Google Git
Sign in
chromium / external / github.com / Maratyszcza / FP16 / a8f140a6bcb6cfa40095e28ff9167fbd20a5b6ce
commita8f140a6bcb6cfa40095e28ff9167fbd20a5b6ce[log] [tgz]
authorMarat Dukhan <maratek@gmail.com>Thu Mar 23 11:44:07 2017
committerMarat Dukhan <maratek@gmail.com>Thu Mar 23 11:44:07 2017
tree7a083c766f92cb89134eca4d48e53648056664e2
parent13dca24b4d8b284bc0220a386e8a20ac67104cd3 [diff]
Disable inline assembly in google benchmark for wider compatibility
1 file changed
tree: 7a083c766f92cb89134eca4d48e53648056664e2
  1. bench/
  2. include/
  3. test/
  4. third-party/
  5. .gitignore
  6. configure.py
  7. confu.yaml
  8. LICENSE
  9. README.md
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.