Dev Builds » 20230612-1841

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 09:28:16 1956791 3337 2873 3 461 +449.43 +/- 15.86
ncm-et-4 09:25:56 1954388 3332 2860 3 469 +445.97 +/- 15.72
ncm-et-9 09:24:47 1956599 3335 2830 4 501 +433.17 +/- 15.21
ncm-et-10 09:28:12 1952754 3331 2863 2 466 +447.89 +/- 15.77
ncm-et-13 09:24:05 1957684 3331 2877 2 452 +453.54 +/- 16.02
ncm-et-15 09:26:32 1955016 3334 2864 0 470 +448.06 +/- 15.69
20000 17167 14 2819 +446.24 +/- 6.4

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
192629 ncm-et-13 2023-06-16 08:22 00:56:46 1949639 331 279 1 51 +424.14 +/- 48.59
192628 ncm-et-4 2023-06-16 08:22 00:57:11 1956492 332 284 1 47 +439.47 +/- 50.75
192627 ncm-et-15 2023-06-16 08:21 00:57:14 1955132 334 283 0 51 +433.08 +/- 48.49
192626 ncm-et-10 2023-06-16 08:21 00:57:23 1946645 336 286 0 50 +437.92 +/- 49.0
192625 ncm-et-3 2023-06-16 08:21 00:57:23 1958327 337 286 2 49 +427.53 +/- 49.68
192624 ncm-et-9 2023-06-16 08:21 00:57:15 1955593 335 281 1 53 +419.4 +/- 47.62
192371 ncm-et-10 2023-06-15 02:23 01:24:05 1957729 500 428 1 71 +441.5 +/- 40.95
192370 ncm-et-3 2023-06-15 02:21 01:23:06 1959549 500 440 0 60 +477.98 +/- 44.67
192369 ncm-et-4 2023-06-15 02:18 01:24:07 1951028 500 426 0 74 +438.95 +/- 40.01
192368 ncm-et-15 2023-06-15 02:17 01:25:11 1957838 500 441 0 59 +481.09 +/- 45.07
192367 ncm-et-13 2023-06-15 02:15 01:25:13 1949828 500 438 0 62 +471.92 +/- 43.9
192366 ncm-et-9 2023-06-15 02:15 01:25:47 1954458 500 430 0 70 +449.35 +/- 41.19
192359 ncm-et-10 2023-06-15 00:56 01:25:41 1954681 500 434 0 66 +460.32 +/- 42.48
192358 ncm-et-3 2023-06-15 00:56 01:24:43 1961752 500 437 0 63 +468.95 +/- 43.54
192357 ncm-et-4 2023-06-15 00:53 01:23:45 1962919 500 427 0 73 +441.5 +/- 40.29
192356 ncm-et-15 2023-06-15 00:52 01:24:47 1952697 500 433 0 67 +457.52 +/- 42.15
192355 ncm-et-13 2023-06-15 00:50 01:24:48 1955014 500 425 1 74 +433.94 +/- 40.08
192354 ncm-et-9 2023-06-15 00:49 01:24:51 1952434 500 416 1 83 +412.8 +/- 37.74
192347 ncm-et-4 2023-06-14 23:30 01:22:42 1962182 500 435 0 65 +463.15 +/- 42.83
192346 ncm-et-3 2023-06-14 23:30 01:25:18 1956373 500 423 1 76 +429.05 +/- 39.52
192345 ncm-et-10 2023-06-14 23:29 01:26:26 1944508 500 434 1 65 +457.52 +/- 42.89
192344 ncm-et-15 2023-06-14 23:26 01:24:56 1952521 500 431 0 69 +452.04 +/- 41.5
192343 ncm-et-9 2023-06-14 23:24 01:24:42 1953343 500 420 2 78 +419.61 +/- 39.03
192342 ncm-et-13 2023-06-14 23:24 01:25:06 1967110 500 436 0 64 +466.03 +/- 43.18
192335 ncm-et-4 2023-06-14 22:04 01:25:19 1943905 500 423 0 77 +431.48 +/- 39.18
192334 ncm-et-10 2023-06-14 22:02 01:26:16 1950402 500 435 0 65 +463.15 +/- 42.83
192333 ncm-et-3 2023-06-14 22:02 01:26:38 1952064 500 438 0 62 +471.92 +/- 43.9
192332 ncm-et-13 2023-06-14 22:02 01:21:28 1962034 500 430 0 70 +449.35 +/- 41.19
192331 ncm-et-15 2023-06-14 22:01 01:24:04 1950394 500 417 0 83 +417.31 +/- 37.67
192330 ncm-et-9 2023-06-14 22:01 01:22:20 1959870 500 415 0 85 +412.8 +/- 37.2
192323 ncm-et-4 2023-06-14 20:38 01:25:20 1957863 500 422 1 77 +426.65 +/- 39.25
192322 ncm-et-10 2023-06-14 20:36 01:25:35 1955727 500 426 0 74 +438.95 +/- 40.01
192321 ncm-et-15 2023-06-14 20:36 01:24:38 1962184 500 431 0 69 +452.04 +/- 41.5
192320 ncm-et-3 2023-06-14 20:36 01:25:41 1954815 500 427 0 73 +441.5 +/- 40.29
192319 ncm-et-13 2023-06-14 20:34 01:26:33 1958778 500 435 0 65 +463.15 +/- 42.83
192318 ncm-et-9 2023-06-14 20:34 01:26:02 1954809 500 430 0 70 +449.35 +/- 41.19
192311 ncm-et-3 2023-06-14 19:10 01:25:27 1954660 500 422 0 78 +429.05 +/- 38.92
192310 ncm-et-9 2023-06-14 19:10 01:23:50 1965689 500 438 0 62 +471.92 +/- 43.9
192309 ncm-et-15 2023-06-14 19:10 01:25:42 1954352 500 428 0 72 +444.08 +/- 40.59
192308 ncm-et-10 2023-06-14 19:10 01:22:46 1959590 495 420 0 75 +434.54 +/- 39.72
192307 ncm-et-4 2023-06-14 19:10 01:27:32 1946328 500 443 1 56 +484.25 +/- 46.39
192306 ncm-et-13 2023-06-14 19:10 01:24:11 1961387 500 434 0 66 +460.32 +/- 42.48

Commit

Commit ID 38e61663d836e062af0bc002814ad5149c4b7729
Author AndrovT
Date 2023-06-12 18:41:27 UTC
Use block sparse input for the first layer. Use block sparse input for the first fully connected layer on architectures with at least SSSE3. Depending on the CPU architecture, this yields a speedup of up to 10%, e.g. ``` Result of 100 runs of 'bench 16 1 13 default depth NNUE' base (...ockfish-base) = 959345 +/- 7477 test (...ckfish-patch) = 1054340 +/- 9640 diff = +94995 +/- 3999 speedup = +0.0990 P(speedup > 0) = 1.0000 CPU: 8 x AMD Ryzen 7 5700U with Radeon Graphics Hyperthreading: on ``` Passed STC: https://tests.stockfishchess.org/tests/view/6485aa0965ffe077ca12409c LLR: 2.93 (-2.94,2.94) <0.00,2.00> Total: 8864 W: 2479 L: 2223 D: 4162 Ptnml(0-2): 13, 829, 2504, 1061, 25 This commit includes a net with reordered weights, to increase the likelihood of block sparse inputs, but otherwise equivalent to the previous master net (nn-ea57bea57e32.nnue). Activation data collected with https://github.com/AndrovT/Stockfish/tree/log-activations, running bench 16 1 13 varied_1000.epd depth NNUE on this data. Net parameters permuted with https://gist.github.com/AndrovT/9e3fbaebb7082734dc84d27e02094cb3. closes https://github.com/official-stockfish/Stockfish/pull/4612 No functional change
Copyright 2011–2024 Next Chess Move LLC