Gitiles
Code Review Sign In
android-review.linaro.org / platform / external / FP16 / refs/heads/main
commitb9cd83b97d074e567bd6acb2912802cedcc9a216[log] [tgz]
authorXin Li <delphij@google.com>Wed Mar 06 09:29:51 2024 -0800
committerXin Li <delphij@google.com>Wed Mar 06 09:29:51 2024 -0800
treecb376e49da4efb0db30089e589c64672c41dd208
parent5fbb3524547540b2d023b3f7bd77ad3ec0b0d546 [diff]
parent58fe764feb9aee390c7efb1bbd184388dacfa6d6 [diff]
Merge Android 14 QPR2 to AOSP main

Bug: 319669529
Merged-In: Ib37391d6c9b63a1512757ec8f6af502c8236837e
Change-Id: Icb75cde2bb0cb517d47e33c9b060394163bcc61d
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.