Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / FP16 / 6b3341613efd596c41858f8873554062c4b05361
commit6b3341613efd596c41858f8873554062c4b05361[log] [tgz]
authorLinux Build Service Account <lnxbuild@localhost>Tue Apr 06 17:23:00 2021 -0700
committerLinux Build Service Account <lnxbuild@localhost>Tue Apr 06 17:23:00 2021 -0700
treebf795792c727b7d3d47af92580a2ac7157e72cd1
parent9b4c00f019d40cca68e728c4664d0c75087b63e5 [diff]
parente7ee08b47783f80e2eb76e944beb4ad5829157f6 [diff]
Merge e7ee08b47783f80e2eb76e944beb4ad5829157f6 on remote branch

Change-Id: I54c65828a24aa16a65d930f3220c403320f9aea2
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.