Google Git
Sign in
android / platform / external / FP16 / 9e3376fdab33d108e5d2dc4918f3692dc9330f61
commit9e3376fdab33d108e5d2dc4918f3692dc9330f61[log] [tgz]
authorXin Li <delphij@google.com>Sat Oct 07 00:20:50 2023 +0000
committerAutomerger Merge Worker <android-build-automerger-merge-worker@system.gserviceaccount.com>Sat Oct 07 00:20:50 2023 +0000
tree5861de0ba4651a75bafe7097b5665fed86eb6903
parent6a0eb56e81ecf6f64a016f8bf25ed06dab301c3a [diff]
parent86a48f72b37fde8946f2a37d12ff03533cbd7e2d [diff]
[automerger skipped] Merge Android 14 am: e04a018721 -s ours am: 3603450018 -s ours am: c2696316bf -s ours am: ba3d2d3b79 -s ours am: 86a48f72b3 -s ours

am skip reason: Merged-In I7ca632fa39c0d5190e10f87fb4fe1419a149a69f with SHA-1 c86276f80b is already in history

Original change: https://android-review.googlesource.com/c/platform/external/FP16/+/2775002

Change-Id: Ib7e62e0c786039af7963ecc46084eb62027da3ce
Signed-off-by: Automerger Merge Worker <android-build-automerger-merge-worker@system.gserviceaccount.com>
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.