Google Git
Sign in
android / platform / external / FP16 / refs/heads/android11-d1-release
commit4039a7ffa4bbee623e1e616efef9a387b2a8b47d[log] [tgz]
authorAutomerger Merge Worker <[email protected]>Wed Feb 12 20:53:13 2020 +0000
committerAutomerger Merge Worker <[email protected]>Wed Feb 12 20:53:13 2020 +0000
treea426169d62e9a749341225ef30aa64fea2af39f2
parentfeb62fe3e061db6b91b5e740e475b8e88f0f2af7 [diff]
parenta27f152ea3ebfcc257a38f956a5713a46955feb9 [diff]
Add OWNERS am: 212fae0646 am: 71f087c388 am: a27f152ea3

Change-Id: Ic6b68b0a083409e35338034165bfc27e92a4d60b
tree: a426169d62e9a749341225ef30aa64fea2af39f2
  1. bench/
  2. cmake/
  3. include/
  4. test/
  5. NOTICE ⇨ LICENSE
  6. .gitignore
  7. Android.bp
  8. CMakeLists.txt
  9. configure.py
  10. confu.yaml
  11. LICENSE
  12. METADATA
  13. MODULE_LICENSE_MIT
  14. OWNERS
  15. README.md
  16. 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.