Google Git
Sign in
eigen / mirror / 24db460503f9cfda0ee0b42a898f2c5caf06f31b / Eigen / src / Core / arch / NEON
  1. c593e9e Fix typos by Mike Taves · 8 months ago
  2. 6331da9 fixing a lot of typos by Frédéric Chapoton · 8 months ago
  3. fb95e90 Add truncation op by Charles Schlosser · 11 months ago
  4. 888fca0 Simd sincos double by Damiano Franzò · 12 months ago
  5. be06c9a Implement float pexp_complex by Damiano Franzò · 1 year, 2 months ago
  6. 7fd7a3f Implement plog_complex by Damiano Franzò · 1 year, 4 months ago
  7. a73970a Fix arm32 issues. by Antonio Sánchez · 1 year, 3 months ago
  8. f38e16c Apply clang-format by Tobias Wood · 1 year, 4 months ago
  9. dd8c71e Fix typecasting for arm32 by Antonio Sánchez · 1 year, 5 months ago
  10. 66b9f4e Fix (u)int64_t->float conversion on arm by Pavel Labath · 1 year, 5 months ago
  11. 81b4806 Fix arm32 float division and related bugs by Charles Schlosser · 1 year, 7 months ago
  12. 6e4d5d4 Add IWYU private pragmas to internal headers. by Antonio Sánchez · 1 year, 8 months ago
  13. 7465b76 Disable FP16 arithmetic for arm32. by Antonio Sánchez · 1 year, 10 months ago
  14. 59b3ef5 Partially Vectorize Cast by Charles Schlosser · 1 year, 10 months ago
  15. df1049d Small packet math cleanup. by Rasmus Munk Larsen · 2 years ago
  16. 2c8011c Fix arm builds. by Antonio Sánchez · 2 years, 1 month ago
  17. 21c49e8 Delete mystery character from Eigen/src/Core/arch/NEON/MathFunctions.h by Rasmus Munk Larsen · 2 years, 1 month ago
  18. d6235d7 Clean up generic packetmath specializations for various backends with the help of a macro. by Rasmus Munk Larsen · 2 years, 1 month ago
  19. ce62177 Vectorize atanh & add a missing definition and unit test for atan. by Rasmus Munk Larsen · 2 years, 2 months ago
  20. 3842699 More NEON packetmath fixes. by Antonio Sánchez · 2 years, 2 months ago
  21. 2dfbf1b Fix NEON make_packet2f. by Antonio Sánchez · 2 years, 2 months ago
  22. 0a5392d Fix MSVC arm build. by Antonio Sánchez · 2 years, 2 months ago
  23. d70b486 issue #2581: review and cleanup of compiler version checks by Sean McBride · 2 years, 3 months ago
  24. c54785b Fix error: unused parameter 'tmp' [-Werror,-Wunused-parameter] on clang/32-bit arm by Martin Burchell · 2 years, 3 months ago
  25. 311cc0f Enable NEON pcmp, plset, and complex psqrt by Arthur · 2 years, 4 months ago
  26. bb6675c Fix incorrect NEON native fp16 multiplication. by Antonio Sánchez · 2 years, 4 months ago
  27. c4fb6af Enable NEON pabs for unsigned int types by Arthur Feeney · 2 years, 4 months ago
  28. d194167 Fix the bug using neon instruction fmla for data type half by Lianhuang Li · 2 years, 4 months ago
  29. 9b6d624 fix neon by Charles Schlosser · 2 years, 5 months ago
  30. 7e398e9 Add missing return keyword in psignbit for NEON. by Rasmus Munk Larsen · 2 years, 5 months ago
  31. 82b152d Add signbit function by Charles Schlosser · 2 years, 5 months ago
  32. c475228 Vectorize atan() for double. by Rasmus Munk Larsen · 2 years, 6 months ago
  33. ed8cda3 Move EIGEN_NEON_GEBP_NR macro to the right place in GeneralBlockPanelKernel.h by Rasmus Munk Larsen · 2 years, 7 months ago
  34. e2ea866 Add a macro to set the nr trait in the BEBP kernel for NEON. by Rasmus Munk Larsen · 2 years, 7 months ago
  35. 2329963 Use 3px8/2px8/1px8/1x8 gebp_kernel on arm64-neon by Lianhuang Li · 2 years, 7 months ago
  36. bd393e1 Vectorize acos, asin, and atan for float. by Rasmus Munk Larsen · 2 years, 7 months ago
  37. 9b9496a Revert "Add AVX512 optimizations for matrix multiply" by Antonio Sánchez · 2 years, 11 months ago
  38. 25db0b4 Add AVX512 optimizations for matrix multiply by aaraujom · 2 years, 11 months ago
  39. f3ba220 Remove EIGEN_EMPTY_STRUCT_CTOR by Tobias Schlüter · 3 years ago
  40. f1b9692 Removed EIGEN_UNUSED decorations from many functions that are in fact used by Sean McBride · 3 years, 1 month ago
  41. 28e008b Fix sqrt/rsqrt for NEON. by Antonio Sánchez · 3 years, 2 months ago
  42. 6b60bd6 Fix 32-bit arm int issue. by Antonio Sánchez · 3 years, 2 months ago
  43. 8d81a23 Reduce usage of reserved names by Kolja Brix · 3 years, 3 months ago
  44. c4b1dd2 Add support for Cray, Fujitsu, and Intel ICX compilers by Matthias Möller · 3 years, 3 months ago
  45. cc11e24 Some further cleanup by Erik Schultheis · 3 years, 4 months ago
  46. 6bb6a6b Vectorize fp16 tanh and logistic functions on Neon by Alex Druinsky · 3 years, 5 months ago
  47. afa616b Fix some typos found by Kolja Brix · 3 years, 7 months ago
  48. d7d0bf8 Issue an error in case of direct inclusion of internal headers. by Rasmus Munk Larsen · 3 years, 7 months ago
  49. ff07a8a GCC 4.8 arm EIGEN_OPTIMIZATION_BARRIER fix (#2315). by Antonio Sanchez · 3 years, 7 months ago
  50. ab28419 optimize predux if architecture is aarch64 by Han-Kuan Chen · 3 years, 8 months ago
  51. 66ca41b Add support for vectorizing logical comparisons. by derekjchow · 3 years, 9 months ago
  52. 9312a5b Implement a generic vectorized version of Smith's algorithms for complex division. by Rasmus Munk Larsen · 3 years, 9 months ago
  53. c81da59 Correct declarations for aarch64-pc-windows-msvc by 大河メタル · 3 years, 9 months ago
  54. bffd267 Small cleanup: Get rid of the macros EIGEN_HAS_SINGLE_INSTRUCTION_CJMADD and CJMADD, which were effectively unused, apart from on x86, where the change results in identically performing code. by Rasmus Munk Larsen · 3 years, 10 months ago
  55. 52a5f98 Get rid of code duplication for conj_helper. For packets where LhsType=RhsType a single generic implementation suffices. For scalars, the generic implementation of pconj automatically forwards to numext::conj, so much of the existing specialization can be avoided. For mixed types we still need specializations. by Rasmus Munk Larsen · 3 years, 10 months ago
  56. fc87e2c Use bit_cast to create -0.0 for floating point types to avoid compiler optimization changing sign with --ffast-math enabled. by Rasmus Munk Larsen · 3 years, 10 months ago
  57. dba753a Add missing NEON ptranspose implementations. by Antonio Sanchez · 3 years, 11 months ago
  58. 722ca0b Revert addition of unused `paddsub<Packet2cf>`. This fixes #2242 by Christoph Hertzberg · 4 years ago
  59. 172db7b Add missing pcmp_lt_or_nan for NEON Packet4bf. by Antonio Sanchez · 4 years ago
  60. 4811e81 Remove yet another comma at end of enum by David Tellenbach · 4 years, 1 month ago
  61. 82d61af Fix rint SSE/NEON again, using optimization barrier. by Antonio Sanchez · 4 years, 1 month ago
  62. 9a66397 Revert "Fix rint for SSE/NEON." by Antonio Sánchez · 4 years, 1 month ago
  63. e72dfeb Fix rint for SSE/NEON. by Antonio Sanchez · 4 years, 1 month ago
  64. 1e0c7d4 Add print for SSE/NEON, use NEON rounding intrinsics if available. by Antonio Sanchez · 4 years, 1 month ago
  65. 29ebd84 Fix NEON sqrt for 32-bit, add prsqrt. by Antonio Sanchez · 4 years, 1 month ago
  66. e19829c Fix floor/ceil for NEON fp16. by Antonio Sanchez · 4 years, 1 month ago
  67. 5529db7 Fix SSE/NEON pfloor/pceil for saturated values. by Antonio Sanchez · 4 years, 1 month ago
  68. 6cf0ab5 Disable fast psqrt for NEON. by Antonio Sanchez · 4 years, 1 month ago
  69. 7ff0b7a Updated pfrexp implementation. by Antonio Sanchez · 4 years, 2 months ago
  70. f702792 missing method in packetmath.h void ptranspose(PacketBlock<Packet16uc, 4>& kernel) by Ashutosh Sharma · 4 years, 2 months ago
  71. 90ee821 Use vrsqrts for rsqrt Newton iterations. by Antonio Sanchez · 4 years, 2 months ago
  72. 7eb07da loop less ptranspose by Ashutosh Sharma · 4 years, 2 months ago
  73. f85038b Fix excessive GEBP register spilling for 32-bit NEON. by Antonio Sanchez · 4 years, 2 months ago
  74. b2126fd Fix pfrexp/pldexp for half. by Antonio Sanchez · 4 years, 3 months ago
  75. 38ae535 1)provide a better generic paddsub op implementation by Guoqiang QI · 4 years, 3 months ago
  76. 0575410 * Add iterative psqrt<double> for AVX and SSE when FMA is available. This provides a ~10% speedup. by Rasmus Munk Larsen · 4 years, 4 months ago
  77. 6cee8d3 Add an additional step of Newton-Raphson for `psqrt<double>` on Arm, which otherwise has an error of ~1000 ulps. by Rasmus Munk Larsen · 4 years, 4 months ago
  78. 55967f8 Fix NEON pmax<PropagateNumbers,Packet4bf>. by Antonio Sanchez · 4 years, 4 months ago
  79. c7eb3a7 Don't guard psqrt for std::complex<float> with EIGEN_ARCH_ARM64 by David Tellenbach · 4 years, 4 months ago
  80. bccf055 Add Armv8 guard on PropagateNumbers implementation. by Everton Constantino · 4 years, 4 months ago
  81. 00be0a7 Fix vectorization of complex sqrt on NEON by David Tellenbach · 4 years, 4 months ago
  82. 8eb461a Remove comma at end of enumerator list in NEON PacketMath by David Tellenbach · 4 years, 4 months ago
  83. baf9d76 - Enabling PropagateNaN and PropagateNumbers for NEON. by Everton Constantino · 4 years, 5 months ago
  84. e2f2146 Special function implementations for half/bfloat16 packets. by Antonio Sanchez · 4 years, 4 months ago
  85. 70fbcf8 Fix typo in `F32MaskToBf16Mask`. by Antonio Sanchez · 4 years, 4 months ago
  86. 2627e2f Fix neon cmp* functions for bf16. by Antonio Sanchez · 4 years, 4 months ago
  87. fd1dcb6 Fixes duplicate symbol when building blas by Antonio Sanchez · 4 years, 5 months ago
  88. 11e4056 Re-enable Arm Neon Eigen::half packets of size 8 by David Tellenbach · 4 years, 5 months ago
  89. e9b55c4 Avoid promotion of Arm __fp16 to float in Neon PacketMath by David Tellenbach · 4 years, 5 months ago
  90. 394f564 Unify Inverse_SSE.h and Inverse_NEON.h into a single generic implementation using PacketMath. by Guoqiang QI · 4 years, 5 months ago
  91. 8324e5e Fix typo in NEON/PacketMath.h by guoqiangqi · 4 years, 5 months ago
  92. e265f7e Add support for Armv8.2-a __fp16 by David Tellenbach · 4 years, 5 months ago
  93. 4e4d3f3 Clean up packetmath tests and fix various bugs to make bfloat16 pass (almost) all packetmath tests with SSE, AVX, and AVX512. by Rasmus Munk Larsen · 4 years, 6 months ago
  94. 9aad16b Fix undefined reference to pset1frombits bug on different platforms by guoqiangqi · 4 years, 7 months ago
  95. 14022f5 Fix more mildly embarrassing typos in ARM intrinsics in PacketMath.h. by Rasmus Munk Larsen · 4 years, 7 months ago
  96. a5b2269 Fix typo in PacketMath.h by Rasmus Munk Larsen · 4 years, 7 months ago
  97. 3af744b Add missing packet op pcmp_lt_or_nan for Packet2d on ARM. by Rasmus Munk Larsen · 4 years, 7 months ago
  98. 880fa43 Add support for CastXML on ARM aarch64 by Brad King · 4 years, 7 months ago
  99. cc0c38a Remove old Clang compiler bug work-arounds. The two LLVM bugs referenced in the comments here have long been fixed. The workarounds were now detrimental because (1) they prevented using fused mul-add on Clang/ARM32 and (2) the unnecessary 'volatile' in 'asm volatile' prevented legitimate reordering by the compiler. by Benoit Jacob · 4 years, 7 months ago
  100. 3012e75 Add plog ops support packet2d for NEON by Guoqiang QI · 4 years, 7 months ago