Google Git
Sign in
android / platform / external / FP16 / 348e89de3646e144ea832236322933918b33ff03
commit348e89de3646e144ea832236322933918b33ff03[log] [tgz]
authorXin Li <[email protected]>Wed Jan 31 02:03:46 2024 +0000
committerAutomerger Merge Worker <[email protected]>Wed Jan 31 02:03:46 2024 +0000
tree5861de0ba4651a75bafe7097b5665fed86eb6903
parent9e3376fdab33d108e5d2dc4918f3692dc9330f61 [diff]
parentcee98717f6ccb0c92d5e0fc77cdc497d5ee223f3 [diff]
[automerger skipped] Merge Android 24Q1 Release (ab/11220357) am: cee98717f6 -s ours

am skip reason: Merged-In Ie2f1b6ab5f20312f84266d98e9cc7f6245469efd with SHA-1 3bd198b649 is already in history

Original change: https://googleplex-android-review.googlesource.com/c/platform/external/FP16/+/25973767

Change-Id: I77bd0ffb19d5fb6e80bb15a7448db624ed2c3fc2
Signed-off-by: Automerger Merge Worker <[email protected]>
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.