Google Git
Sign in
android / platform / external / FP16 / refs/heads/android13-s1-release
commitfc8d111c5fbe65583e0fdaf6a491be326e6b09a1[log] [tgz]
authorAndroid Build Coastguard Worker <[email protected]>Sat Mar 19 23:02:30 2022 +0000
committerAndroid Build Coastguard Worker <[email protected]>Sat Mar 19 23:02:30 2022 +0000
tree2c085b686688abfc7135e4727b9ff04a6e23a4b6
parent0ea72199b8d3262edf0b37bfc2901403fd7030ef [diff]
parent8217ca7fcb18ecd3b12e952f6e41c5b9316fe411 [diff]
Snap for 8326520 from 8217ca7fcb18ecd3b12e952f6e41c5b9316fe411 to tm-release

Change-Id: I9538c6deb8bf584794c44964292c299730bbfcd7
tree: 2c085b686688abfc7135e4727b9ff04a6e23a4b6
  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.