LondonFrau Admin
Posts : 1292Reputation : 3534Join date : 2010-02-27Location : ??? Subject: Latest abrok Development Version December 19th (SF The gift that keeps on giving Wed Dec 20, 2023 3:16 pm !! latest version !! Windows x64 for Haswell CPUs Windows x64 for modern computers + AVX2 Windows x64 for modern computers Windows x64 + SSSE3 Windows x64 Linux x64 for Haswell CPUs Linux x64 for modern computers + AVX2 Linux x64 for modern computers Linux x64 + SSSE3 Linux x64 Author: Michael Chaly Date: Tue Dec 19 18:22:10 2023 +0100 Timestamp: 1703006530 Adjust return value in qsearch after fail high Instead of returning strict fail soft fail high return value between value from search and beta (somewhat by analogy to futility pruning and probcut). This seems to be somewhat depth sensitive heuristic which performed much worse at LTC while performing much better at STC if it is more aggressive, passed version is the least aggressive one. Passed STC: https://tests.stockfishchess.org/tests/view/657b06414d789acf40ab1475 LLR: 2.95 (-2.94,2.94) <0.00,2.00> Total: 212352 W: 53900 L: 53315 D: 105137 Elo +0.96 Ptnml(0-2): 809, 25236, 53520, 25783, 828 Passed LTC: https://tests.stockfishchess.org/tests/view/657ce36f393ac02e79120a7c LLR: 2.94 (-2.94,2.94) <0.50,2.50> Total: 319362 W: 79541 L: 78630 D: 161191 Elo +0.99 Ptnml(0-2): 202, 35839, 86709, 36708, 223 closes https://github.com/official-stockfish/Stockfish/pull/4928 Bench: 974739 see source Windows x64 for Haswell CPUs Windows x64 for modern computers + AVX2 Windows x64 for modern computers Windows x64 + SSSE3 Windows x64 Linux x64 for Haswell CPUs Linux x64 for modern computers + AVX2 Linux x64 for modern computers Linux x64 + SSSE3 Linux x64 Author: FauziAkram Date: Tue Dec 19 18:22:10 2023 +0100 Timestamp: 1703006530 Use std::abs over abs closes https://github.com/official-stockfish/Stockfish/pull/4926 closes https://github.com/official-stockfish/Stockfish/pull/4909 No functional change Co-Authored-By: fffelix-huang <72808219+> see source Windows x64 for Haswell CPUs Windows x64 for modern computers + AVX2 Windows x64 for modern computers Windows x64 + SSSE3 Windows x64 Linux x64 for Haswell CPUs Linux x64 for modern computers + AVX2 Linux x64 for modern computers Linux x64 + SSSE3 Linux x64 Author: FauziAkram Date: Tue Dec 19 18:22:10 2023 +0100 Timestamp: 1703006530 Improvement of Time Management Parameters Passed STC: https://tests.stockfishchess.org/tests/view/6579c5574d789acf40aaf914 LLR: 2.93 (-2.94,2.94) <0.00,2.00> Total: 44672 W: 11354 L: 11030 D: 22288 Elo +2.52 Ptnml(0-2): 140, 5033, 11685, 5319, 159 Passed LTC: https://tests.stockfishchess.org/tests/view/657ad7f44d789acf40ab105e LLR: 2.94 (-2.94,2.94) <0.50,2.50> Total: 40932 W: 10275 L: 9950 D: 20707 Elo +2.76 Ptnml(0-2): 21, 4316, 11473, 4629, 27 Passed non-regression Sudden death 10+0: https://tests.stockfishchess.org/tests/view/657b9b9e393ac02e7911f1a8 LLR: 2.93 (-2.94,2.94) <-1.75,0.25> Total: 21384 W: 5171 L: 4925 D: 11288 Elo +4.00 Ptnml(0-2): 112, 2420, 5409, 2612, 139 closes https://github.com/official-stockfish/Stockfish/pull/4923 No functional change see source Windows x64 for Haswell CPUs Windows x64 for modern computers + AVX2 Windows x64 for modern computers Windows x64 + SSSE3 Windows x64 Linux x64 for Haswell CPUs Linux x64 for modern computers + AVX2 Linux x64 for modern computers Linux x64 + SSSE3 Linux x64 Author: Michael Chaly Date: Tue Dec 19 18:22:10 2023 +0100 Timestamp: 1703006530 Adjust stand pat in qsearch on pv nodes Instead of immediately returning a fail high do this only at non-pv nodes, for pv nodes adjust bestValue to value between alpha and beta and continue searching. Idea is to do it the same way as it's done in search where we don't return positive beta cutoffs after ttHits / zero window search at PvNodes and instead fully search lines. Passed STC: https://tests.stockfishchess.org/tests/view/65739b0af09ce1261f122f33 LLR: 2.94 (-2.94,2.94) <0.00,2.00> Total: 189216 W: 48142 L: 47598 D: 93476 Elo +1.00 Ptnml(0-2): 584, 22463, 48051, 22845, 665 Passed LTC: https://tests.stockfishchess.org/tests/view/657701214d789acf40aac194 LLR: 2.94 (-2.94,2.94) <0.50,2.50> Total: 82506 W: 20689 L: 20269 D: 41548 Elo +1.77 Ptnml(0-2): 56, 9236, 22268, 9618, 75 Two issues had to be resolved: - in rare cases it set alpha to the same value as beta and thus broke some asserts; - messed up with returning tb win values. Fix passed non-regression LTC vs this patch: https://tests.stockfishchess.org/tests/view/6578113b4d789acf40aad544 LLR: 2.96 (-2.94,2.94) <-1.75,0.25> Total: 277308 W: 68839 L: 68880 D: 139589 Elo -0.05 Ptnml(0-2): 167, 31580, 75212, 31517, 178 closes https://github.com/official-stockfish/Stockfish/pull/4922 Bench: 1069503 Co-Authored-By: Muzhen Gaming <61100393+> Co-Authored-By: Shahin M. Shahin <41402573+> Co-Authored-By: fffelix-huang <72808219+> see source
_________________ Bettina...............The greatest happiness of life is the conviction that we are loved. Victor Hugo
vv56 likes this post
Like 1 Dislike