math/universal 3.75 DragonFly

The goal of Universal Numbers, or unums, is to replace IEEE floating-point with a number system that is more efficient and mathematically consistent in concurrent execution environments.

The motivation to replace IEEE floating-point had been brewing in the HPC community since the late 90's as most algorithms became memory bound. The inefficiency of IEEE floating-point has been measured and agreed upon, but it was the AI Deep Learning community that moved first and replaced IEEE with number systems that are tailored to the application to yield speed-ups of two to three orders of magnitude.

The Universal library is a ready-to-use header-only library that provides plug-in replacement for native types, and provides a low-friction environment to start exploring alternatives to IEEE floating-point in your own algorithms.