Google Git
Sign in
android / platform / external / FP16 / cee98717f6ccb0c92d5e0fc77cdc497d5ee223f3
commitcee98717f6ccb0c92d5e0fc77cdc497d5ee223f3[log] [tgz]
authorXin Li <[email protected]>Wed Jan 17 22:13:21 2024 -0800
committerXin Li <[email protected]>Wed Jan 17 22:13:21 2024 -0800
tree5861de0ba4651a75bafe7097b5665fed86eb6903
parent86a48f72b37fde8946f2a37d12ff03533cbd7e2d [diff]
parent9e3376fdab33d108e5d2dc4918f3692dc9330f61 [diff]
Merge Android 24Q1 Release (ab/11220357)

Bug: 319669529
Merged-In: Ie2f1b6ab5f20312f84266d98e9cc7f6245469efd
Change-Id: I073e72469467354d16e6d1844c0a7a150466b2e1
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.