ebc59029781006e In terms of opportunities for future work opened up by this patch, the ProbCut rBeta formula could probably be tuned to gain more Elo. Closes https://github.com/official-stockfish/Stockfish/pull/1485 Bench: 5328254
Copyright 2011–2024 Next Chess Move LLC