Dev Builds » 20190927-1046

You are viewing an old NCM Stockfish dev build test. You may find the most recent dev build tests using Stockfish 15 as the baseline here.

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 08:30:58 1970866 3333 2027 60 1246 +235.53 ± 9.55
ncm-et-4 08:28:48 1973545 3311 1998 57 1256 +233.44 ± 9.49
ncm-et-9 08:32:15 1967775 3365 2105 67 1193 +243.91 ± 9.79
ncm-et-10 08:23:37 1963061 3298 1943 47 1308 +227.5 ± 9.26
ncm-et-13 08:31:11 1976469 3349 2000 75 1274 +227.45 ± 9.45
ncm-et-15 08:31:23 1972066 3344 2034 66 1244 +234.66 ± 9.57
20000 12107 372 7521 +233.72 ± 3.88

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
97708 ncm-et-10 2019-09-27 22:29 00:45:03 1985052 298 164 6 128 +205.13 ± 29.65
97707 ncm-et-4 2019-09-27 22:26 00:48:13 1969860 311 196 5 110 +248.61 ± 32.36
97706 ncm-et-3 2019-09-27 22:24 00:50:11 1963691 333 213 4 116 +256.23 ± 31.44
97705 ncm-et-15 2019-09-27 22:22 00:52:42 1974828 344 202 7 135 +223.36 ± 29.08
97704 ncm-et-13 2019-09-27 22:21 00:53:38 1969410 349 214 9 126 +234.06 ± 30.33
97703 ncm-et-9 2019-09-27 22:18 00:55:53 1968468 365 219 8 138 +229.16 ± 28.85
97702 ncm-et-10 2019-09-27 21:12 01:16:04 1963840 500 303 11 186 +232.26 ± 24.84
97701 ncm-et-4 2019-09-27 21:08 01:17:12 1967418 500 282 8 210 +213.85 ± 23.08
97700 ncm-et-3 2019-09-27 21:06 01:16:42 1973443 500 306 4 190 +243.0 ± 24.3
97699 ncm-et-15 2019-09-27 21:04 01:16:43 1972654 500 328 11 161 +259.91 ± 26.87
97698 ncm-et-13 2019-09-27 21:03 01:16:30 1988647 500 293 10 197 +222.92 ± 24.02
97697 ncm-et-9 2019-09-27 21:01 01:16:18 1966655 500 302 10 188 +232.26 ± 24.67
97696 ncm-et-10 2019-09-27 19:52 01:18:21 1927336 500 293 7 200 +226.0 ± 23.71
97695 ncm-et-4 2019-09-27 19:51 01:15:31 1971726 500 317 8 175 +250.77 ± 25.61
97694 ncm-et-3 2019-09-27 19:48 01:17:15 1973438 500 290 8 202 +221.9 ± 23.61
97693 ncm-et-15 2019-09-27 19:46 01:16:43 1972183 500 301 11 188 +230.16 ± 24.69
97692 ncm-et-13 2019-09-27 19:45 01:16:59 1971104 500 293 9 198 +223.94 ± 23.92
97691 ncm-et-9 2019-09-27 19:43 01:16:38 1970169 500 296 8 196 +228.08 ± 24.03
97690 ncm-et-10 2019-09-27 18:33 01:18:16 1956345 500 317 5 178 +254.16 ± 25.26
97689 ncm-et-4 2019-09-27 18:33 01:17:19 1968006 500 302 6 192 +236.51 ± 24.24
97688 ncm-et-3 2019-09-27 18:30 01:16:43 1968479 500 301 12 187 +229.12 ± 24.79
97687 ncm-et-13 2019-09-27 18:28 01:15:29 1975305 500 300 14 186 +226.0 ± 24.91
97686 ncm-et-15 2019-09-27 18:28 01:16:35 1971879 500 306 10 184 +236.51 ± 24.97
97685 ncm-et-9 2019-09-27 18:26 01:15:50 1964303 500 312 9 179 +244.1 ± 25.32
97684 ncm-et-10 2019-09-27 17:16 01:15:47 1975620 500 291 4 205 +227.04 ± 23.25
97683 ncm-et-4 2019-09-27 17:15 01:16:29 1981539 500 295 15 190 +219.87 ± 24.64
97682 ncm-et-15 2019-09-27 17:12 01:15:26 1975209 500 305 13 182 +232.26 ± 25.19
97681 ncm-et-13 2019-09-27 17:12 01:15:47 1992943 500 293 9 198 +223.94 ± 23.92
97680 ncm-et-3 2019-09-27 17:12 01:17:31 1966773 500 315 9 176 +247.41 ± 25.56
97679 ncm-et-9 2019-09-27 17:09 01:15:56 1967583 500 319 10 171 +250.77 ± 25.99
97678 ncm-et-10 2019-09-27 16:00 01:14:59 1966001 500 297 6 197 +231.21 ± 23.88
97677 ncm-et-4 2019-09-27 15:58 01:16:33 1976248 500 295 9 196 +226.0 ± 24.06
97676 ncm-et-13 2019-09-27 15:54 01:16:59 1969093 500 290 12 198 +217.85 ± 24.01
97675 ncm-et-3 2019-09-27 15:53 01:17:13 1983134 500 290 11 199 +218.86 ± 23.91
97674 ncm-et-15 2019-09-27 15:53 01:17:43 1971098 500 279 11 210 +207.95 ± 23.18
97673 ncm-et-9 2019-09-27 15:52 01:16:10 1967849 500 338 8 154 +275.45 ± 27.45
97672 ncm-et-10 2019-09-27 14:44 01:15:07 1967235 500 278 8 214 +209.91 ± 22.83
97671 ncm-et-4 2019-09-27 14:39 01:17:31 1980019 500 311 6 183 +246.3 ± 24.91
97670 ncm-et-3 2019-09-27 14:37 01:15:23 1967104 500 312 12 176 +240.82 ± 25.63
97669 ncm-et-13 2019-09-27 14:37 01:15:49 1968783 500 317 12 171 +246.3 ± 26.04
97668 ncm-et-15 2019-09-27 14:37 01:15:31 1966614 500 313 3 184 +251.89 ± 24.71
97667 ncm-et-9 2019-09-27 14:36 01:15:30 1969404 500 319 14 167 +246.3 ± 26.4

Commit

Commit ID 3a3ca6af0390d74427c218f29cb5fe1a913efb42
Author 31m059
Date 2019-09-27 10:46:47 UTC
Extend castling independently of singular extension A curious feature of Stockfish's current extension code is its repeated use of "else if." In most cases, this makes no functional difference, because no more than one extension is applied; once one extension has been applied, the remaining ones can be safely ignored. However, if most singular extension search conditions are true, except "value < singularBeta", no non-singular extensions (e.g., castling) can be performed! Three tests were submitted, for three of Stockfish's four non-singular extensions. I excluded the shuffle extension, because historically there have been concerns about the fragility of its conditions, and I did not want to risk causing any serious search problems. - Modifying the passed pawn extension appeared roughly neutral at STC. At best, it appeared to be an improvement of less than 1 Elo. - Modifying check extension performed very poorly at STC - Modifying castling extension (this patch) produced a long "yellow" run at STC (insufficient to pass, but positive score) and a strong LTC. In simple terms, prior to this patch castling extension was occasionally not applied during search--on castling moves. The effect of this patch is to perform castling extension on more castling moves. It does so without adding any code complexity, simply by replacing an "else if" with "if" and reordering some existing code. STC: LLR: -2.96 (-2.94,2.94) [0.00,4.00] Total: 108114 W: 23877 L: 23615 D: 60622 http://tests.stockfishchess.org/tests/view/5d8d86bd0ebc590f3beb0c88 LTC: LLR: 2.96 (-2.94,2.94) [0.00,4.00] Total: 20862 W: 3517 L: 3298 D: 14047 http://tests.stockfishchess.org/tests/view/5d8d99cd0ebc590f3beb1899 Bench: 3728191 -------- Where do we go from here? - It seems strange to me that check extension performed so poorly -- clearly some of the singular extension conditions are also very important for check extension. I am not an expert in search, and I do not have any intuition about which of the eight conditions is/are the culprit. I will try a succession of eight STC tests to identify the relevant conditions, then try to replicate this PR for check extension. - Recent tests interacting with the castle extension may deserve retesting. I will shortly resubmit a few of my recent castling extension tweaks, rebased on this PR/commit. My deepest thanks to @noobpwnftw for the extraordinary CPU donation, and to all our other fishtest volunteers, who made it possible for a speculative LTC to pass in 70 minutes! Closes https://github.com/official-stockfish/Stockfish/pull/2331
Copyright 2011–2024 Next Chess Move LLC