Google Git
Sign in
android / platform / external / FP16 / refs/heads/android-t-preview-1
commit3e17c7a0be5a4afbdb330d583d71a14f6e7dee1c[log] [tgz]
authorIan Hua <[email protected]>Tue Jan 11 11:46:08 2022 +0000
committerIan Hua <[email protected]>Tue Jan 11 11:46:08 2022 +0000
tree902ebbb006074e18653102fdd5be27b04e0d0a88
parente22b19eb0cfaa40394d1d70eaa84cdff6a1ef67b [diff]
Update OWNERS for external/FP16.

Test: n/a
Change-Id: I6c5048aac5478d1cca010d8a56f5d422897e3817
1 file changed
tree: 902ebbb006074e18653102fdd5be27b04e0d0a88
  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.