Google Git
Sign in
android / platform / external / FP16 / refs/heads/android12-mainline-art-release
commitccaaa4a74e6731571976b31afa30a336e61b8364[log] [tgz]
authorAndroid Build Coastguard Worker <[email protected]>Wed Nov 17 15:59:24 2021 +0000
committerAndroid Build Coastguard Worker <[email protected]>Wed Nov 17 15:59:24 2021 +0000
treedf61340909aacfee61ba9b551e63df7ae3d3eceb
parent692d39568e39535f3772b5ffaa9be64e605ef24a [diff]
parent2cc626c378b7a58cc8ca25d3686fa963e765f6ee [diff]
Snap for 7924065 from 2cc626c378b7a58cc8ca25d3686fa963e765f6ee to mainline-art-release

Change-Id: Id8acf96f8e5bf4eb602ece7a812be5dd99258e0b
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.