Dev Builds » 20200908-2053

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 08:42:04 1963732 3376 2527 26 823 +330.86 +/- 11.84
ncm-et-4 08:41:46 1967147 3308 2483 36 789 +330.02 +/- 12.12
ncm-et-9 08:41:50 1968575 3337 2539 30 768 +339.53 +/- 12.28
ncm-et-10 08:42:10 1959388 3309 2534 30 745 +343.44 +/- 12.47
ncm-et-13 08:41:59 1961098 3334 2540 33 761 +339.59 +/- 12.34
ncm-et-15 08:41:49 1966895 3337 2464 41 832 +319.79 +/- 11.8
20001 15087 196 4718 +333.72 +/- 4.95

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
137863 ncm-et-4 2020-09-10 03:17 00:48:20 1974826 308 254 2 52 +400.0 +/- 48.11 ↓
137862 ncm-et-10 2020-09-10 03:16 00:49:48 1958184 309 244 4 61 +360.29 +/- 44.26 ↓
137861 ncm-et-13 2020-09-10 03:13 00:52:53 1962000 334 253 3 78 +336.85 +/- 38.91 ↓
137860 ncm-et-15 2020-09-10 03:13 00:52:50 1975137 337 242 5 90 +303.56 +/- 36.15 ↓
137859 ncm-et-9 2020-09-10 03:12 00:53:27 1962459 337 247 4 86 +316.12 +/- 37.0 ↓
137858 ncm-et-3 2020-09-10 03:07 00:58:20 1965380 376 284 4 88 +333.85 +/- 36.61 ↓
137857 ncm-et-4 2020-09-10 01:57 01:18:45 1966162 500 376 2 122 +336.46 +/- 30.86 ↓
137856 ncm-et-10 2020-09-10 01:56 01:19:20 1966173 500 380 4 116 +339.63 +/- 31.78 ↓
137855 ncm-et-13 2020-09-10 01:54 01:17:42 1961534 500 383 3 114 +346.12 +/- 32.03 ↓
137854 ncm-et-15 2020-09-10 01:54 01:17:52 1961690 500 371 6 123 +322.67 +/- 30.87 ↓
137853 ncm-et-9 2020-09-10 01:52 01:19:10 1966942 500 369 8 123 +316.8 +/- 30.9 ↓
137852 ncm-et-3 2020-09-10 01:49 01:17:06 1963227 500 363 3 134 +315.35 +/- 29.4 ↓
137851 ncm-et-4 2020-09-10 00:38 01:18:07 1970188 500 374 8 118 +324.17 +/- 31.56 ↓
137850 ncm-et-10 2020-09-10 00:35 01:19:39 1957274 500 364 2 134 +318.25 +/- 29.36 ↓
137849 ncm-et-15 2020-09-10 00:34 01:18:25 1973750 500 369 5 126 +321.19 +/- 30.45 ↓
137848 ncm-et-13 2020-09-10 00:34 01:18:56 1959695 500 394 5 101 +361.43 +/- 34.18 ↓
137847 ncm-et-9 2020-09-10 00:32 01:18:57 1961845 500 387 3 110 +352.8 +/- 32.64 ↓
137846 ncm-et-3 2020-09-10 00:31 01:17:15 1965842 500 375 6 119 +328.7 +/- 31.4 ↓
137845 ncm-et-4 2020-09-09 23:18 01:18:41 1974686 500 371 6 123 +322.67 +/- 30.87 ↓
137844 ncm-et-15 2020-09-09 23:16 01:17:15 1966626 500 376 7 117 +328.7 +/- 31.7 ↓
137843 ncm-et-10 2020-09-09 23:15 01:18:32 1961863 500 388 8 104 +346.12 +/- 33.67 ↓
137842 ncm-et-13 2020-09-09 23:13 01:19:34 1958622 500 401 12 87 +361.43 +/- 36.62 ↓
137841 ncm-et-9 2020-09-09 23:13 01:17:59 1971569 500 392 2 106 +363.2 +/- 33.24 ↓
137840 ncm-et-3 2020-09-09 23:12 01:17:51 1963387 500 370 1 129 +328.7 +/- 29.9 ↓
137839 ncm-et-4 2020-09-09 21:57 01:20:26 1955720 500 362 5 133 +311.06 +/- 29.6 ↓
137838 ncm-et-15 2020-09-09 21:57 01:18:23 1961074 500 368 5 127 +319.72 +/- 30.33 ↓
137837 ncm-et-10 2020-09-09 21:56 01:18:14 1956787 500 379 3 118 +339.63 +/- 31.46 ↓
137836 ncm-et-9 2020-09-09 21:55 01:16:22 1968186 500 376 6 118 +330.23 +/- 31.54 ↓
137835 ncm-et-13 2020-09-09 21:55 01:17:30 1961841 500 374 7 119 +325.67 +/- 31.42 ↓
137834 ncm-et-3 2020-09-09 21:54 01:16:57 1963839 500 380 5 115 +338.04 +/- 31.95 ↓
137833 ncm-et-9 2020-09-09 20:37 01:17:02 1974202 500 381 4 115 +341.24 +/- 31.92 ↓
137832 ncm-et-4 2020-09-09 20:37 01:18:43 1964765 500 374 9 117 +322.67 +/- 31.7 ↓
137831 ncm-et-15 2020-09-09 20:36 01:19:17 1965842 500 368 8 124 +315.35 +/- 30.77 ↓
137830 ncm-et-3 2020-09-09 20:36 01:17:20 1961536 500 381 3 116 +342.85 +/- 31.74 ↓
137829 ncm-et-10 2020-09-09 20:35 01:19:25 1953892 500 386 6 108 +346.12 +/- 33.02 ↓
137828 ncm-et-13 2020-09-09 20:35 01:18:27 1960921 500 365 2 133 +319.72 +/- 29.47 ↓
137827 ncm-et-9 2020-09-09 19:17 01:18:53 1974827 500 387 3 110 +352.8 +/- 32.64 ↓
137826 ncm-et-15 2020-09-09 19:17 01:17:47 1964148 500 370 5 125 +322.67 +/- 30.58 ↓
137825 ncm-et-3 2020-09-09 19:17 01:17:15 1962919 500 374 4 122 +330.23 +/- 30.94 ↓
137824 ncm-et-13 2020-09-09 19:17 01:16:57 1963073 500 370 1 129 +328.7 +/- 29.9 ↓
137823 ncm-et-10 2020-09-09 19:17 01:17:12 1961543 500 393 3 104 +363.2 +/- 33.62 ↓
137822 ncm-et-4 2020-09-09 19:17 01:18:44 1963686 500 372 4 124 +327.18 +/- 30.68 ↓

Commit

Commit ID fc27d158c012341593518a05abf51903ecbcb495
Author syzygy1
Date 2020-09-08 20:53:17 UTC
Bug fix in do_null_move() and NNUE simplification. This fixes #3108 and removes some NNUE code that is currently not used. At the moment, do_null_move() copies the accumulator from the previous state into the new state, which is correct. It then clears the "computed_score" flag because the side to move has changed, and with the other side to move NNUE will return a completely different evaluation (normally with changed sign but also with different NNUE-internal tempo bonus). The problem is that do_null_move() clears the wrong flag. It clears the computed_score flag of the old state, not of the new state. It turns out that this almost never affects the search. For example, fixing it does not change the current bench (but it does change the previous bench). This is because the search code usually avoids calling evaluate() after a null move. This PR corrects do_null_move() by removing the computed_score flag altogether. The flag is not needed because nnue_evaluate() is never called twice on a position. This PR also removes some unnecessary {}s and inserts a few blank lines in the modified NNUE files in line with SF coding style. Resulf ot STC non-regression test: LLR: 2.95 (-2.94,2.94) {-1.25,0.25} Total: 26328 W: 3118 L: 3012 D: 20198 Ptnml(0-2): 126, 2208, 8397, 2300, 133 https://tests.stockfishchess.org/tests/view/5f553ccc2d02727c56b36db1 closes https://github.com/official-stockfish/Stockfish/pull/3109 bench: 4109324
Copyright 2011–2024 Next Chess Move LLC
Feedback & Support Terms of Service Privacy Policy