Dev Builds » 20240424-1638

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 06:51:26 584048 4000 1475 593 1932 +77.89 ± 5.02 3 122 881 978 16 +161.71 ± 11.44
ncm-dbt-02 06:51:19 587022 4002 1502 578 1922 +81.69 ± 5.0 3 116 849 1020 13 +171.36 ± 11.66
ncm-dbt-03 06:49:32 587045 4000 1456 603 1941 +75.25 ± 5.01 1 138 877 975 9 +156.6 ± 11.48
ncm-dbt-04 06:49:24 571852 4000 1455 591 1954 +76.25 ± 4.95 2 119 906 959 14 +158.08 ± 11.25
ncm-dbt-05 06:51:41 583939 3998 1483 572 1943 +80.58 ± 4.97 1 114 875 991 18 +167.19 ± 11.47
20000 7371 2937 9692 +78.33 ± 2.23 10 609 4388 4923 70 +162.95 ± 5.12

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
382573 ncm-dbt-02 588260 2 1 0 1 +189.7 ± 55.98 0 0 0 1 0 +1129.65 ± 376.02
382572 ncm-dbt-05 583950 498 184 58 256 +89.85 ± 14.04 0 13 99 135 2 +189.92 ± 34.36
382571 ncm-dbt-01 586054 500 175 80 245 +66.82 ± 15.16 0 24 111 111 4 +132.54 ± 32.44
382570 ncm-dbt-04 572961 500 181 67 252 +80.63 ± 13.33 0 9 121 117 3 +165.8 ± 30.41
382569 ncm-dbt-02 584579 500 183 74 243 +76.98 ± 14.41 2 13 110 124 1 +164.07 ± 32.48
382568 ncm-dbt-03 589283 500 190 69 241 +85.78 ± 14.22 0 15 101 132 2 +179.9 ± 34.03
382567 ncm-dbt-04 568792 500 178 85 237 +65.38 ± 14.83 1 21 114 112 2 +134.15 ± 31.97
382566 ncm-dbt-05 584369 500 180 66 254 +80.63 ± 13.94 0 14 110 124 2 +167.53 ± 32.45
382565 ncm-dbt-01 584034 500 190 75 235 +81.36 ± 13.66 0 12 113 123 2 +169.27 ± 31.88
382564 ncm-dbt-02 584159 500 200 70 230 +92.45 ± 14.2 0 15 91 143 1 +198.34 ± 35.92
382563 ncm-dbt-03 588004 500 171 84 245 +61.07 ± 14.52 0 25 113 112 0 +126.18 ± 32.14
382562 ncm-dbt-04 571592 500 187 76 237 +78.44 ± 15.02 1 19 100 128 2 +164.07 ± 34.22
382561 ncm-dbt-01 583908 500 183 68 249 +81.36 ± 13.96 0 14 109 125 2 +169.27 ± 32.62
382560 ncm-dbt-05 584790 500 192 71 237 +85.78 ± 13.92 0 12 108 127 3 +178.11 ± 32.72
382559 ncm-dbt-02 587240 500 183 78 239 +74.06 ± 14.32 0 20 105 125 0 +155.54 ± 33.38
382558 ncm-dbt-03 585421 500 187 87 226 +70.44 ± 14.2 0 20 110 120 0 +147.19 ± 32.57
382557 ncm-dbt-01 583070 500 187 83 230 +73.34 ± 14.58 2 14 114 118 2 +153.86 ± 31.85
382556 ncm-dbt-04 572921 500 181 70 249 +78.43 ± 13.58 0 12 117 119 2 +162.35 ± 31.23
382555 ncm-dbt-02 589197 500 185 69 246 +82.1 ± 13.83 0 13 110 125 2 +171.02 ± 32.42
382554 ncm-dbt-05 580904 500 170 69 261 +71.16 ± 13.64 0 15 120 114 1 +147.19 ± 30.88
382553 ncm-dbt-03 586731 500 181 78 241 +72.61 ± 13.84 0 15 119 114 2 +148.85 ± 31.04
382552 ncm-dbt-04 569390 500 184 73 243 +78.43 ± 13.58 0 13 114 122 1 +164.07 ± 31.75
382551 ncm-dbt-02 587792 500 188 88 224 +70.43 ± 14.34 0 18 117 112 3 +142.26 ± 31.44
382550 ncm-dbt-01 585970 500 190 68 242 +86.52 ± 14.23 0 14 103 130 3 +179.9 ± 33.65
382549 ncm-dbt-03 587325 500 180 67 253 +79.9 ± 13.92 1 12 111 125 1 +169.27 ± 32.25
382548 ncm-dbt-05 584706 500 191 78 231 +79.9 ± 13.77 0 14 110 125 1 +167.53 ± 32.45
382547 ncm-dbt-04 573365 500 191 78 231 +79.9 ± 13.31 0 11 116 122 1 +167.53 ± 31.34
382546 ncm-dbt-02 586477 500 190 66 244 +88.0 ± 13.35 1 7 110 131 1 +189.0 ± 32.17
382545 ncm-dbt-03 586435 500 180 75 245 +74.06 ± 13.89 0 17 111 122 0 +155.54 ± 32.36
382544 ncm-dbt-05 581569 500 189 74 237 +81.36 ± 14.4 0 17 103 128 2 +169.27 ± 33.7
382543 ncm-dbt-01 580779 500 179 77 244 +71.89 ± 14.25 1 16 114 118 1 +150.51 ± 31.88
382542 ncm-dbt-04 572638 500 184 64 252 +85.04 ± 13.75 0 13 105 131 1 +179.9 ± 33.27
382541 ncm-dbt-03 583279 500 189 70 241 +84.3 ± 14.33 0 14 107 125 4 +172.78 ± 32.96
382540 ncm-dbt-01 585042 500 183 70 247 +79.9 ± 13.77 0 14 110 125 1 +167.53 ± 32.45
382539 ncm-dbt-02 587537 500 185 57 258 +90.96 ± 14.62 0 15 96 135 4 +189.0 ± 34.95
382538 ncm-dbt-05 584664 500 180 73 247 +75.52 ± 14.37 1 15 112 120 2 +157.24 ± 32.18
382537 ncm-dbt-04 573163 500 169 78 253 +63.94 ± 14.37 0 21 119 108 2 +129.35 ± 31.2
382536 ncm-dbt-03 589882 500 178 73 249 +74.06 ± 14.32 0 20 105 125 0 +155.54 ± 33.38
382535 ncm-dbt-05 586562 500 197 83 220 +80.63 ± 14.38 0 14 113 118 5 +162.35 ± 31.95
382534 ncm-dbt-02 587962 500 187 76 237 +78.43 ± 13.88 0 15 110 124 1 +164.07 ± 32.48
382533 ncm-dbt-01 583531 500 188 72 240 +82.1 ± 13.83 0 14 107 128 1 +172.78 ± 32.96

Commit

Commit ID 49ef4c935a5cb0e4d94096e6354caa06b36b3e3c
Author gab8192
Date 2024-04-24 16:38:20 UTC
Implement accumulator refresh table For each thread persist an accumulator cache for the network, where each cache contains multiple entries for each of the possible king squares. When the accumulator needs to be refreshed, the cached entry is used to more efficiently update the accumulator, instead of rebuilding it from scratch. This idea, was first described by Luecx (author of Koivisto) and is commonly referred to as "Finny Tables". When the accumulator needs to be refreshed, instead of filling it with biases and adding every piece from scratch, we... 1. Take the `AccumulatorRefreshEntry` associated with the new king bucket 2. Calculate the features to activate and deactivate (from differences between bitboards in the entry and bitboards of the actual position) 3. Apply the updates on the refresh entry 4. Copy the content of the refresh entry accumulator to the accumulator we were refreshing 5. Copy the bitboards from the position to the refresh entry, to match the newly updated accumulator Results at STC: https://tests.stockfishchess.org/tests/view/662301573fe04ce4cefc1386 (first version) https://tests.stockfishchess.org/tests/view/6627fa063fe04ce4cefc6560 (final) Non-Regression between first and final: https://tests.stockfishchess.org/tests/view/662801e33fe04ce4cefc660a STC SMP: https://tests.stockfishchess.org/tests/view/662808133fe04ce4cefc667c closes https://github.com/official-stockfish/Stockfish/pull/5183 No functional change
Copyright 2011–2024 Next Chess Move LLC