Bug: 148976166

Clone this repo:
  1. 3e17c7a Update OWNERS for external/FP16. by Ian Hua · 11 days ago master
  2. e22b19e Add host_supported, min_sdk_version and apex_available configs to Android.bp for FP16. am: 80907cb9be am: cbadf74d7b am: 919d1bd8ea by Ian Hua · 5 months ago
  3. 919d1bd Add host_supported, min_sdk_version and apex_available configs to Android.bp for FP16. am: 80907cb9be am: cbadf74d7b by Ian Hua · 5 months ago
  4. cbadf74 Add host_supported, min_sdk_version and apex_available configs to Android.bp for FP16. am: 80907cb9be by Ian Hua · 5 months ago android-games-sdk-release android-s-v2-preview-1 android-s-v2-beta-2 android-s-v2-preview-2
  5. 80907cb Add host_supported, min_sdk_version and apex_available configs to Android.bp for FP16. by Ian Hua · 5 months ago android-s-beta-5

FP16

Header-only library for conversion to/from half-precision floating point formats

Features

  • Supports IEEE and ARM alternative half-precision floating-point format
    • Property converts infinities and NaNs
    • Properly converts denormal numbers, even on systems without denormal support
  • Header-only library, no installation or build required
  • Compatible with C99 and C++11
  • Fully covered with unit tests and microbenchmarks

Acknowledgements

HPC Garage logo Georgia Tech College of Computing logo

The library is developed by Marat Dukhan of Georgia Tech. FP16 is a research project at Richard Vuduc's HPC Garage lab in the Georgia Institute of Technology, College of Computing, School of Computational Science and Engineering.

This material is based upon work supported by the U.S. National Science Foundation (NSF) Award Number 1339745. Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect those of NSF.