Google Git
Sign in
android / platform / external / FP16 / refs/heads/android14-mainline-art-release
commit0124030a2de5798c9f94ea2da17afab94b580121[log] [tgz]
authorAndroid Build Coastguard Worker <[email protected]>Mon Aug 07 10:36:14 2023 +0000
committerAndroid Build Coastguard Worker <[email protected]>Mon Aug 07 10:36:14 2023 +0000
tree5861de0ba4651a75bafe7097b5665fed86eb6903
parent44f1a58223f7b73786824b7af72d26233306c931 [diff]
parent3bd198b649a0cb30bd01f2daf5da7a4d00cfffb2 [diff]
Snap for 10620181 from 3bd198b649a0cb30bd01f2daf5da7a4d00cfffb2 to mainline-art-release

Change-Id: Ib8a4204775792793401fd64e01c5c578ef173d62
tree: 5861de0ba4651a75bafe7097b5665fed86eb6903
  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.