Dev Builds » 20180330-2316

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 06:45:16 2001755 2683 1265 95 1323 +162.38 +/- 9.14
ncm-et-4 04:13:50 1990033 1663 788 58 817 +163.62 +/- 11.63
ncm-et-6 02:31:29 2027047 1000 456 31 513 +157.66 +/- 14.56
ncm-et-7 02:30:48 2027292 1000 471 30 499 +164.5 +/- 14.81
ncm-et-9 06:43:55 2004189 2682 1286 81 1315 +168.09 +/- 9.14
ncm-et-10 06:44:33 1983481 2650 1196 78 1376 +156.34 +/- 8.86
ncm-et-11 02:31:17 2018801 997 463 30 504 +161.62 +/- 14.71
ncm-et-13 06:44:11 1998637 2666 1238 86 1342 +160.68 +/- 9.03
ncm-et-14 02:30:38 2018558 1000 462 37 501 +157.66 +/- 14.84
ncm-et-15 06:45:08 1995250 2659 1191 87 1381 +153.52 +/- 8.86
ncm-et-16 02:30:59 1995475 1000 437 29 534 +150.51 +/- 14.15
20000 9253 642 10105 +160.02 +/- 3.29

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
51588 ncm-et-10 2018-08-28 20:57 00:23:21 1979983 150 63 3 84 +147.19 +/- 35.2
51587 ncm-et-4 2018-08-28 20:55 00:24:43 1989401 163 76 8 79 +154.36 +/- 37.95
51586 ncm-et-15 2018-08-28 20:55 00:25:07 1983420 166 81 6 79 +169.19 +/- 37.8
51585 ncm-et-13 2018-08-28 20:54 00:25:45 1948575 160 72 9 79 +144.61 +/- 37.98
51584 ncm-et-9 2018-08-28 20:52 00:27:43 1989400 177 91 7 79 +179.26 +/- 38.12
51583 ncm-et-3 2018-08-28 20:52 00:28:15 1988928 184 87 6 91 +164.16 +/- 34.95
51582 ncm-et-10 2018-08-28 19:40 01:15:59 1979347 500 211 15 274 +143.89 +/- 19.68
51581 ncm-et-4 2018-08-28 19:39 01:15:42 1991139 500 229 15 256 +158.93 +/- 20.62
51580 ncm-et-13 2018-08-28 19:38 01:14:46 1987040 500 230 14 256 +160.64 +/- 20.59
51579 ncm-et-15 2018-08-28 19:37 01:16:52 1981381 500 205 19 276 +135.76 +/- 19.7
51578 ncm-et-9 2018-08-28 19:36 01:14:31 1987981 500 227 17 256 +155.54 +/- 20.68
51577 ncm-et-3 2018-08-28 19:36 01:14:41 1989558 500 229 21 250 +153.86 +/- 21.1
51576 ncm-et-10 2018-08-28 18:22 01:16:34 1953874 500 248 15 237 +175.44 +/- 21.67
51575 ncm-et-4 2018-08-28 18:21 01:16:17 1989086 500 240 23 237 +161.49 +/- 21.85
51574 ncm-et-13 2018-08-28 18:21 01:16:39 1986722 500 243 16 241 +170.15 +/- 21.47
51573 ncm-et-9 2018-08-28 18:20 01:15:26 1986722 500 239 17 244 +165.8 +/- 21.33
51572 ncm-et-15 2018-08-28 18:20 01:16:22 1983263 500 239 20 241 +163.21 +/- 21.56
51571 ncm-et-3 2018-08-28 18:19 01:15:37 1990032 500 228 17 255 +156.39 +/- 20.73
51570 ncm-et-9 2018-08-28 17:03 01:15:32 1987193 500 237 10 253 +170.15 +/- 20.62
51569 ncm-et-10 2018-08-28 17:03 01:17:50 1946988 500 205 19 276 +135.76 +/- 19.7
51568 ncm-et-15 2018-08-28 17:03 01:15:26 1984204 500 229 13 258 +160.64 +/- 20.45
51567 ncm-et-3 2018-08-28 17:03 01:15:15 1989559 500 233 24 243 +154.7 +/- 21.54
51566 ncm-et-4 2018-08-28 17:03 01:17:08 1990506 500 243 12 245 +173.67 +/- 21.13
51565 ncm-et-13 2018-08-28 17:03 01:16:19 1988296 500 238 16 246 +165.8 +/- 21.19
14167 ncm-et-9 2018-04-01 03:20 00:01:03 2026556 5 3 0 2 +240.65 +/- 570.11
14166 ncm-et-13 2018-04-01 03:19 00:01:18 2028032 6 2 1 3 +58.45 +/- 226.57
14165 ncm-et-10 2018-04-01 02:06 01:14:22 2020021 500 236 9 255 +170.15 +/- 20.48
14164 ncm-et-3 2018-04-01 02:06 01:14:52 2027210 499 239 14 246 +168.8 +/- 21.13
14163 ncm-et-16 2018-04-01 02:05 01:14:22 2021162 500 202 15 283 +136.56 +/- 19.23
14162 ncm-et-6 2018-04-01 02:05 01:15:17 2026884 500 228 22 250 +152.18 +/- 21.12
14161 ncm-et-11 2018-04-01 02:05 01:15:14 2019370 497 226 19 252 +154.07 +/- 20.93
14160 ncm-et-15 2018-04-01 02:04 01:15:47 2017745 493 221 11 261 +158.07 +/- 20.18
14159 ncm-et-14 2018-04-01 02:04 01:15:17 2017421 500 232 18 250 +158.93 +/- 21.03
14158 ncm-et-13 2018-04-01 02:04 01:14:01 2025572 500 219 18 263 +148.02 +/- 20.34
14157 ncm-et-7 2018-04-01 02:04 01:15:48 2026554 500 229 18 253 +156.39 +/- 20.86
14156 ncm-et-9 2018-04-01 02:03 01:15:20 2025572 500 249 11 240 +179.9 +/- 21.37
14155 ncm-et-10 2018-04-01 00:48 01:16:27 2020674 500 233 17 250 +160.64 +/- 21.0
14154 ncm-et-11 2018-04-01 00:48 01:16:03 2018232 500 237 11 252 +169.27 +/- 20.71
14153 ncm-et-3 2018-04-01 00:48 01:16:36 2025245 500 249 13 238 +178.11 +/- 21.55
14152 ncm-et-14 2018-04-01 00:48 01:15:21 2019696 500 230 19 251 +156.39 +/- 21.0
14151 ncm-et-6 2018-04-01 00:48 01:16:12 2027210 500 228 9 263 +163.21 +/- 20.05
14150 ncm-et-9 2018-04-01 00:48 01:14:20 2025900 500 240 19 241 +164.93 +/- 21.54
14149 ncm-et-13 2018-04-01 00:48 01:15:23 2026226 500 234 12 254 +165.8 +/- 20.64
14148 ncm-et-15 2018-04-01 00:48 01:15:34 2021488 500 216 18 266 +145.54 +/- 20.18
14147 ncm-et-7 2018-04-01 00:48 01:15:00 2028030 500 242 12 246 +172.78 +/- 21.07
14146 ncm-et-16 2018-04-01 00:48 01:16:37 1969789 500 235 14 251 +164.93 +/- 20.86

Commit

Commit ID d5e3e7d207538fe3ff0c86df9d9f95912f5266e9
Author Alain SAVARD
Date 2018-03-30 23:16:51 UTC
Candidate Passed Pawn Include some not fully supported levers in the (candidate) passed pawns bitboard, if otherwise unblocked. Maybe levers are usually very short lived, and some inaccuracy in the lever balance for the definition of candidate passed pawns just triggers a deeper search. Here is a example of a case where the patch has an effect on the definition of candidate passers: White c5/e5 pawns, against Black d6 pawn. Let's say we want to test if e5 is a candidate passer. The previous master looks only at files d, e and f (which is already very good) and reject e5 as a candidate. However, the lever d6 is challenged by 2 pawns, so it should not fully count. Indirectly, this patch will view such case (and a few more) to be scored as candidates. STC http://tests.stockfishchess.org/tests/view/5abcd55d0ebc5902926cf1e1 LLR: 2.95 (-2.94,2.94) [0.00,4.00] Total: 16492 W: 3419 L: 3198 D: 9875 LTC http://tests.stockfishchess.org/tests/view/5abce1360ebc5902926cf1e6 LLR: 2.95 (-2.94,2.94) [0.00,4.00] Total: 21156 W: 3201 L: 2990 D: 14965 This was inspired by this test of Jerry Donald Watson, except the case of zero supporting pawns against two levers is excluded, and it seems that not excluding that case is bad, while excluding is it beneficial. See the following tests on fishtest: https://github.com/official-stockfish/Stockfish/pull/1519 http://tests.stockfishchess.org/tests/view/5abccd850ebc5902926cf1dd http://tests.stockfishchess.org/tests/view/5abcdd490ebc5902926cf1e4 Closes https://github.com/official-stockfish/Stockfish/pull/1521 Bench: 5568461 ---- Comments by Jerry Donald Watson: > My thinking as to why this works: > > The evaluation is either called in an interior node or in the qsearch. > The calls at the end of the qsearch are the more important as they > ultimately determine the scoring of each move, whereas the internal > values are mainly used for pruning decisions with a margin. Some strong > engines don't even call the eval at all nodes. Now the whole point of > the qsearch is to find quiet positions where captures do not change the > evaluation of the position with regards to the search bounds - i.e. if > there were good captures they would be tried.* So when a candidate lever > appears in the evaluation at the end of the qsearch, the qsearch has > guaranteed that it cannot just be captured, or if it can, this does not > take the score past the search bounds. Practically this may mean that > the side with the candidate lever has the turn, or perhaps the stopping > lever pawn is pinned, or that side is forced for other reasons to make > some other move (e.g. d6 can only take one of the pawns in the example > above). > > Hence granting the full score for only one lever defender makes some > sense, at least, to me. > > IMO this is also why huge bonuses for possible captures in the evaluation > (e.g. threat on queen and our turn), etc. don't tend to work. Such things > are best left to the search to figure out.
Copyright 2011–2024 Next Chess Move LLC