Google Git
Sign in
android / platform / external / FP16 / refs/heads/android-s-beta-4
commita738f586f15db2566055279ac2de08559d40b8d1[log] [tgz]
authorStephen Hines <[email protected]>Thu Feb 18 22:25:25 2021 +0000
committerAutomerger Merge Worker <[email protected]>Thu Feb 18 22:25:25 2021 +0000
treebf795792c727b7d3d47af92580a2ac7157e72cd1
parentde7ecd98d97b69bdade712e289e39cc47fbfd445 [diff]
parentc2b391b0768dbf3ddf46ab0472e8f79afdf1eba6 [diff]
Fix undefined behavior (UB) from calling builtin_clz(0) am: c2b391b076

Original change: https://android-review.googlesource.com/c/platform/external/FP16/+/1592859

MUST ONLY BE SUBMITTED BY AUTOMERGER

Change-Id: I60fa6df65faf5bc160772bf1727264b3bdc514fe
tree: bf795792c727b7d3d47af92580a2ac7157e72cd1
  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.