Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / ruy / 3a48f4ee63c98b96fc776a42698fcafc02bf5a75 / kernel.h
  1. 6593992 Ruy: Modify guards to use X86 platform in some places. by Alex Stark · 5 years ago
  2. 8221a67 Ruy: Improve includes. by Alex Stark · 5 years ago
  3. ac7834c Ruy: Reorganize pack and kernel headers. by Alex Stark · 5 years ago
  4. 7f75028 Ruy: Introduce x86 (AVX-512) code. by Alex Stark · 5 years ago
  5. a4aa186 Introduce a SidePair concept allowing us to rewrite much internal by Benoit Jacob · 5 years ago
  6. 78ea2fe Fix performance regression (b/137615815) introduced by new platform by Benoit Jacob · 5 years ago
  7. 70d6007 Ruy - Float Kernel in ARM32 asm. by T.J. Alumbaugh · 5 years ago
  8. 6552ad2 Ruy - Float Kernel in ARM32 asm. by Ruy Contributors · 5 years ago
  9. 12e5fc7 Ruy - Float Kernel in ARM32 asm. by T.J. Alumbaugh · 5 years ago
  10. 8af0706 Ruy - Float Kernel in ARM32 asm. by T.J. Alumbaugh · 5 years ago
  11. 75f72e9 Ruy - Float Kernel in ARM32 asm. by T.J. Alumbaugh · 5 years ago
  12. 86e430c Make the kStandardCpp kernel layout and the cache-friendly traversal by Benoit Jacob · 5 years ago
  13. e081f4a Enable ruy optimizations in a less error-prone way. by Sean Silva · 5 years ago
  14. bf2e9f0 C++ kernels don't need relaxed_atomic_store. This is needed only when multiple threads might be concurrently writing the same value to the same memory location. This situation occurs in packing code, but not in kernels, as each tile of the destination matrix is computed only once. by Benoit Jacob · 5 years ago
  15. fc2c2a0 Skip over left-shift instructions in the common case where they by Benoit Jacob · 6 years ago
  16. 6532f45 Clean up Ruy's internal dependency structure. by Ruy Contributors · 6 years ago
  17. 30a287a Clean up Ruy's internal dependency structure. by Sean Silva · 6 years ago
  18. b214d15 Add ruy dst==int32 support for in order cores for dot prod kernel. by Renjie Liu · 6 years ago
  19. b254f9e Clarify and enforce the contract around kernels and destination matrix dimensions that aren't a multiple of kernel dimensions. by Benoit Jacob · 6 years ago
  20. 9b97ec4 Add ruy dst==int32 support for out order cores for dot prod kernel. by Renjie Liu · 6 years ago
  21. 74595d0 Better fix for b/132351025: by Benoit Jacob · 6 years ago
  22. 8dc4be3 Fix the kernel layout description for the float NEON kernels. by Benoit Jacob · 6 years ago
  23. 99fd663 Add ruy dst==int32 support for in order cores. by Renjie Liu · 6 years ago
  24. fc2ed74 ruy: support dst == int32 for neon out of order case. by Renjie Liu · 6 years ago
  25. b6f4a49 ruy: support dst == int32, currently only kStandardCpp path. by Renjie Liu · 6 years ago
  26. a8907c5 PR #28392: [aarch64] make TensorFlow build for aarch64 linux natively with bazel by Ruy Contributors · 6 years ago
  27. 50fa2c4 Detemplatize TrMul and introduce type-erased TrMulParams. by Sean Silva · 6 years ago
  28. d9646e5 Fix a couple of ruy compile issues found by TensorFlow continuous integration. by Benoit Jacob · 6 years ago
  29. 13689fe Opensource Ruy by Jared Duke · 6 years ago
  30. c8dc0cd Internal change by Benoit Jacob · 6 years ago
  31. c43c0d3 Internal change. by Sean Silva · 6 years ago
  32. 1514393 Internal change. by Sean Silva · 6 years ago
  33. a048fe8 Internal Change by Benoit Jacob · 6 years ago
  34. a0ba3ac internal change by Benoit Jacob · 6 years ago