1*5f32b710SXin Li# FP16 2*5f32b710SXin LiHeader-only library for conversion to/from half-precision floating point formats 3*5f32b710SXin Li 4*5f32b710SXin Li## Features 5*5f32b710SXin Li 6*5f32b710SXin Li- Supports IEEE and ARM alternative half-precision floating-point format 7*5f32b710SXin Li - Property converts infinities and NaNs 8*5f32b710SXin Li - Properly converts denormal numbers, even on systems without denormal support 9*5f32b710SXin Li- Header-only library, no installation or build required 10*5f32b710SXin Li- Compatible with C99 and C++11 11*5f32b710SXin Li- Fully covered with unit tests and microbenchmarks 12*5f32b710SXin Li 13*5f32b710SXin Li## Acknowledgements 14*5f32b710SXin Li 15*5f32b710SXin Li[](http://hpcgarage.org) 16*5f32b710SXin Li[](http://www.cse.gatech.edu/) 17*5f32b710SXin Li 18*5f32b710SXin LiThe library is developed by [Marat Dukhan](http://www.maratdukhan.com) of Georgia Tech. FP16 is a research project at [Richard Vuduc](http://vuduc.org)'s HPC Garage lab in the Georgia Institute of Technology, College of Computing, School of Computational Science and Engineering. 19*5f32b710SXin Li 20*5f32b710SXin LiThis 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. 21
){kind=link}
){kind=link}