Google Git
Sign in
android / platform / external / FP16 / refs/heads/android12-mainline-permission-release
commit6f8049e14175fe3a1b642e00de73be18bc826329[log] [tgz]
authorAndroid Build Coastguard Worker <[email protected]>Tue Nov 16 06:36:52 2021 +0000
committerAndroid Build Coastguard Worker <[email protected]>Tue Nov 16 06:36:52 2021 +0000
treedf61340909aacfee61ba9b551e63df7ae3d3eceb
parent46ea4d748c3e9321fad3db18104526e9f5dc7b8f [diff]
parent2cc626c378b7a58cc8ca25d3686fa963e765f6ee [diff]
Snap for 7919077 from 2cc626c378b7a58cc8ca25d3686fa963e765f6ee to mainline-permission-release

Change-Id: I58c888f433f0c0d7124de7e67e9806bf42aa928c
tree: df61340909aacfee61ba9b551e63df7ae3d3eceb
  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.