Google Git
Sign in
android / platform / external / FP16 / refs/heads/android12-s3-release
commitbe25eab55985132276bf89e33c303839d0576413[log] [tgz]
authorandroid-build-team Robot <[email protected]>Sat Feb 20 02:03:02 2021 +0000
committerandroid-build-team Robot <[email protected]>Sat Feb 20 02:03:02 2021 +0000
treebf795792c727b7d3d47af92580a2ac7157e72cd1
parent821f0883646b7a96bc199300f9583791dc39e3b9 [diff]
parent015fd0f2a06d43d5274e6d9a9675f7d575df0316 [diff]
Snap for 7158291 from 015fd0f2a06d43d5274e6d9a9675f7d575df0316 to sc-release

Change-Id: If16d4ab3483cc9d5baf6579a97486e541f280390
tree: bf795792c727b7d3d47af92580a2ac7157e72cd1
  1. bench/
  2. cmake/
  3. include/
  4. test/
  5. .gitignore
  6. Android.bp
  7. CMakeLists.txt
  8. configure.py
  9. confu.yaml
  10. LICENSE
  11. METADATA
  12. MODULE_LICENSE_MIT
  13. OWNERS
  14. README.md
  15. TEST_MAPPING
README.md

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.