Dev Builds » 20210526-2321

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 08:13:43 1967036 3364 2742 13 609 +392.82 +/- 13.79
ncm-et-4 08:13:55 1966052 3341 2698 21 622 +382.91 +/- 13.66
ncm-et-9 08:13:41 1962400 3328 2717 12 599 +394.42 +/- 13.91
ncm-et-10 08:13:40 1953968 3296 2693 11 592 +395.36 +/- 13.99
ncm-et-13 08:13:46 1961949 3351 2737 25 589 +390.87 +/- 14.05
ncm-et-15 08:13:46 1956986 3320 2697 15 608 +389.39 +/- 13.81
20000 16284 97 3619 +390.91 +/- 5.65

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
147642 ncm-et-10 2021-05-27 21:05 00:44:50 1951459 296 232 1 63 +363.56 +/- 43.35
147641 ncm-et-15 2021-05-27 21:01 00:48:43 1954871 320 257 0 63 +384.73 +/- 43.29
147640 ncm-et-9 2021-05-27 21:01 00:48:45 1966309 328 267 3 58 +386.46 +/- 45.46
147639 ncm-et-4 2021-05-27 20:58 00:51:20 1963999 341 274 3 64 +376.66 +/- 43.19
147638 ncm-et-13 2021-05-27 20:58 00:51:51 1960623 351 282 1 68 +382.25 +/- 41.74
147637 ncm-et-3 2021-05-27 20:56 00:53:44 1971584 364 298 1 65 +397.65 +/- 42.77
147636 ncm-et-10 2021-05-27 19:49 01:14:50 1953437 500 412 0 88 +406.2 +/- 36.53
147635 ncm-et-9 2021-05-27 19:46 01:13:51 1973024 500 394 1 105 +368.59 +/- 33.35
147634 ncm-et-15 2021-05-27 19:46 01:14:04 1959236 500 413 1 86 +406.2 +/- 37.04
147633 ncm-et-4 2021-05-27 19:44 01:13:35 1963231 500 409 1 90 +397.72 +/- 36.17
147632 ncm-et-13 2021-05-27 19:44 01:13:17 1969247 500 410 6 84 +389.56 +/- 37.56
147631 ncm-et-3 2021-05-27 19:43 01:12:14 1972366 500 411 3 86 +397.72 +/- 37.12
147630 ncm-et-10 2021-05-27 18:34 01:14:44 1949648 500 420 1 79 +421.93 +/- 38.73
147629 ncm-et-9 2021-05-27 18:31 01:14:21 1961393 500 415 4 81 +404.05 +/- 38.29
147628 ncm-et-15 2021-05-27 18:31 01:14:43 1954670 500 404 1 95 +387.57 +/- 35.16
147627 ncm-et-13 2021-05-27 18:30 01:12:52 1962312 500 408 2 90 +393.6 +/- 36.22
147626 ncm-et-4 2021-05-27 18:30 01:13:47 1971442 500 387 3 110 +352.8 +/- 32.64
147625 ncm-et-3 2021-05-27 18:29 01:13:37 1970790 500 408 4 88 +389.56 +/- 36.69
147624 ncm-et-10 2021-05-27 17:18 01:15:11 1958017 500 403 2 95 +383.64 +/- 35.21
147623 ncm-et-9 2021-05-27 17:17 01:13:58 1946160 500 411 0 89 +404.05 +/- 36.31
147622 ncm-et-13 2021-05-27 17:16 01:13:12 1959557 500 398 4 98 +370.41 +/- 34.7
147621 ncm-et-15 2021-05-27 17:15 01:14:39 1958954 500 400 3 97 +375.98 +/- 34.86
147620 ncm-et-4 2021-05-27 17:14 01:14:28 1969246 500 402 4 94 +377.87 +/- 35.46
147619 ncm-et-3 2021-05-27 17:14 01:13:59 1959261 500 411 1 88 +401.92 +/- 36.6
147618 ncm-et-10 2021-05-27 16:03 01:14:39 1956026 500 406 4 90 +385.59 +/- 36.27
147617 ncm-et-9 2021-05-27 16:02 01:14:00 1963230 500 412 1 87 +404.05 +/- 36.82
147616 ncm-et-4 2021-05-27 16:02 01:12:04 1974360 500 405 4 91 +383.64 +/- 36.06
147615 ncm-et-13 2021-05-27 16:02 01:13:57 1956950 500 411 3 86 +397.72 +/- 37.12
147614 ncm-et-15 2021-05-27 16:01 01:14:02 1959562 500 397 3 100 +370.41 +/- 34.31
147613 ncm-et-3 2021-05-27 16:00 01:13:41 1962273 500 405 2 93 +387.57 +/- 35.6
147612 ncm-et-9 2021-05-27 14:48 01:13:05 1971117 500 408 2 90 +393.6 +/- 36.22
147611 ncm-et-13 2021-05-27 14:47 01:13:55 1968015 500 420 1 79 +421.93 +/- 38.73
147610 ncm-et-10 2021-05-27 14:47 01:14:59 1950717 500 413 1 86 +406.2 +/- 37.04
147609 ncm-et-4 2021-05-27 14:46 01:14:41 1957413 500 414 3 83 +404.05 +/- 37.81
147608 ncm-et-3 2021-05-27 14:46 01:12:48 1971903 500 406 1 93 +391.57 +/- 35.55
147607 ncm-et-15 2021-05-27 14:46 01:13:55 1953444 500 403 5 92 +377.87 +/- 35.86
147606 ncm-et-3 2021-05-27 13:32 01:13:40 1961075 500 403 1 96 +385.59 +/- 34.96
147605 ncm-et-10 2021-05-27 13:32 01:14:27 1958474 500 407 2 91 +391.57 +/- 36.01
147604 ncm-et-9 2021-05-27 13:32 01:15:41 1955567 500 410 1 89 +399.81 +/- 36.38
147603 ncm-et-15 2021-05-27 13:32 01:13:40 1958165 500 423 2 75 +426.65 +/- 39.83
147602 ncm-et-13 2021-05-27 13:32 01:14:42 1956940 500 408 8 84 +381.7 +/- 37.49
147601 ncm-et-4 2021-05-27 13:32 01:14:00 1962678 500 407 3 90 +389.56 +/- 36.25

Commit

Commit ID f193778446acc6e60d7f0f99c6eb01489f89e962
Author Stéphane Nicolet
Date 2021-05-26 23:21:56 UTC
Do not use lazy evaluation inside NNUE This simplification patch implements two changes: 1. it simplifies away the so-called "lazy" path in the NNUE evaluation internals, where we trusted the psqt head alone to avoid the costly "positional" head in some cases; 2. it raises a little bit the NNUEThreshold1 in evaluate.cpp (from 682 to 800), which increases the limit where we switched from NNUE eval to Classical eval. Both effects increase the number of positional evaluations done by our new net architecture, but the results of our tests below seem to indicate that the loss of speed will be compensated by the gain of eval quality. STC: LLR: 2.95 (-2.94,2.94) <-2.50,0.50> Total: 26280 W: 2244 L: 2137 D: 21899 Ptnml(0-2): 72, 1755, 9405, 1810, 98 https://tests.stockfishchess.org/tests/view/60ae73f112066fd299795a51 LTC: LLR: 2.95 (-2.94,2.94) <-2.50,0.50> Total: 20592 W: 750 L: 677 D: 19165 Ptnml(0-2): 9, 614, 8980, 681, 12 https://tests.stockfishchess.org/tests/view/60ae88e812066fd299795a82 closes https://github.com/official-stockfish/Stockfish/pull/3503 Bench: 3817907
Copyright 2011–2024 Next Chess Move LLC