Google Git
Sign in
android / platform / external / FP16 / 174e833f96e5c123bde20fcc43062bec66942407
commit174e833f96e5c123bde20fcc43062bec66942407[log] [tgz]
authorAndroid Build Coastguard Worker <[email protected]>Tue Aug 17 03:02:39 2021 +0000
committerAndroid Build Coastguard Worker <[email protected]>Tue Aug 17 03:02:39 2021 +0000
treedf61340909aacfee61ba9b551e63df7ae3d3eceb
parent0b407d4ff750d0a2e7e855edcaa22c8a0e2d4c3b [diff]
parentba83e6721903bc6db3e1c107057ea5bdd48c61b3 [diff]
Snap for 7647598 from ba83e6721903bc6db3e1c107057ea5bdd48c61b3 to tm-release

Change-Id: I83b9b4bbbf4083f83c3bd07cf036d7c2bbde2591
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.