Google Git
Sign in
android / platform / external / FP16 / refs/heads/android14-qpr3-s11-release
commit1ab38e28378728b705bd20e7be271270ad5b28ee[log] [tgz]
authorAndroid Build Coastguard Worker <[email protected]>Sat Feb 24 02:06:05 2024 +0000
committerAndroid Build Coastguard Worker <[email protected]>Sat Feb 24 02:06:05 2024 +0000
treecb376e49da4efb0db30089e589c64672c41dd208
parent96afbf8987cd5aeeddbd9fe8eaaf558228ea4d2e [diff]
parentef03046e79727f21547dc22cb9a3ae6e78e1a917 [diff]
Snap for 11491255 from ef03046e79727f21547dc22cb9a3ae6e78e1a917 to 24Q2-release

Change-Id: I19daf490d232340f7b74c159c72dfde7d66b4071
tree: cb376e49da4efb0db30089e589c64672c41dd208
  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.