Dev Builds » 20180310-1104

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 06:10:01 2001866 2440 1057 70 1313 +149.06 +/- 9.0
ncm-et-4 06:11:24 1993382 2449 1116 80 1253 +156.82 +/- 9.33
ncm-et-5 01:56:16 2021896 769 351 20 398 +159.97 +/- 16.45
ncm-et-6 01:56:40 2023773 778 366 36 376 +157.3 +/- 17.3
ncm-et-7 01:56:27 2011894 770 346 24 400 +154.78 +/- 16.46
ncm-et-8 01:55:20 1988835 749 353 21 375 +165.48 +/- 17.06
ncm-et-9 06:12:13 2000689 2440 1085 79 1276 +152.31 +/- 9.21
ncm-et-10 06:13:13 1984868 2402 1119 67 1216 +163.2 +/- 9.45
ncm-et-11 01:55:32 2012177 767 356 14 397 +166.62 +/- 16.36
ncm-et-12 01:55:35 2020510 773 354 22 397 +159.57 +/- 16.52
ncm-et-13 06:13:50 1972688 2418 1088 69 1261 +156.14 +/- 9.24
ncm-et-14 01:55:40 2020185 778 371 20 387 +168.91 +/- 16.78
ncm-et-15 06:14:31 1993392 2467 1119 87 1261 +154.84 +/- 9.31
20000 9081 609 10310 +157.07 +/- 3.24

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
52746 ncm-et-10 2018-09-07 02:56 00:19:34 1979037 127 57 2 68 +161.1 +/- 39.41
52745 ncm-et-13 2018-09-07 02:54 00:21:56 1930693 135 57 2 76 +150.27 +/- 36.79
52744 ncm-et-9 2018-09-07 02:50 00:26:06 1988927 168 83 7 78 +169.44 +/- 38.26
52743 ncm-et-3 2018-09-07 02:48 00:28:02 1991297 183 85 3 95 +167.57 +/- 33.5
52742 ncm-et-15 2018-09-07 02:47 00:28:37 1982008 190 83 5 102 +151.57 +/- 32.39
52741 ncm-et-4 2018-09-07 02:46 00:29:34 1989876 196 85 10 101 +140.07 +/- 33.29
52740 ncm-et-10 2018-09-07 01:37 01:17:43 1981694 500 234 15 251 +163.21 +/- 20.89
52739 ncm-et-13 2018-09-07 01:34 01:18:23 1942908 500 225 14 261 +156.39 +/- 20.33
52738 ncm-et-9 2018-09-07 01:33 01:15:26 1987981 500 221 14 265 +153.02 +/- 20.12
52737 ncm-et-3 2018-09-07 01:33 01:13:51 1992405 500 216 17 267 +146.36 +/- 20.1
52736 ncm-et-15 2018-09-07 01:31 01:15:26 1981538 500 229 17 254 +157.24 +/- 20.78
52735 ncm-et-4 2018-09-07 01:28 01:16:35 1990664 500 240 8 252 +174.55 +/- 20.6
52734 ncm-et-13 2018-09-07 00:18 01:15:39 1976779 500 233 14 253 +163.21 +/- 20.75
52733 ncm-et-10 2018-09-07 00:17 01:19:22 1952880 500 238 10 252 +171.02 +/- 20.68
52732 ncm-et-9 2018-09-07 00:15 01:16:59 1988611 500 218 18 264 +147.19 +/- 20.28
52731 ncm-et-15 2018-09-07 00:14 01:15:15 1979972 500 245 17 238 +171.02 +/- 21.66
52730 ncm-et-3 2018-09-07 00:14 01:17:55 1990190 500 210 21 269 +138.18 +/- 20.1
52729 ncm-et-4 2018-09-07 00:13 01:14:13 1990032 500 221 15 264 +152.18 +/- 20.2
52728 ncm-et-15 2018-09-06 22:57 01:16:37 1978254 500 208 19 273 +138.18 +/- 19.85
52727 ncm-et-4 2018-09-06 22:56 01:15:40 1990507 500 233 14 253 +163.21 +/- 20.75
52726 ncm-et-3 2018-09-06 22:56 01:16:18 1988613 500 207 12 281 +143.07 +/- 19.23
52725 ncm-et-10 2018-09-06 22:56 01:19:30 1952779 500 254 11 235 +184.42 +/- 21.66
52724 ncm-et-9 2018-09-06 22:56 01:17:37 1988614 500 219 19 262 +147.19 +/- 20.41
52723 ncm-et-13 2018-09-06 22:56 01:20:14 1934281 500 224 19 257 +151.35 +/- 20.68
7228 ncm-et-3 2018-03-11 04:43 00:38:18 2024591 257 108 4 145 +149.13 +/- 26.57
7227 ncm-et-8 2018-03-11 04:43 00:38:44 1986020 249 125 10 114 +173.6 +/- 31.6
7226 ncm-et-11 2018-03-11 04:42 00:39:48 2005310 267 129 3 135 +178.07 +/- 28.05
7225 ncm-et-4 2018-03-11 04:42 00:40:01 1978531 253 95 10 148 +121.44 +/- 26.6
7224 ncm-et-12 2018-03-11 04:42 00:39:58 2021162 273 125 7 141 +160.74 +/- 27.7
7223 ncm-et-5 2018-03-11 04:41 00:41:09 2021651 269 132 8 129 +173.21 +/- 29.37
7222 ncm-et-7 2018-03-11 04:41 00:41:03 2000173 270 131 9 130 +169.21 +/- 29.31
7221 ncm-et-9 2018-03-11 04:40 00:41:13 2025086 272 128 8 136 +164.58 +/- 28.43
7220 ncm-et-14 2018-03-11 04:40 00:41:27 2020999 278 132 3 143 +174.56 +/- 27.15
7219 ncm-et-6 2018-03-11 04:40 00:41:56 2022629 278 143 11 124 +179.37 +/- 30.35
7218 ncm-et-10 2018-03-11 04:40 00:42:06 2023285 275 117 13 145 +138.26 +/- 27.6
7217 ncm-et-13 2018-03-11 04:39 00:42:22 2025735 283 126 10 147 +151.3 +/- 27.28
7216 ncm-et-15 2018-03-11 04:39 00:42:22 2018233 277 135 7 135 +173.71 +/- 28.56
7215 ncm-et-3 2018-03-11 03:26 01:15:37 2024100 500 231 13 256 +162.35 +/- 20.56
7214 ncm-et-4 2018-03-11 03:25 01:15:21 2020683 500 242 23 235 +163.21 +/- 21.96
7213 ncm-et-8 2018-03-11 03:25 01:16:36 1991650 500 228 11 261 +161.49 +/- 20.23
7212 ncm-et-12 2018-03-11 03:25 01:15:37 2019859 500 229 15 256 +158.93 +/- 20.62
7211 ncm-et-11 2018-03-11 03:25 01:15:44 2019045 500 227 11 262 +160.64 +/- 20.18
7210 ncm-et-14 2018-03-11 03:25 01:14:13 2019371 500 239 17 244 +165.8 +/- 21.33
7209 ncm-et-9 2018-03-11 03:24 01:14:52 2024917 500 216 13 271 +149.68 +/- 19.78
7208 ncm-et-5 2018-03-11 03:24 01:15:07 2022141 500 219 12 269 +153.02 +/- 19.85
7207 ncm-et-7 2018-03-11 03:24 01:15:24 2023616 500 215 15 270 +147.19 +/- 19.89
7206 ncm-et-6 2018-03-11 03:24 01:14:44 2024918 500 223 25 252 +145.54 +/- 21.07
7205 ncm-et-10 2018-03-11 03:24 01:14:58 2019536 500 219 16 265 +149.68 +/- 20.18
7204 ncm-et-13 2018-03-11 03:23 01:15:16 2025735 500 223 10 267 +158.08 +/- 19.88
7203 ncm-et-15 2018-03-11 03:22 01:16:14 2020347 500 219 22 259 +144.72 +/- 20.64

Commit

Commit ID 5dc381a56692a378674fe61e9ee2d64900aa3aa6
Author protonspring
Date 2018-03-10 11:04:03 UTC
Implement an old Russian proverb "Loose pieces drop, in blitz keep everything protected" Adding a small S(2,2) bonus for knights, bishops, rooks, and queens that are "connected" to each other (in the sense that they are under attack by our own pieces) apparently is a good thing. It probably helps the pieces work together a bit better. STC LLR: 2.96 (-2.94,2.94) [0.00,5.00] Total: 12317 W: 2655 L: 2467 D: 7195 http://tests.stockfishchess.org/tests/view/5aa2d86b0ebc590297cb6474 LTC LLR: 2.96 (-2.94,2.94) [0.00,5.00] Total: 35725 W: 5516 L: 5263 D: 24946 http://tests.stockfishchess.org/tests/view/5aa2fc6f0ebc590297cb64a8 How to continue from there (by Stefan Geschwentner)? • First we should identify all other eval terms which have an overlap with new connectivity bonus (like the outpost bonus). A simple way would be subtract the connectivity bonus from them and look if this better, or use a SPSA session for these terms. • Tuning Connectivity himself with SPSA seems not so promising because of the small range which is useful. Here manual testing changes of Connectivity like +-1 seems better. • The eg value is more important because in endgame the position gets more open and so attacks on pieces are easier. Another important point is that when defending/fortress-like positions each defending piece needs a protection, otherwise attacks on them can break defense. Closes https://github.com/official-stockfish/Stockfish/pull/1474 Bench: 5318575
Copyright 2011–2024 Next Chess Move LLC