Dev Builds » 20200908-2053

Use this dev build

NCM plays each Stockfish dev build 20,000 times against Stockfish 15. This yields an approximate Elo difference and establishes confidence in the strength of the dev builds.

Summary

Host Duration Avg Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo
ncm-dbt-01 06:28:40 581040 4000 155 1951 1894 -167.97 ± 4.38 111 1582 299 8 0 -427.25 ± 19.75
ncm-dbt-02 06:27:56 585607 4000 172 1963 1865 -167.42 ± 4.55 121 1563 302 14 0 -418.46 ± 19.66
ncm-dbt-03 06:27:38 586497 4000 174 1971 1855 -168.08 ± 4.45 117 1572 302 9 0 -424.28 ± 19.65
ncm-dbt-04 06:27:19 568037 4000 182 1984 1834 -168.62 ± 4.48 127 1552 317 4 0 -421.35 ± 19.15
ncm-dbt-05 06:28:27 581930 4000 161 1957 1882 -167.97 ± 4.63 131 1547 309 13 0 -415.61 ± 19.43
20000 844 9826 9330 -168.01 ± 2.01 607 7816 1529 48 0 -421.35 ± 8.71

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
440022 ncm-dbt-05 586097 500 17 255 228 -179.9 ± 13.12 20 200 28 2 0 -466.04 ± 66.79
440021 ncm-dbt-01 584790 500 18 239 243 -164.93 ± 11.69 10 202 37 1 0 -429.05 ± 57.55
440020 ncm-dbt-04 568912 500 16 248 236 -174.55 ± 12.12 15 203 31 1 0 -460.32 ± 63.35
440019 ncm-dbt-02 586816 500 14 255 231 -182.6 ± 11.6 17 207 26 0 0 -504.31 ± 69.71
440018 ncm-dbt-03 584537 500 22 254 224 -174.55 ± 13.97 20 196 30 4 0 -433.94 ± 63.67
440017 ncm-dbt-05 579248 500 22 245 233 -166.66 ± 12.97 15 195 38 2 0 -415.05 ± 56.74
440016 ncm-dbt-02 587749 500 26 238 236 -157.23 ± 14.12 15 187 43 5 0 -370.41 ± 52.84
440015 ncm-dbt-01 581194 500 12 242 246 -172.78 ± 10.99 11 208 31 0 0 -471.91 ± 63.24
440014 ncm-dbt-03 586435 500 24 249 227 -168.4 ± 12.14 14 197 39 0 0 -429.05 ± 55.79
440013 ncm-dbt-04 566138 500 20 259 221 -180.8 ± 13.28 23 193 34 0 0 -454.75 ± 60.11
440012 ncm-dbt-01 580282 500 24 244 232 -164.07 ± 11.76 10 201 38 1 0 -424.28 ± 56.73
440011 ncm-dbt-05 584495 500 18 229 253 -156.39 ± 13.77 14 187 45 4 0 -370.41 ± 51.77
440010 ncm-dbt-02 584369 500 21 242 237 -164.93 ± 12.41 12 199 37 2 0 -419.61 ± 57.55
440009 ncm-dbt-03 585042 500 26 252 222 -169.27 ± 13.24 18 191 40 1 0 -415.05 ± 55.18
440008 ncm-dbt-04 565312 500 15 254 231 -180.8 ± 12.8 21 197 32 0 0 -466.03 ± 62.14
440007 ncm-dbt-05 581194 500 26 248 226 -165.79 ± 13.46 18 187 44 1 0 -397.72 ± 52.42
440006 ncm-dbt-01 582861 500 26 248 226 -165.79 ± 12.8 14 196 38 2 0 -415.05 ± 56.74
440005 ncm-dbt-02 587028 500 14 244 242 -172.78 ± 13.66 20 192 36 2 0 -424.28 ± 58.4
440004 ncm-dbt-03 588387 500 23 240 237 -161.49 ± 12.41 12 194 43 1 0 -401.92 ± 53.07
440003 ncm-dbt-04 567165 500 30 243 227 -158.08 ± 13.27 15 184 50 1 0 -374.11 ± 48.94
440002 ncm-dbt-05 579620 500 16 242 242 -169.27 ± 12.32 14 199 36 1 0 -433.94 ± 58.41
440001 ncm-dbt-03 586435 500 24 243 233 -163.21 ± 12.97 15 190 44 1 0 -397.72 ± 52.42
440000 ncm-dbt-01 580282 500 21 247 232 -169.27 ± 13.67 20 187 42 1 0 -406.2 ± 53.75
439999 ncm-dbt-02 586731 500 23 243 234 -164.07 ± 13.35 16 190 42 2 0 -397.72 ± 53.79
439998 ncm-dbt-04 570789 500 20 249 231 -171.9 ± 12.1 15 199 36 0 0 -444.08 ± 58.27
439997 ncm-dbt-04 570068 500 33 247 220 -158.93 ± 12.58 13 188 49 0 0 -385.59 ± 49.31
439996 ncm-dbt-05 580365 500 19 248 233 -171.9 ± 12.35 16 197 37 0 0 -438.94 ± 57.41
439995 ncm-dbt-01 572437 500 16 240 244 -167.53 ± 13.13 16 194 38 2 0 -415.05 ± 56.74
439994 ncm-dbt-02 586816 500 23 247 230 -167.53 ± 12.21 14 196 40 0 0 -424.27 ± 55.03
439993 ncm-dbt-03 584916 500 13 247 240 -176.32 ± 10.89 12 210 28 0 0 -490.7 ± 66.89
439992 ncm-dbt-05 582485 500 26 247 227 -164.93 ± 14.34 21 181 46 2 0 -381.7 ± 51.25
439991 ncm-dbt-04 567126 500 20 242 238 -165.79 ± 12.34 13 197 39 1 0 -419.61 ± 55.94
439990 ncm-dbt-01 583866 500 18 251 231 -175.43 ± 12.54 17 200 32 1 0 -454.76 ± 62.27
439989 ncm-dbt-02 583405 500 23 244 233 -164.93 ± 13.08 15 193 40 2 0 -406.2 ± 55.21
439988 ncm-dbt-03 586985 500 27 243 230 -160.64 ± 13.13 14 190 44 2 0 -389.56 ± 52.48
439987 ncm-dbt-04 568792 500 28 242 230 -158.93 ± 12.58 12 191 46 1 0 -389.56 ± 51.18
439986 ncm-dbt-05 581943 500 17 243 240 -169.27 ± 12.07 13 201 35 1 0 -438.95 ± 59.31
439985 ncm-dbt-03 589240 500 15 243 242 -171.02 ± 11.42 12 204 34 0 0 -454.75 ± 60.11
439984 ncm-dbt-02 581943 500 28 250 222 -165.79 ± 12.11 12 199 38 1 0 -424.28 ± 56.73
439983 ncm-dbt-01 582611 500 20 240 240 -164.07 ± 12.23 13 194 43 0 0 -410.57 ± 52.91

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 2011–2025 Next Chess Move LLC