Google Git
Sign in
android / platform / external / FP16 / refs/heads/android14-mainline-resolv-release
commit649b97da888beefb09e79041fea5d5b297584c88[log] [tgz]
authorAndroid Build Coastguard Worker <[email protected]>Wed Aug 30 19:04:54 2023 +0000
committerAndroid Build Coastguard Worker <[email protected]>Wed Aug 30 19:04:54 2023 +0000
tree5861de0ba4651a75bafe7097b5665fed86eb6903
parentcfa1ada34c94a748928bd368c48316e6c9b818f6 [diff]
parent3bd198b649a0cb30bd01f2daf5da7a4d00cfffb2 [diff]
Snap for 10737070 from 3bd198b649a0cb30bd01f2daf5da7a4d00cfffb2 to mainline-resolv-release

Change-Id: I1e5d6ca4dee9d796c4aff87ea968eeadc3bb6af4
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.