Dev Builds » 20171218-1532

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 06:20:09 2000365 2595 966 91 1538 +121.92 +/- 8.17
ncm-et-4 06:20:29 1991201 2564 1007 100 1457 +128.45 +/- 8.51
ncm-et-5 02:12:23 2004882 911 357 29 525 +130.96 +/- 14.07
ncm-et-6 02:12:07 2022467 907 356 30 521 +130.71 +/- 14.15
ncm-et-7 02:12:10 1997718 897 340 29 528 +125.67 +/- 13.96
ncm-et-8 02:12:13 2024590 894 329 43 522 +115.19 +/- 14.22
ncm-et-9 06:20:28 2001486 2560 1007 86 1467 +130.85 +/- 8.43
ncm-et-10 06:20:22 1986378 2537 980 71 1486 +130.26 +/- 8.3
ncm-et-11 02:12:00 1996493 898 340 31 527 +124.64 +/- 14.0
ncm-et-12 02:12:12 1989757 883 334 27 522 +126.05 +/- 14.0
ncm-et-13 06:20:18 1995184 2571 1020 85 1466 +132.41 +/- 8.44
ncm-et-14 00:17:45 2019696 118 57 5 56 +164.36 +/- 45.17
ncm-et-15 04:08:03 1974717 1665 657 59 949 +130.61 +/- 10.52
20000 7750 686 11564 +128.24 +/- 3.0

Test Detail

ID Host Started (UTC) Duration Base NPS Games Wins Losses Draws Elo CLI PGN
54468 ncm-et-10 2018-09-20 06:29 00:21:18 1962552 142 59 6 77 +136.26 +/- 37.68 ↓
54467 ncm-et-15 2018-09-20 06:26 00:23:32 1982636 165 57 6 102 +111.02 +/- 31.47 ↓
54466 ncm-et-9 2018-09-20 06:26 00:23:28 1989085 158 63 5 90 +133.78 +/- 34.2 ↓
54465 ncm-et-13 2018-09-20 06:25 00:25:15 1981939 167 52 8 107 +93.75 +/- 30.58 ↓
54464 ncm-et-4 2018-09-20 06:24 00:25:50 1989716 177 68 6 103 +127.08 +/- 31.83 ↓
54463 ncm-et-3 2018-09-20 06:21 00:28:33 1984991 191 74 5 112 +131.45 +/- 30.26 ↓
54462 ncm-et-10 2018-09-20 05:12 01:15:12 1980757 500 173 10 317 +117.55 +/- 17.38 ↓
54461 ncm-et-15 2018-09-20 05:12 01:13:33 1982478 500 205 16 279 +138.18 +/- 19.46 ↓
54460 ncm-et-9 2018-09-20 05:10 01:15:01 1989086 500 181 18 301 +117.55 +/- 18.43 ↓
54459 ncm-et-13 2018-09-20 05:09 01:14:40 1990033 500 194 20 286 +126.18 +/- 19.22 ↓
54458 ncm-et-4 2018-09-20 05:09 01:14:08 1988296 500 199 14 287 +134.95 +/- 19.0 ↓
54457 ncm-et-3 2018-09-20 05:08 01:11:42 1988927 500 188 20 292 +121.46 +/- 18.92 ↓
54456 ncm-et-10 2018-09-20 03:56 01:15:21 1983890 500 199 17 284 +132.54 +/- 19.24 ↓
54455 ncm-et-15 2018-09-20 03:56 01:14:34 1984204 500 182 19 299 +117.55 +/- 18.55 ↓
54454 ncm-et-9 2018-09-20 03:54 01:14:33 1991140 500 209 18 273 +139.81 +/- 19.82 ↓
54453 ncm-et-13 2018-09-20 03:54 01:13:29 1986880 500 201 16 283 +134.95 +/- 19.26 ↓
54452 ncm-et-4 2018-09-20 03:54 01:14:05 1989717 500 202 23 275 +130.15 +/- 19.85 ↓
54451 ncm-et-3 2018-09-20 03:52 01:15:02 1988138 500 172 19 309 +109.83 +/- 18.06 ↓
54450 ncm-et-3 2018-09-20 02:38 01:12:41 1987352 500 182 14 304 +121.46 +/- 18.16 ↓
54449 ncm-et-10 2018-09-20 02:38 01:16:24 1949402 500 197 16 287 +131.74 +/- 19.06 ↓
54448 ncm-et-15 2018-09-20 02:38 01:16:24 1949553 500 213 18 269 +143.07 +/- 20.03 ↓
54447 ncm-et-13 2018-09-20 02:38 01:14:54 1961277 500 200 15 285 +134.95 +/- 19.13 ↓
54446 ncm-et-9 2018-09-20 02:38 01:15:01 1988298 500 203 14 283 +138.18 +/- 19.2 ↓
54445 ncm-et-4 2018-09-20 02:38 01:14:15 1990981 500 195 24 281 +123.81 +/- 19.57 ↓
4074 ncm-et-14 2018-03-01 21:28 00:17:45 2019696 118 57 5 56 +164.36 +/- 45.17 ↓
4073 ncm-et-12 2018-03-01 20:50 00:55:56 2022306 383 156 11 216 +138.42 +/- 22.02 ↓
4072 ncm-et-4 2018-03-01 20:49 00:57:48 2022629 387 154 14 219 +131.65 +/- 21.97 ↓
4071 ncm-et-3 2018-03-01 20:48 00:58:10 2027702 404 154 18 232 +121.7 +/- 21.39 ↓
4070 ncm-et-11 2018-03-01 20:48 00:58:07 2013531 398 139 10 249 +116.82 +/- 19.82 ↓
4069 ncm-et-8 2018-03-01 20:48 00:58:23 2024590 394 147 19 228 +117.12 +/- 21.59 ↓
4068 ncm-et-6 2018-03-01 20:48 00:58:26 2022142 407 155 14 238 +125.56 +/- 20.88 ↓
4067 ncm-et-10 2018-03-01 20:48 00:58:18 2021325 395 159 10 226 +137.87 +/- 21.39 ↓
4066 ncm-et-13 2018-03-01 20:48 00:58:37 2025736 404 166 12 226 +139.48 +/- 21.58 ↓
4065 ncm-et-7 2018-03-01 20:48 00:58:44 1996974 397 152 15 230 +125.03 +/- 21.34 ↓
4064 ncm-et-9 2018-03-01 20:47 00:59:44 2024754 402 159 17 226 +128.25 +/- 21.75 ↓
4063 ncm-et-5 2018-03-01 20:46 01:00:02 2023610 411 166 14 231 +134.88 +/- 21.39 ↓
4062 ncm-et-13 2018-03-01 19:33 01:13:23 2025244 500 207 14 279 +141.44 +/- 19.4 ↓
4061 ncm-et-11 2018-03-01 19:33 01:13:53 1979456 500 201 21 278 +130.94 +/- 19.65 ↓
4060 ncm-et-6 2018-03-01 19:33 01:13:41 2022793 500 201 16 283 +134.95 +/- 19.26 ↓
4059 ncm-et-4 2018-03-01 19:33 01:14:23 1965871 500 189 19 292 +123.02 +/- 18.9 ↓
4058 ncm-et-7 2018-03-01 19:33 01:13:26 1998463 500 188 14 298 +126.18 +/- 18.45 ↓
4057 ncm-et-12 2018-03-01 19:33 01:16:16 1957209 500 178 16 306 +116.78 +/- 18.12 ↓
4056 ncm-et-3 2018-03-01 19:33 01:14:01 2025082 500 196 15 289 +131.74 +/- 18.93 ↓
4055 ncm-et-5 2018-03-01 19:33 01:12:21 1986154 500 191 15 294 +127.76 +/- 18.68 ↓
4054 ncm-et-8 2018-03-01 19:33 01:13:50 2024590 500 182 24 294 +113.68 +/- 18.93 ↓
4053 ncm-et-9 2018-03-01 19:33 01:12:41 2026554 500 192 14 294 +129.35 +/- 18.65 ↓
4052 ncm-et-10 2018-03-01 19:33 01:13:49 2020347 500 193 12 295 +131.74 +/- 18.53 ↓

Commit

Commit ID 2198cd0524574f0d9df8c0ec9aaf14ad8c94402b
Author Joost VandeVondele
Date 2017-12-18 15:32:21 UTC
Allow for general transposition table sizes. (#1341) For efficiency reasons current master only allows for transposition table sizes that are N = 2^k in size, the index computation can be done efficiently as (hash % N) can be written instead as (hash & 2^k - 1). On a typical computer (with 4, 8... etc Gb of RAM), this implies roughly half the RAM is left unused in analysis. This issue was mentioned on fishcooking by Mindbreaker: http://tests.stockfishchess.org/tests/view/5a3587de0ebc590ccbb8be04 Recently a neat trick was proposed to map a hash into the range [0,N[ more efficiently than (hash % N) for general N, nearly as efficiently as (hash % 2^k): https://lemire.me/blog/2016/06/27/a-fast-alternative-to-the-modulo-reduction/ namely computing (hash * N / 2^32) for 32 bit hashes. This patch implements this trick and now allows for general hash sizes. Note that for N = 2^k this just amounts to using a different subset of bits from the hash. Master will use the lower k bits, this trick will use the upper k bits (of the 32 bit hash). There is no slowdown as measured with [-3, 1] test: http://tests.stockfishchess.org/tests/view/5a3587de0ebc590ccbb8be04 LLR: 2.96 (-2.94,2.94) [-3.00,1.00] Total: 128498 W: 23332 L: 23395 D: 81771 There are two (smaller) caveats: 1) the patch is implemented for a 32 bit hash (so that a 64 bit multiply can be used), this effectively limits the number of clusters that can be used to 2^32 or to 128Gb of transpostion table. That's a change in the maximum allowed TT size, which could bother those using 256Gb or more regularly. 2) Already in master, an excluded move is hashed into the position key in rather simple way, essentially only affecting the lower 16 bits of the key. This is OK in master, since bits 0-15 end up in the index, but not in the new scheme, which picks the higher bits. This is 'fixed' by shifting the excluded move a few bits up. Eventually a better hashing scheme seems wise. Despite these two caveats, I think this is a nice improvement in usability. Bench: 5346341
Copyright 2011–2024 Next Chess Move LLC
Feedback & Support Terms of Service Privacy Policy