Dev Builds » 20211121-2018

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 08:30:44 1958301 3353 2873 7 473 +442.48 +/- 15.67
ncm-et-4 08:31:14 1919060 3244 2839 9 396 +466.59 +/- 17.15
ncm-et-9 08:31:07 1963723 3377 2891 10 476 +440.38 +/- 15.63
ncm-et-10 08:31:16 1955638 3331 2876 3 452 +452.72 +/- 16.02
ncm-et-13 08:30:55 1960735 3367 2880 8 479 +440.2 +/- 15.57
ncm-et-15 08:31:15 1956087 3328 2889 4 435 +458.76 +/- 16.34
20000 17248 41 2711 +449.82 +/- 6.54

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
153765 ncm-et-4 2021-11-22 14:14 00:37:54 1960947 244 203 0 41 +415.0 +/- 54.27
153764 ncm-et-15 2021-11-22 14:02 00:50:15 1950711 328 280 0 48 +441.06 +/- 50.07
153763 ncm-et-10 2021-11-22 14:01 00:51:47 1953179 331 276 0 55 +417.12 +/- 46.57
153762 ncm-et-3 2021-11-22 13:59 00:52:54 1969086 353 307 0 46 +462.71 +/- 51.27
153761 ncm-et-9 2021-11-22 13:56 00:56:05 1960165 377 308 1 68 +395.98 +/- 41.78
153760 ncm-et-13 2021-11-22 13:56 00:56:29 1955568 367 313 1 53 +436.6 +/- 47.66
153759 ncm-et-4 2021-11-22 12:54 01:19:54 1889998 500 430 2 68 +444.09 +/- 41.92
153758 ncm-et-15 2021-11-22 12:44 01:17:46 1955263 500 431 1 68 +449.35 +/- 41.89
153757 ncm-et-3 2021-11-22 12:42 01:15:58 1973898 500 434 1 65 +457.52 +/- 42.89
153756 ncm-et-10 2021-11-22 12:42 01:17:55 1957716 500 439 0 61 +474.93 +/- 44.28
153755 ncm-et-13 2021-11-22 12:40 01:15:20 1966003 500 419 0 81 +421.93 +/- 38.15
153754 ncm-et-9 2021-11-22 12:38 01:17:01 1966785 500 433 1 66 +454.76 +/- 42.55
153753 ncm-et-4 2021-11-22 11:38 01:15:30 1968777 500 447 1 52 +497.4 +/- 48.23
153752 ncm-et-15 2021-11-22 11:26 01:16:40 1953151 500 438 0 62 +471.92 +/- 43.9
153751 ncm-et-10 2021-11-22 11:24 01:17:09 1948877 500 436 1 63 +463.16 +/- 43.6
153750 ncm-et-3 2021-11-22 11:24 01:18:05 1946710 500 426 2 72 +433.94 +/- 40.69
153749 ncm-et-13 2021-11-22 11:23 01:16:02 1960782 500 434 3 63 +452.04 +/- 43.62
153748 ncm-et-9 2021-11-22 11:22 01:15:41 1962925 500 431 2 67 +446.7 +/- 42.25
153747 ncm-et-4 2021-11-22 10:19 01:18:28 1943433 500 428 1 71 +441.5 +/- 40.95
153746 ncm-et-15 2021-11-22 10:09 01:16:50 1955417 500 430 1 69 +446.7 +/- 41.57
153745 ncm-et-10 2021-11-22 10:08 01:16:05 1963378 500 444 0 56 +490.71 +/- 46.32
153744 ncm-et-3 2021-11-22 10:06 01:16:49 1948428 500 427 1 72 +438.95 +/- 40.66
153743 ncm-et-13 2021-11-22 10:06 01:16:33 1956947 500 443 2 55 +481.1 +/- 46.83
153742 ncm-et-9 2021-11-22 10:06 01:15:42 1963073 500 426 2 72 +433.94 +/- 40.69
153741 ncm-et-4 2021-11-22 08:57 01:21:09 1870019 500 447 1 52 +497.4 +/- 48.23
153740 ncm-et-15 2021-11-22 08:52 01:16:29 1960641 500 432 0 68 +454.76 +/- 41.82
153739 ncm-et-10 2021-11-22 08:51 01:16:22 1947219 500 426 1 73 +436.43 +/- 40.36
153738 ncm-et-13 2021-11-22 08:50 01:14:57 1961234 500 427 2 71 +436.43 +/- 40.99
153737 ncm-et-9 2021-11-22 08:49 01:15:42 1961254 500 425 2 73 +431.48 +/- 40.4
153736 ncm-et-3 2021-11-22 08:49 01:16:28 1952583 500 426 2 72 +433.94 +/- 40.69
153735 ncm-et-15 2021-11-22 07:35 01:16:09 1962926 500 437 1 62 +466.04 +/- 43.97
153734 ncm-et-4 2021-11-22 07:34 01:22:04 1830544 500 445 3 52 +484.25 +/- 48.17
153733 ncm-et-13 2021-11-22 07:34 01:15:40 1960614 500 415 0 85 +412.8 +/- 37.2
153732 ncm-et-3 2021-11-22 07:34 01:14:50 1960614 500 428 1 71 +441.5 +/- 40.95
153731 ncm-et-10 2021-11-22 07:33 01:16:46 1958932 500 434 0 66 +460.32 +/- 42.48
153730 ncm-et-9 2021-11-22 07:33 01:15:37 1968322 500 430 1 69 +446.7 +/- 41.57
153729 ncm-et-3 2021-11-22 06:17 01:15:40 1956791 500 425 0 75 +436.43 +/- 39.73
153728 ncm-et-10 2021-11-22 06:17 01:15:12 1960166 500 421 1 78 +424.28 +/- 38.99
153727 ncm-et-13 2021-11-22 06:17 01:15:54 1963998 500 429 0 71 +446.7 +/- 40.89
153726 ncm-et-4 2021-11-22 06:17 01:16:15 1969708 500 439 1 60 +471.92 +/- 44.73
153725 ncm-et-9 2021-11-22 06:17 01:15:19 1963537 500 438 1 61 +468.96 +/- 44.35
153724 ncm-et-15 2021-11-22 06:17 01:17:06 1954502 500 441 1 58 +477.99 +/- 45.54

Commit

Commit ID a5a89b27c8e3225fb453d603bc4515d32bb351c3
Author Stéphane Nicolet
Date 2021-11-21 20:18:08 UTC
Introduce Optimism Current master implements a scaling of the raw NNUE output value with a formula equivalent to 'eval = alpha * NNUE_output', where the scale factor alpha varies between 1.8 (for early middle game) and 0.9 (for pure endgames). This feature allows Stockfish to keep material on the board when she thinks she has the advantage, and to seek exchanges and simplifications when she thinks she has to defend. This patch slightly offsets the turning point between these two strategies, by adding to Stockfish's evaluation a small "optimism" value before actually doing the scaling. The effect is that SF will play a little bit more risky, trying to keep the tension a little bit longer when she is defending, and keeping even more material on the board when she has an advantage. We note that this patch is similar in spirit to the old "Contempt" idea we used to have in classical Stockfish, but this implementation differs in two key points: a) it has been tested as an Elo-gainer against master; b) the values output by the search are not changed on average by the implementation (in other words, the optimism value changes the tension/exchange strategy, but a displayed value of 1.0 pawn has the same signification before and after the patch). See the old comment https://github.com/official-stockfish/Stockfish/pull/1361#issuecomment-359165141 for some images illustrating the ideas. ------- finished yellow at STC: LLR: -2.94 (-2.94,2.94) <0.00,2.50> Total: 165048 W: 41705 L: 41611 D: 81732 Ptnml(0-2): 565, 18959, 43245, 19327, 428 https://tests.stockfishchess.org/tests/view/61942a3dcd645dc8291c876b passed LTC: LLR: 2.95 (-2.94,2.94) <0.50,3.00> Total: 121656 W: 30762 L: 30287 D: 60607 Ptnml(0-2): 87, 12558, 35032, 13095, 56 https://tests.stockfishchess.org/tests/view/61962c58cd645dc8291c8877 ------- How to continue from there? a) the shape (slope and amplitude) of the sigmoid used to compute the optimism value could be tweaked to try to gain more Elo, so the parameters of the sigmoid function in line 391 of search.cpp could be tuned with SPSA. Manual tweaking is also possible using this Desmos page: https://www.desmos.com/calculator/jhh83sqq92 b) in a similar vein, with two recents patches affecting the scaling of the NNUE evaluation in evaluate.cpp, now could be a good time to try a round of SPSA tuning of the NNUE network; c) this patch will tend to keep tension in middlegame a little bit longer, so any patch improving the defensive aspect of play via search extensions in risky, tactical positions would be welcome. ------- closes https://github.com/official-stockfish/Stockfish/pull/3797 Bench: 6184852
Copyright 2011–2024 Next Chess Move LLC