Clone this repo:
  1. e6e24bf Update copyright year to 2018. by Paul Robinson · 2 days ago master
  2. 9662704 r600/fmin: Flush denormals before calling builtin. by Jan Vesely · 12 days ago
  3. aa96171 r600/fmax: Flush denormals before calling builtin. by Jan Vesely · 12 days ago
  4. 8f66cbc math/fma: Add fp32 software implementation by Jan Vesely · 12 days ago
  5. a2118d5 Add initial support for half precision builtins by Jan Vesely · 5 weeks ago
  6. bac1578 rootn: Use denormal path only by Jan Vesely · 5 weeks ago
  7. 569bb0e remquo: Flush denormals if not supported by Jan Vesely · 7 weeks ago
  8. bebb1c2 remquo: Port from amd builtins by Jan Vesely · 7 weeks ago
  9. a787281 math: Add helper function to flush denormals if not supported. by Jan Vesely · 7 weeks ago
  10. cb89446 clc_sqrt: Reuse unary_decl.inc by Jan Vesely · 7 weeks ago
  11. e4c03a8 relational/select: Condition types for half are short/ushort, not char/uchar by Jan Vesely · 8 weeks ago
  12. 20c7963 log10: Use sw implementation from amd builtins by Jan Vesely · 8 weeks ago
  13. 942243a powr: Use denormal path only by Jan Vesely · 9 weeks ago
  14. b7ccb44 pown: Use denormal path only by Jan Vesely · 9 weeks ago
  15. 0af7f47 pow: Use denormal path only by Jan Vesely · 9 weeks ago
  16. 24ce5c5 amdgcn/fmin: Fix typos that reduced precision by Jan Vesely · 9 weeks ago
  17. 3a7cab5 exp10: Port from amd builtins by Jan Vesely · 9 weeks ago
  18. 2a1e8aa hypot: Port from amd builtins by Jan Vesely · 2 months ago
  19. f66afab select: simplify implementation and fix fp16 by Jan Vesely · 2 months ago
  20. d70af3b fmod: Port from amd_builtins by Jan Vesely · 2 months ago