Google Git
Sign in
android / platform / external / FP16 / d83872f97c2a3a1175f4215b44151a5acd3bd1bd
commitd83872f97c2a3a1175f4215b44151a5acd3bd1bd[log] [tgz]
authorAutomerger Merge Worker <android-build-automerger-merge-worker@system.gserviceaccount.com>Wed Feb 12 02:32:00 2020 +0000
committerAutomerger Merge Worker <android-build-automerger-merge-worker@system.gserviceaccount.com>Wed Feb 12 02:32:00 2020 +0000
treef5dd6f49d5a89ff3ebf8d17554810ea01542504c
parent08a172d6da973d8750165c40316df6fade8ab364 [diff]
parent8456625b65287d70f101e95bb194e90fe17e465e [diff]
Merge remote-tracking branch 'aosp/upstream-master' into mymerge am: 88c2703c71 am: 8456625b65

Change-Id: I4edabd2c9400c47f6bf4c900321fbd394b46f7a0
tree: f5dd6f49d5a89ff3ebf8d17554810ea01542504c
  1. bench/
  2. cmake/
  3. include/
  4. jni/
  5. test/
  6. third-party/
  7. .gitignore
  8. CMakeLists.txt
  9. configure.py
  10. confu.yaml
  11. LICENSE
  12. README.md
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.