Software for testing compact representations of floating-point values

This software tests the performance of the compact floating point representations described in the following paper:

Neal, R. M. (2015) ``Representing numeric data in 32 bits while preserving 64-bit precision'', Technical Report, 16 pages: abstract, pdf.

The programs are written in C.

First (and probably only) release of 8 April 2015: tar archive.

Back to Radford Neal's home page