1. 13d5770 Port fixedpoint header to WebAssembly SIMD (#202) by Marat Dukhan · 2 days ago master
  2. fda83bd Large-scale change: Mark static const class/struct members as constexpr. This change fixes declarations that have initial values but are technically not definitions by marking them constexpr (which counts as a definition). This enables, among other things, the modified constants to be passed into functions and function templates that accept arguments by reference. Without this change, such functions would cause linker errors. by Danny van der Rijn · 8 months ago
  3. 023c190 Add 16-bit support for AVX2 by T.J. Alumbaugh · 8 months ago
  4. 2483d84 added template specialization for int8_t saturating add by Lamar · 12 months ago
  5. fcdb1b3 fixed doc comments for Mul and Sub by Lamar · 12 months ago
  6. 08c5dd3 fix recent regression by Benoit Jacob · 1 year, 1 month ago
  7. e25e353 StoreFinalOutputImpl::Run is reloading MatrixMap::data_ and by Clement Courbet · 1 year, 1 month ago
  8. 6a007c7 Add some extra documentation to the magic numbers used in by Nick Kreeger · 1 year, 2 months ago
  9. dc69acd fix avx header include by Benoit Jacob · 1 year, 5 months ago
  10. a227af1 Fix the type to avoid comparison of integers of different signs. The compiling error is reported when -Wsign-compared is added. by Chao Mei · 1 year, 6 months ago
  11. 447abda Increase ProfilingStack::kMaxSize by Sean Silva · 1 year, 7 months ago
  12. 12fed0c Argh, we can't update MultiThreadGemm to use the new by Benoit Jacob · 1 year, 9 months ago
  13. 5495284 Another take on what a sane Execute method would look like, by Benoit Jacob · 1 year, 9 months ago
  14. f0077a5 unused variable by Benoit Jacob · 1 year, 9 months ago
  15. 948468b Rename Execute to LegacyExecuteAndDestroyTasks. Add better Execute overload. by Benoit Jacob · 1 year, 9 months ago
  16. 892c70f Making scripts compatible with Python 3 by dibyaran · 1 year, 9 months ago
  17. 36212ad Changed Microsoft-specific suffix for 'long long int' type to the C++ standard for the same type explained in the section '5.13 Literals'. by Arritmic · 1 year, 10 months ago
  18. dec2b7d Fix pack_neon to use signed instruction. by Suharsh Sivakumar · 1 year, 10 months ago
  19. 1bf3b9c Fix ShiftLeft int16x8_t impl. by Suharsh Sivakumar · 1 year, 10 months ago
  20. 1ff8bdd fix the deps for the :fixedpoint target by Benoit Jacob · 1 year, 10 months ago
  21. 82acdad Do not register worker threads for profiling. by Benoit Jacob · 1 year, 10 months ago
  22. 1b76f37 make SSE 128bit specializations available also when AVX is available, as AVX per se only defines 256bit specializations, and user code may depend specifically on 128bit, as is the case in TFLite by Benoit Jacob · 1 year, 10 months ago
  23. 664f14a Per-channel output rescale and int8 input support for NEON. by Suharsh Sivakumar · 1 year, 10 months ago
  24. 2390b74 Fix documentation on OutputScalar in GemmWithOutputPipeline by Tetsuo Kiso · 1 year, 11 months ago
  25. 58825b1 MIPS: Improve StoreFinalOutput for 8x4 and 8x8 blocks of uint8. by Alexey Frunze · 2 years ago
  26. 5e071e0 Add a stage truncating int32 to uint8. by Alexey Frunze · 2 years ago
  27. 1949d3c MIPS: Implement an int8/LhsAlwaysNonzero kernel by Alexey Frunze · 2 years ago
  28. 1762ea1 Fix markdown format in kernel.md by Tetsuo Kiso · 2 years ago
  29. 2233f5e fix warnings about unused parameters by Paul Idstein · 2 years ago
  30. 55680df Fix typo by Tetsuo Kiso · 2 years ago
  31. 7aff33d Fix typo by Tetsuo Kiso · 2 years ago
  32. 42c5318 use GEMMLOWP_NOINLINE to avoid windows build break. by linkerzhang · 2 years ago
  33. 6a692f0 Add --fix option to allow fixing up wrong encodings in place, useful when mutating already-encoded code by Benoit Jacob · 2 years, 1 month ago
  34. a3df028 Fix typo in comments by Tetsuo Kiso · 2 years, 1 month ago
  35. 1490d29 add an experiment program to study how to interprete ARM PMU cache event counters. by Benoit Jacob · 2 years, 1 month ago
  36. 3fb5c17 MIPS: Implement ImplSaturatingRoundingMultiplyByPOT<,,-1> by Alexey Frunze · 2 years, 2 months ago
  37. 18e4cc0 support encoding sdot as well as udot by Benoit Jacob · 2 years, 2 months ago
  38. 75c19af MIPS: Implement RoundingDivideByPOT(). by Alexey Frunze · 2 years, 2 months ago
  39. 6db6c2a MIPS: Improve saturation. by Alexey Frunze · 2 years, 2 months ago
  40. 8416bab Fix #136 (test_blocking_counter) by Alexey Frunze · 2 years, 3 months ago
  41. 7d6d123 Import the new kernel contributed by ARM using the new dot-product by Benoit Jacob · 2 years, 3 months ago
  42. 3f87bab tiny fix to encode.py by Benoit Jacob · 2 years, 3 months ago
  43. f183d88 further improvements on encode.py by Benoit Jacob · 2 years, 3 months ago
  44. b1d3c06 Minor touchups on standalone programs by Benoit Jacob · 2 years, 3 months ago
  45. f3c2b90 Add a tool to encode udot instructions in asm code so we compile by Benoit Jacob · 2 years, 3 months ago
  46. 2c33e33 make udot kernels build on current toolchains by manually encoding the udot instruction, and add another udot kernel, "_narrow" by Benoit Jacob · 2 years, 4 months ago
  47. 7c5f9be Fix a use-after-free issue with the TLS data in the profiler, by Benoit Jacob · 2 years, 4 months ago
  48. 3aad165 Fix a regression of the profiler introduced in commit by Benoit Jacob · 2 years, 4 months ago
  49. 719139c fix comment by Benoit Jacob · 2 years, 4 months ago
  50. d877b64 Fix undefined behavior in the overflow case in ShiftLeft. by Benoit Jacob · 2 years, 4 months ago
  51. 4f09871 lint by Benoit Jacob · 2 years, 4 months ago
  52. 1080dba Fixing missing header. (Patch by Ben Vanik) by Benoit Jacob · 2 years, 4 months ago
  53. fa27313 Make gemmlowp multithreading correct: by Benoit Jacob · 2 years, 5 months ago
  54. dea287e Added math header to fixed point for round() declaration by Pete Warden · 2 years, 5 months ago
  55. a47fd48 Separated platform detection into a new header, so fixed point routines can use the right architecture after previous removal of common.h header by Pete Warden · 2 years, 5 months ago
  56. e88547b Removed fixed point dependency on wider gemmlowp headers, and added explicit casting for platforms where ints are 16-bits, to help with microcontroller portability by Pete Warden · 2 years, 5 months ago
  57. 07eb865 fix warnings about unused parameters by Bin Li · 2 years, 5 months ago
  58. 8724b85 [MSVC] Enable SSE4 when AVX is turned on by Loo Rong Jie · 2 years, 7 months ago
  59. cd685fe [Bazel] Don't set -lpthread for Windows by Loo Rong Jie · 2 years, 7 months ago
  60. 6c8f5d4 add missing inline keywords to fix linkage issues by bjacob · 2 years, 7 months ago
  61. 697ceea Add 'cc' and 'memory' to the clobbers list of NEON_64bit_GEMM_Int7Operands_AccumEightWithin16Bits and NEON_64bit_GEMM_Int425Operands kernels since they are being modified. by Marie White · 2 years, 7 months ago
  62. 38ebac7 Add malloc.h for Windows by Yong Tang · 2 years, 9 months ago
  63. fc0ebb1 Fix compilation error in Windows by Yong Tang · 2 years, 9 months ago
  64. 85179c4 add Richart Winterton to CONTRIBUTORS, clang-format and convert to unix line endings by Benoit Jacob · 2 years, 9 months ago
  65. d74760e Avx2optimizations (#122) by rrwinterton · 2 years, 9 months ago
  66. ad03972 Add profiling support to benchmark_all_sizes by Benoit Jacob · 2 years, 10 months ago
  67. e08d0f6 Merge pull request #132 from jalexstark/fixedpoint-constexpr by bjacob · 2 years, 10 months ago
  68. c033219 Adjust use of static const / static constexpr in fixedpoint. by Alex Stark · 2 years, 10 months ago
  69. 87510e5 Merge pull request #131 from MIPS/msa_width_major by bjacob · 2 years, 10 months ago
  70. c318e3f MIPS: Improve the kernel by switching to width-major. by Alexey Frunze · 2 years, 10 months ago
  71. f59a96b Merge pull request #130 from ajtulloch/gemmlowp-wait-for-variable-change-spurious-wakeup by bjacob · 2 years, 10 months ago
  72. 1a4582c multi_thread_gemm.h::WaitForVariableChange: Handle spurious wakeups correctly by Andrew Tulloch · 2 years, 10 months ago
  73. d6fe01d Merge pull request #129 from ajtulloch/gemmlowp-fix-undefined-behaviour-in-left-shift by bjacob · 2 years, 10 months ago
  74. ea3a4bf UBSAN: Fix fixedpoint.h::ShiftLeft to not invoke undefined behaviour by Andrew Tulloch · 2 years, 10 months ago
  75. 5b40e38 Merge pull request #128 from mariecwhite/sevenbit by bjacob · 2 years, 11 months ago
  76. 3463777 Add mariewhite@ to list of contributors by Marie White · 2 years, 11 months ago
  77. abc0fdb Add 4-bit GEMM asm and intrinsics kernel by Marie White · 2 years, 11 months ago
  78. 73184f4 Add 7-bit GEMM asm and intrinsics kernel by Marie White · 2 years, 11 months ago
  79. d41349d Add kernel operand ranges to standalone benchmark by Marie White · 2 years, 11 months ago
  80. ecae4d1 Merge pull request #125 from MIPS/mips_cast16 by bjacob · 2 years, 11 months ago
  81. c9bd438 MIPS: Add output stage casting to int16 by Alexey Frunze · 2 years, 11 months ago
  82. 7c7c744 Merge pull request #124 from bjacob/updates-for-fast-lstm by bjacob · 3 years ago
  83. 416992a add OutputStageScaleInt32ByFixedPointAndExponent by Benoit Jacob · 3 years ago
  84. 9d7ae82 Add output stage casting to int16 by Benoit Jacob · 3 years ago
  85. 15a66ad rename: s/OutputStageQuantizeDownInt32ToUint8ScaleByFixedPoint/OutputStageQuantizeDownInt32ByFixedPoint/g - and keep the old name supported for compatibility by Benoit Jacob · 3 years ago
  86. d4d1e29 compilation fixes by Benoit Jacob · 3 years ago
  87. c45f8f5 Merge pull request #123 from MIPS/mips_fixedpoint16bit by bjacob · 3 years ago
  88. 95af7df fix compilation by Benoit Jacob · 3 years ago
  89. 80c9ee5 MIPS: Add 16-bit support to gemmlowp/fixedpoint & math functions by Alexey Frunze · 3 years ago
  90. f4763b6 compilation fixes by Benoit Jacob · 3 years ago
  91. 304ac74 fix trailing whitespace in script by Benoit Jacob · 3 years ago
  92. d77359f Merge branch 'master' of github.com:google/gemmlowp by Benoit Jacob · 3 years ago
  93. 6da0669 clang-format by Benoit Jacob · 3 years ago
  94. 4c5cbde Merge pull request #113 from achille/master by bjacob · 3 years ago
  95. 3977925 fix typo by Benoit Jacob · 3 years ago
  96. 8711341 Merge pull request #121 from bjacob/fixedpoint16bit by bjacob · 3 years ago
  97. 1a292d2 Merge branch 'master' into fixedpoint16bit by bjacob · 3 years ago
  98. 76de7f7 Add 16-bit support to gemmlowp/fixedpoint & math functions by Benoit Jacob · 3 years ago
  99. c915905 Merge pull request #120 from MIPS/mips_msa by bjacob · 3 years ago
  100. ca372a1 MIPS: Add support for MIPS MSA SIMD. by Alexey Frunze · 3 years ago