| commit | ef664cfb655e748463eba7907bcf12f46aade41d | [log] [tgz] |
|---|---|---|
| author | Android Build Role Account android-build-prod <android-build-team-robot@google.com> | Sat Feb 22 03:25:07 2020 +0000 |
| committer | Android Build Role Account android-build-prod <android-build-team-robot@google.com> | Sat Feb 22 03:25:07 2020 +0000 |
| tree | a426169d62e9a749341225ef30aa64fea2af39f2 | |
| parent | 08a172d6da973d8750165c40316df6fade8ab364 [diff] | |
| parent | 8e0c0b31a917d79952b2e87d3b93790d306ac378 [diff] |
Snap for 6227608 from 8e0c0b31a917d79952b2e87d3b93790d306ac378 to r-keystone-qcom-release Change-Id: I1030ec1691b75baa513e8178c53a58b1131a615f
tree: a426169d62e9a749341225ef30aa64fea2af39f2
- bench/
- cmake/
- include/
- test/
- NOTICE ⇨ LICENSE
- .gitignore
- Android.bp
- CMakeLists.txt
- configure.py
- confu.yaml
- LICENSE
- METADATA
- MODULE_LICENSE_MIT
- OWNERS
- README.md
- 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
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.