| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: DU-jdto
Date: Wed Jun 13 21:52:53 2018 +0200 Timestamp: 1528919573 Remove lmrDepth restriction on quiet see pruning And tweak the threshold value. With this threshold and the current piece values, this permits see pruning on quiets to be done up to an lmrDepth of 9 (beyond that the threshold is below -QueenValueMg and see_ge will pass unconditionally). STC: LLR: 2.96 (-2.94,2.94) [-3.00,1.00] Total: 110316 W: 24612 L: 24667 D: 61037 Elo -0.17 http://tests.stockfishchess.org/tests/view/5b20aa760ebc5902ab9c9c1d LTC: LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 17352 W: 2968 L: 2842 D: 11542 Elo +2.52 http://tests.stockfishchess.org/tests/view/5b20cf1e0ebc5902ab9c9fb6 Closes https://github.com/official-stockfish/Stockfish/pull/1651 Bench: 5069074 see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: Joost VandeVondele
Date: Wed Jun 13 21:46:12 2018 +0200 Timestamp: 1528919172 Remove depth condition for pruning captures. The SEE condition alone is sufficient. passed STC: LLR: 2.96 (-2.94,2.94) [-3.00,1.00] Total: 109863 W: 24339 L: 24392 D: 61132 Elo -0.17 http://tests.stockfishchess.org/tests/view/5b1f5b000ebc5902ab9c8668 passed LTC: LLR: 2.96 (-2.94,2.94) [-3.00,1.00] Total: 23390 W: 4020 L: 3903 D: 15467 Elo +1.74 http://tests.stockfishchess.org/tests/view/5b1f94b40ebc5902ab9c8b5e Closes https://github.com/official-stockfish/Stockfish/pull/1648 Bench: 4834747 see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: protonspring
Date: Mon Jun 11 10:22:55 2018 +0200 Timestamp: 1528705375 Optimize an expression in endgame.cpp I believe using foward_file_bb() here is fewer instructions. a) Fewer instructions and probably more clear (debatable). b) Possible that a lookup is slower than a few local operations, but the forward_file_bb table is probably used often enough that it is always cached. Passed LLR: 2.96 (-2.94,2.94) [-3.00,1.00] Total: 21004 W: 4263 L: 4141 D: 12600 Elo +2.02 http://tests.stockfishchess.org/tests/view/5b1cad830ebc5902ab9c6239 Closes https://github.com/official-stockfish/Stockfish/pull/1644 No functional change. see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: candirufish
Date: Mon Jun 11 10:03:33 2018 +0200 Timestamp: 1528704213 Simplify capture pruning margin formula Using just `PawnValueEg * depth` as Capture Prune Margin. There was a bunch of patches that passed recently regarding captures, maybe this part of the master code redundant? The patch was tested as a simplification: STC: LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 20833 W: 4218 L: 4096 D: 12519 Elo +2.03 http://tests.stockfishchess.org/tests/view/5b1cf2100ebc5902ab9c6651 LTC: LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 27050 W: 3975 L: 3864 D: 19211 Elo +1.43 http://tests.stockfishchess.org/tests/view/5b1cfdc80ebc5902ab9c6776 Closes https://github.com/official-stockfish/Stockfish/pull/1643 Bench: 4980482 see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: Stefan Geschwentner
Date: Mon Jun 11 09:54:32 2018 +0200 Timestamp: 1528703672 Optimize contempt value (21) After several tests it seems best to increase contempt from 12 to 21. This does not regress against contempt=0 and gives a gain of around 7-8 elo against SF 7 in comparison to current default contempt. STC: Test for non-regression contempt=21 against contempt=0 LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 71250 W: 13956 L: 13926 D: 43368 Elo +0.15 http://tests.stockfishchess.org/tests/view/5b19a58d0ebc5902ab9c3bfa STC: Test contempt 21 against SF 7 ELO: 190.06 +-2.8 (95%) LOS: 100.0% Total: 40000 W: 22608 L: 2676 D: 14716 Elo +190.06 http://tests.stockfishchess.org/tests/view/5b19a6520ebc5902ab9c3c0e STC: Test master against SF 7 for comparison ELO: 182.95 +-2.7 (95%) LOS: 100.0% Total: 40000 W: 21905 L: 2595 D: 15500 Elo +182.95 http://tests.stockfishchess.org/tests/view/5b16f5bc0ebc59214346d5ca LTC: Test for non-regression contempt=21 against contempt=0 LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 47666 W: 6914 L: 6832 D: 33920 Elo +0.60 http://tests.stockfishchess.org/tests/view/5b1a170b0ebc5902ab9c3fde LTC: Test contempt 21 against SF 7 ELO: 203.92 +-2.6 (95%) LOS: 100.0% Total: 40000 W: 22447 L: 1340 D: 16213 Elo +203.92 http://tests.stockfishchess.org/tests/view/5b1a174b0ebc5902ab9c3fe1 LTC: Test master against SF 7 for comparison ELO: 196.08 +-2.6 (95%) LOS: 100.0% Total: 40000 W: 21639 L: 1191 D: 17170 Elo +196.08 http://tests.stockfishchess.org/tests/view/5b1a17e40ebc5902ab9c3fe4 Closes https://github.com/official-stockfish/Stockfish/pull/1646 Bench: 4786912 see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: Joost VandeVondele
Date: Mon Jun 11 09:46:26 2018 +0200 Timestamp: 1528703186 Remove failedLow from the mainThread struct failedLow is in fact a local variable in Thread::search(). Also clean some spaces and tabs in code. No functional change. see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: candirufish
Date: Wed Jun 6 10:16:48 2018 +0200 Timestamp: 1528273008 Quiet move soft fail high bonus Extra bonus for quiet move creating a huge soft fail high (triggered in 21% of quiet bestmoves on a normal bench run). Pb00067 original idea using PawnValueMg. Passed STC: LLR: 2.95 (-2.94,2.94) [0.00,5.00] Total: 138207 W: 28060 L: 27295 D: 82852 Elo +1.92 http://tests.stockfishchess.org/tests/view/5b14471b0ebc5902a81689c1 Passe LTC: LLR: 2.94 (-2.94,2.94) [0.00,5.00] Total: 157289 W: 23200 L: 22518 D: 111571 Elo +1.51 http://tests.stockfishchess.org/tests/view/5b149dde0ebc5902a8b41c5a bench: 4441320 see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: VoyagerOne
Date: Wed Jun 6 09:58:34 2018 +0200 Timestamp: 1528271914 Simplify capture bonus Simplify capture bonus by simply adding ONE_DEPTH instead of being dependent on BestValue. STC: LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 24419 W: 4939 L: 4824 D: 14656 Elo +1.64 http://tests.stockfishchess.org/tests/view/5b16b2040ebc5963ba37e2a5 LTC: LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 44560 W: 6524 L: 6438 D: 31598 Elo +0.67 http://tests.stockfishchess.org/tests/view/5b16ccc00ebc59214346d403 Closes https://github.com/official-stockfish/Stockfish/pull/1640 Bench: 4782637 see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: Stéphane Nicolet
Date: Tue Jun 5 18:15:54 2018 +0200 Timestamp: 1528215354 Call cycle detection before qsearch() This has the property of raising alpha before calling qsearch(), thus maybe giving some more cuts during qsearch(). The patch is equivalent to the use of cycle detection inside qsearch() at depth 0, but is in fact implemented by re-ordering code inside search(), which explains the [0..4] bounds in the following tests. STC (interrupted after 124250 games, with LLR=0.87): http://tests.stockfishchess.org/tests/view/5b1500bd0ebc5902a8b420bf LLR: 0.87 (-2.94,2.94) [0.00,4.00] Total: 124250 W: 24973 L: 24470 D: 74807 Elo +1.41 LTC: http://tests.stockfishchess.org/tests/view/5b1590eb0ebc5902a84dcd09 LLR: 2.96 (-2.94,2.94) [0.00,4.00] Total: 74234 W: 11098 L: 10733 D: 52403 Elo +1.71 Closes https://github.com/official-stockfish/Stockfish/pull/1635 Bench: 4326784 see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: Joost VandeVondele
Date: Tue Jun 5 18:06:13 2018 +0200 Timestamp: 1528214773 Stabilize AppVeyor CI After a helpful suggestion from AppVeyor support staff, moving the Stockfish execution from ps to cmd seems to work. Alternative to PR #1624 tested in PR #1637. No functional change. see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: Guenther Demetz
Date: Tue Jun 5 10:10:32 2018 +0200 Timestamp: 1528186232 Remove a superfluous subtrahend The '- 1' subtrahend was introduced for guarding against null move search at root, which would be nonsense. But this is actually already guaranteed by the !PvNode condition. This followed from the discussion in pull request 1609: https://github.com/official-stockfish/Stockfish/pull/1609 No functional change see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: VoyagerOne
Date: Tue Jun 5 09:58:10 2018 +0200 Timestamp: 1528185490 Simplify LMR for captures Simplify LMR for captures by removing capture's stat score logic for reduction. STC: LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 37068 W: 7462 L: 7370 D: 22236 Elo +0.86 http://tests.stockfishchess.org/tests/view/5b115bc30ebc591af58a6fd2 LTC: LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 80061 W: 11706 L: 11671 D: 56684 Elo +0.15 http://tests.stockfishchess.org/tests/view/5b117f590ebc59033d2d5315 Closes https://github.com/official-stockfish/Stockfish/pull/1631 Bench: 4470519 see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: joergoster
Date: Mon Jun 4 07:45:12 2018 +0200 Timestamp: 1528091112 Bugfix of Position::has_repeated() The function Position::has_repeated() is used by Tablebases::root_probe() to determine whether we can rank all winning moves with the same value, or if we need to strictly rank by dtz in case the position has already been repeated once, and we are risking to run into the 50-move rule and thus losing the win (especially critical in some very complicated endgames). To check whether the current position or one of the previous positions after the last zeroing move has already been occured once, we start looking for a repetition of the current position, and if that is not the case, we step one position back and repeat the check for that position, and so on. If you now look at how this was done before the new root ranking patch was merged two months ago, it seems quite obvious that it is a simple oversight: https://github.com/official-stockfish/Stockfish/commit/108f0da4d7f993732aa2e854b8f3fa8ca6d3b46c More specifically, after we stepped one position back with ``` stc = stc->previous; ``` we now have to start checking for a repetition with ``` StateInfo* stp = stc->previous->previous; ``` and not with ``` StateInfo* stp = st->previous->previous; ``` Closes https://github.com/official-stockfish/Stockfish/pull/1625 No functional change see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: Stéphane Nicolet
Date: Sun Jun 3 09:19:13 2018 +0200 Timestamp: 1528010353 Fix overload ambiguity Fix an error when compiling current master with MSVC due to the ambiguity of which operator* overload was intended (reported by Jarrod Torriero). No functional change. see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: pb00068
Date: Sat Jun 2 18:02:35 2018 +0200 Timestamp: 1527955355 Extra bonus for capture creating a huge fail high STC: http://tests.stockfishchess.org/tests/view/5b114f3d0ebc596e9e0881f6 LLR: 2.95 (-2.94,2.94) [0.00,5.00] Total: 13007 W: 2730 L: 2541 D: 7736 Elo +5.05 LTC: http://tests.stockfishchess.org/tests/view/5b1176740ebc59033d2d52c6 LLR: 2.96 (-2.94,2.94) [0.00,5.00] Total: 15594 W: 2417 L: 2239 D: 10938 Elo +3.97 See https://github.com/official-stockfish/Stockfish/pull/1627 Bench: 4790240 see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: Stéphane Nicolet
Date: Sat Jun 2 17:41:37 2018 +0200 Timestamp: 1527954097 Fix comments, rename variables Thanks everybody for the various hints in the perpetual renaming thread: https://github.com/official-stockfish/Stockfish/issues/1426 No functional change see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: Joost VandeVondele
Date: Sat Jun 2 17:03:01 2018 +0200 Timestamp: 1527951781 Reallocate TT on threadpool resize. Makes sure the potential benefit of first touch does not depend on the order of the UCI commands Threads and Hash, by reallocating the hash if a Threads is issued. The cost is zeroing the TT once more than needed. In case the prefered order (first Threads than Hash) is employed, this amounts to zeroing the default sized TT (16Mb), which is essentially instantaneous. Follow up for https://github.com/official-stockfish/Stockfish/pull/1601 where additional data and discussion is available. Closes https://github.com/official-stockfish/Stockfish/pull/1620 No functional change. see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: Joost VandeVondele
Date: Wed May 30 13:38:11 2018 +0200 Timestamp: 1527680291 Simplify depth reduction in IID. Use a constant reduction instead of a depth dependent reduction. passed STC: http://tests.stockfishchess.org/tests/view/5b06eb600ebc5914abc12ba8 LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 58086 W: 11710 L: 11657 D: 34719 Elo +0.32 passed LTC: http://tests.stockfishchess.org/tests/view/5b07b25e0ebc5914abc12c6d LLR: 3.09 (-2.94,2.94) [-3.00,1.00] Total: 110414 W: 16217 L: 16219 D: 77978 Elo -0.01 Bench: 4521056 see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: Rodrigo Exterckötter Tjäder
Date: Wed May 30 13:07:18 2018 +0200 Timestamp: 1527678438 Use threads to clear the TT faster. Stockfish currently takes a while to clear the TT when using larger hash sizes. On one machine with 128 GB hash it takes about 50 seconds with a single thread, allowing it to use all allocated cores brought that time down to 4 seconds on some Linux systems. The patch was further tested on Windows and refined with NUMA binding of the hash initializing threads (we refer to pull request #1601 for the complete discussion and the speed measurements). Closes https://github.com/official-stockfish/Stockfish/pull/1601 No functional change see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: VoyagerOne
Date: Thu May 24 19:06:24 2018 +0200 Timestamp: 1527181584 LMR Capture Tweak Increase capture reduction by comparing opponent's stat score STC: (Yellow) LLR: -2.96 (-2.94,2.94) [0.00,5.00] Total: 92291 W: 18647 L: 18350 D: 55294 Elo +1.12 http://tests.stockfishchess.org/tests/view/5b04225e0ebc5914abc1291b LTC: LLR: 2.95 (-2.94,2.94) [0.00,5.00] Total: 20176 W: 3061 L: 2868 D: 14247 Elo +3.32 http://tests.stockfishchess.org/tests/view/5b04bb3f0ebc5914abc129a3 How to continue? • we could try a different value for the capture history threeshold Closes https://github.com/official-stockfish/Stockfish/pull/1612 Bench: 5167159 see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: protonspring
Date: Thu May 24 18:46:55 2018 +0200 Timestamp: 1527180415 Simplify BlockedByPawn to one dimension I was able to get this to pass which reduces BlockedByPawn to one dimension with NO distance from edge offset. GOOD) It's more simple and may provide additional clarity for further simplifications. Facilitates migrating unblocked to one dimension as well. BAD) If there is indeed a distance component to BlockedStorm (may or may not be the case), this obfuscates this component into ShelterStrength and UnblockedStorm. This may be more convoluted. Also, it may be more convenient to have each of the three arrays (ShelterStrength, BlockedStorm, and UnBlocked) be the same size. STC: LLR: 2.96 (-2.94,2.94) [-3.00,1.00] Total: 96173 W: 19326 L: 19343 D: 57504 Elo -0.06 http://tests.stockfishchess.org/tests/view/5b04544d0ebc5914abc12965 LTC: LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 49818 W: 7441 L: 7363 D: 35014 Elo +0.54 http://tests.stockfishchess.org/tests/view/5b0487d50ebc5914abc12990 Closes https://github.com/official-stockfish/Stockfish/pull/1611 Bench: 5133208 see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: Joost VandeVondele
Date: Thu May 24 18:26:34 2018 +0200 Timestamp: 1527179194 Correctly prevent recursive verification in nmp. As discussed with @pb00068, the condition to prevent recursive verification was not completely correct. This patch corrects that condition, and adds an assert. In the current implementation, recursive verification needs to be avoided in order not to break the verification of the move closer to the root (i.e. to not set thisThread->nmp_min_ply to zero prematurely). This patch is tested as a bug fix, based on and tested against PR #1609 . passed STC: http://tests.stockfishchess.org/tests/view/5b050f170ebc5914abc129c5 LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 7875 W: 1669 L: 1523 D: 4683 Elo +6.44 passed LTC: http://tests.stockfishchess.org/tests/view/5b0513970ebc5914abc129cd LLR: 3.00 (-2.94,2.94) [-3.00,1.00] Total: 24048 W: 3525 L: 3407 D: 17116 Elo +1.70 Bench changes at high depth. Closes https://github.com/official-stockfish/Stockfish/pull/1610 and https://github.com/official-stockfish/Stockfish/pull/1609 Bench: 4484288 see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: Joost VandeVondele
Date: Tue May 22 22:43:17 2018 +0200 Timestamp: 1527021797 Simplify color usage in search. define Color us and use this instead of pos.side_to_move() and nmp_odd. The latter allows to clarify the nmp verification criterion. Tested for no regression: LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 76713 W: 15303 L: 15284 D: 46126 Elo +0.09 http://tests.stockfishchess.org/tests/view/5b046a0d0ebc5914abc12971 No functional change. see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: VoyagerOne
Date: Tue May 22 07:23:04 2018 +0200 Timestamp: 1526966584 Simply reset StatScore to zero at beta cutoff STC: LLR: 2.96 (-2.94,2.94) [-3.00,1.00] Total: 43154 W: 8706 L: 8625 D: 25823 Elo +0.65 http://tests.stockfishchess.org/tests/view/5b01a7660ebc5914abc1271d LTC: LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 48155 W: 7036 L: 6955 D: 34164 Elo +0.58 http://tests.stockfishchess.org/tests/view/5b01b2e50ebc5914abc1272c Closes https://github.com/official-stockfish/Stockfish/pull/1608 Bench: 4484288 see source |
| Windows x64 for Haswell CPUs Windows x64 for modern computers Windows x64 Windows 32 Linux x64 for Haswell CPUs Linux x64 for modern computers Linux x64 | Author: protonspring
Date: Mon May 21 09:51:05 2018 +0200 Timestamp: 1526889065 Simplify shelter: always do strength and danger This check of pawns before subtracting danger can be removed. STC LLR: 2.96 (-2.94,2.94) [-3.00,1.00] Total: 21174 W: 4361 L: 4239 D: 12574 Elo +2.00 http://tests.stockfishchess.org/tests/view/5b00b9f90ebc5914abc12680 LTC: LLR: 2.95 (-2.94,2.94) [-3.00,1.00] Total: 56980 W: 8377 L: 8309 D: 40294 Elo +0.41 http://tests.stockfishchess.org/tests/view/5b00ca750ebc5914abc12683 Closes https://github.com/official-stockfish/Stockfish/pull/1607 Bench: 4746692 see source |