Dev Builds » 20210319-1058

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 08:23:36 1964473 3332 2669 21 642 +376.66 +/- 13.44
ncm-et-4 08:23:21 1962865 3361 2715 18 628 +384.06 +/- 13.59
ncm-et-9 08:24:00 1956112 3332 2698 12 622 +387.69 +/- 13.64
ncm-et-10 08:23:41 1955320 3308 2659 13 636 +381.58 +/- 13.49
ncm-et-13 08:23:20 1962542 3347 2644 14 689 +368.39 +/- 12.95
ncm-et-15 08:23:24 1952169 3320 2681 26 613 +381.41 +/- 13.77
20000 16066 104 3830 +379.87 +/- 5.49

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
145505 ncm-et-3 2021-03-20 04:42 00:50:44 1964007 337 270 1 66 +379.99 +/- 42.38
145504 ncm-et-15 2021-03-20 04:42 00:50:39 1956336 335 270 2 63 +381.7 +/- 43.51
145503 ncm-et-10 2021-03-20 04:42 00:50:43 1955265 334 261 0 73 +364.47 +/- 40.05
145502 ncm-et-4 2021-03-20 04:42 00:50:54 1958319 338 272 2 64 +380.56 +/- 43.15
145501 ncm-et-13 2021-03-20 04:42 00:50:47 1963993 339 277 2 60 +392.8 +/- 44.65
145500 ncm-et-9 2021-03-20 04:42 00:50:58 1960780 340 264 1 75 +357.53 +/- 39.6
145457 ncm-et-13 2021-03-19 20:14 00:02:09 1966160 8 5 0 3 +254.61 +/- 287.37
145456 ncm-et-4 2021-03-19 20:11 00:04:14 1956837 23 15 2 6 +222.52 +/- 152.52
145455 ncm-et-10 2021-03-19 19:04 01:11:54 1958317 474 383 3 88 +383.33 +/- 36.66
145454 ncm-et-15 2021-03-19 19:02 01:13:37 1957560 485 389 4 92 +375.81 +/- 35.85
145453 ncm-et-9 2021-03-19 19:02 01:13:58 1949792 492 401 1 90 +394.63 +/- 36.16
145452 ncm-et-3 2021-03-19 19:02 01:13:58 1970815 495 399 2 94 +383.66 +/- 35.4
145451 ncm-et-4 2021-03-19 18:57 01:13:45 1973909 500 396 3 101 +368.59 +/- 34.13
145450 ncm-et-13 2021-03-19 18:57 01:16:20 1957705 500 393 3 104 +363.2 +/- 33.62
145449 ncm-et-10 2021-03-19 17:47 01:16:06 1950698 500 397 0 103 +375.98 +/- 33.63
145448 ncm-et-9 2021-03-19 17:47 01:14:09 1961237 500 404 1 95 +387.57 +/- 35.16
145447 ncm-et-3 2021-03-19 17:46 01:14:36 1976862 500 396 1 103 +372.25 +/- 33.69
145446 ncm-et-15 2021-03-19 17:46 01:14:59 1948739 500 407 3 90 +389.56 +/- 36.25
145445 ncm-et-13 2021-03-19 17:43 01:13:31 1965391 500 378 1 121 +341.24 +/- 30.94
145444 ncm-et-4 2021-03-19 17:42 01:14:54 1969710 500 404 3 93 +383.64 +/- 35.64
145443 ncm-et-3 2021-03-19 16:31 01:15:14 1966613 500 404 2 94 +385.59 +/- 35.4
145442 ncm-et-9 2021-03-19 16:30 01:16:24 1948256 500 396 1 103 +372.25 +/- 33.69
145441 ncm-et-15 2021-03-19 16:30 01:15:59 1956651 500 396 8 96 +359.68 +/- 35.06
145440 ncm-et-10 2021-03-19 16:29 01:17:03 1954810 500 402 4 94 +377.87 +/- 35.46
145439 ncm-et-13 2021-03-19 16:28 01:14:28 1962307 500 399 2 99 +375.98 +/- 34.46
145438 ncm-et-4 2021-03-19 16:25 01:15:52 1946109 500 410 3 87 +395.65 +/- 36.9
145437 ncm-et-9 2021-03-19 15:14 01:15:31 1958833 500 420 2 78 +419.61 +/- 39.03
145436 ncm-et-10 2021-03-19 15:13 01:15:48 1959082 500 420 4 76 +415.05 +/- 39.57
145435 ncm-et-3 2021-03-19 15:13 01:17:19 1953433 500 400 5 95 +372.25 +/- 35.28
145434 ncm-et-15 2021-03-19 15:12 01:16:43 1940233 500 399 6 95 +368.59 +/- 35.27
145433 ncm-et-13 2021-03-19 15:12 01:14:41 1962462 500 392 3 105 +361.43 +/- 33.45
145432 ncm-et-4 2021-03-19 15:10 01:14:33 1958473 500 414 1 85 +408.38 +/- 37.27
145431 ncm-et-15 2021-03-19 13:56 01:15:24 1951009 500 418 1 81 +417.32 +/- 38.22
145430 ncm-et-3 2021-03-19 13:56 01:16:15 1949864 500 390 6 104 +352.8 +/- 33.67
145429 ncm-et-9 2021-03-19 13:56 01:17:33 1958628 500 417 2 81 +412.8 +/- 38.27
145428 ncm-et-4 2021-03-19 13:55 01:14:05 1968932 500 403 3 94 +381.7 +/- 35.44
145427 ncm-et-10 2021-03-19 13:55 01:17:07 1949349 500 404 1 95 +387.57 +/- 35.16
145426 ncm-et-13 2021-03-19 13:54 01:17:27 1958629 500 404 2 94 +385.59 +/- 35.4
145425 ncm-et-15 2021-03-19 12:40 01:16:03 1954661 500 402 2 96 +381.7 +/- 35.02
145424 ncm-et-3 2021-03-19 12:40 01:15:30 1969717 500 410 4 86 +393.6 +/- 37.13
145423 ncm-et-9 2021-03-19 12:40 01:15:27 1955264 500 396 4 100 +366.78 +/- 34.34
145422 ncm-et-10 2021-03-19 12:40 01:15:00 1959719 500 392 1 107 +364.98 +/- 33.02
145421 ncm-et-13 2021-03-19 12:40 01:13:57 1963696 500 396 1 103 +372.25 +/- 33.69
145420 ncm-et-4 2021-03-19 12:40 01:15:04 1970633 500 401 1 98 +381.7 +/- 34.59

Commit

Commit ID ace9632c6713e346ccca73b3e2edc046c1f5527c
Author Vizvezdenec
Date 2021-03-19 10:58:17 UTC
Add a specific FRC correction from classical to NNUE our net currently is not trained on FRC games, and so doesn't know about the important pattern of a bishop that is cornered in FRC. This patch introduces a term we have in the classical evaluation for this case, and adds it to the NNUE eval. Since fishtest doesn't support FRC right now, the patch was tested locally at STC conditions, starting from the book of FRC starting positions. Score of master vs patch: 993 - 2226 - 6781 [0.438] 10000 Which corresponds to approximately 40 Elo The patch passes non-regression testing for traditional chess (where it adds one branch). passed STC: https://tests.stockfishchess.org/tests/view/604fa2532433018de7a38b67 LLR: 2.95 (-2.94,2.94) {-1.25,0.25} Total: 30560 W: 2701 L: 2636 D: 25223 Ptnml(0-2): 88, 2056, 10921, 2133, 82 passed STC also in an earlier version: https://tests.stockfishchess.org/tests/view/604f61282433018de7a38b4d closes https://github.com/official-stockfish/Stockfish/pull/3398 No functional change
Copyright 2011–2024 Next Chess Move LLC