Google Git
Sign in
android / platform / external / FP16 / refs/heads/android13-mainline-scheduling-release
commit464665ad0a3836a3e3c97251f5b740c4f79da0be[log] [tgz]
authorAndroid Build Coastguard Worker <[email protected]>Fri May 20 19:45:28 2022 +0000
committerAndroid Build Coastguard Worker <[email protected]>Fri May 20 19:45:28 2022 +0000
tree0592ff9f13f19172fb00472d1f317c7a7fa31724
parentb4bb3c3b96688fbc81df2dad4854524ac2f5513e [diff]
parentbb37826c27934afb93fd799c5c908d1660f3b473 [diff]
Snap for 8622994 from bb37826c27934afb93fd799c5c908d1660f3b473 to mainline-scheduling-release

Change-Id: Ib5078b7d12f0bb032e10a67f2aacf1a531e6d1ef
tree: 0592ff9f13f19172fb00472d1f317c7a7fa31724
  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.