]> git.sesse.net Git - stockfish/history - src/nnue
Smooth improving
[stockfish] / src / nnue /
2021-10-07 xoto10Small clean-up, Sept 2021
2021-09-21 Michael ChalyCombo of various parameter tweaks
2021-08-20 Tomasz SobczykOptimize and tidy up affine transform code.
2021-08-15 Tomasz SobczykNew NNUE architecture and net
2021-07-30 Tomasz SobczykAvoid unnecessary stores in the affine transform
2021-07-03 Stéphane NicoletSimplify format_cp_aligned_dot()
2021-06-21 Joost VandeVondeleFix build error on OSX
2021-06-19 Tomasz SobczykChange trace with NNUE eval support
2021-06-14 Tomasz SobczykReduce the number of accumulator states
2021-06-13 Tomasz SobczykRegister count for feature transformer
2021-06-13 Stéphane NicoletClarify use of UCI options
2021-06-13 Tomasz SobczykRead NNUE net faster
2021-06-04 Stéphane NicoletClean SIMD code a bit
2021-05-30 Tomasz SobczykFix export of the feature transformer.
2021-05-26 Stéphane NicoletDo not use lazy evaluation inside NNUE
2021-05-25 Tomasz SobczykExpose the lazy threshold for the feature transformer...
2021-05-22 Stéphane NicoletSometimes change the (materialist, positional) balance
2021-05-19 Fanael LinithienUse packed 32-bit MMX operations for updating the PSQT...
2021-05-18 Tomasz SobczykNew NNUE architecture and net
2021-05-17 Stéphane NicoletSmall clean-ups
2021-05-11 Tomasz SobczykExporting the currently loaded network file
2021-04-25 Tomasz SobczykCleanup and simplify NNUE code.
2021-04-24 Tomasz SobczykUnify naming convention of the NNUE code
2021-04-15 Tomasz SobczykDocumentation patch: AppendChangedIndices
2021-03-24 Stéphane NicoletSmall cleanups (march 2021)
2021-03-19 Guy VreulsUse reference instead of pointer for pop_lsb() signature
2021-03-07 Dieter DobbelaereAdd Stockfish namespace.
2021-01-11 MaximMolchanovAffine transform robust implementation
2021-01-08 Joost VandeVondeleUpdate copyright years
2021-01-08 MaximMolchanovAffine transform refactoring.
2020-12-14 mstemberaAVX512, AVX2 and SSSE3 speedups
2020-12-12 Fanael LinithienUse arithmetic right shift for sign extension in MMX...
2020-11-29 mstemberaCleaned up MakeIndex()
2020-11-25 MaximMolchanovCalculate sum from first elements
2020-11-23 Stéphane NicoletSmall cleanups 13
2020-11-07 Tomasz SobczykAVX-512 for smaller affine and feature transforms.
2020-11-04 Tomasz SobczykManually align arrays on the stack
2020-11-02 Tomasz SobczykOptimize affine transform for SSSE3 and higher targets.
2020-10-22 syzygy1More incremental accumulator updates
2020-09-28 noobpwnftwUse incremental updates more often
2020-09-21 Stéphane NicoletSmall cleanups 12
2020-09-21 Sami KiminkiAdd large page support for NNUE weights and simplify...
2020-09-17 syzygy1Use tiling to speed up accumulator refreshes and updates
2020-09-08 syzygy1Bug fix in do_null_move() and NNUE simplification.
2020-08-29 Stéphane NicoletEmbed default net, and simplify using non-default nets
2020-08-26 syzygy1Remove EvalList
2020-08-24 mstemberaSupport VNNI on 256bit vectors
2020-08-23 syzygy1Skip the alignment bug workaround for Clang
2020-08-17 Stéphane NicoletReformat code in little-endian patch
2020-08-16 syzygy1Assume network file is in little-endian byte order
2020-08-14 mstemberaTry to match relative magnitude of NNUE eval to classical
2020-08-13 mstemberaAdd support for VNNI
2020-08-11 Joost VandeVondeleRestrict avx2 hack to windows target
2020-08-11 mstemberaFix AVX512 build with older compilers
2020-08-10 Fanael LinithienProvide vectorized NNUE code for SSE2 and MMX targets
2020-08-10 mstemberaCleanup and optimize SSE/AVX code
2020-08-10 mstemberaWorkaround using unaligned loads for gcc < 9
2020-08-10 Joost VandeVondeleRevert "Avoid special casing for MinGW"
2020-08-09 Dariusz OrzechowskiAvoid special casing for MinGW
2020-08-06 nodchipAdd NNUE evaluation