Universal Numbers (Unum) and floating points are complicated. Here’s an explainer on the subject.
Error-Free Computing: Unums Save Both Real and Virtual Battles
VR World chats with John Gustafson about the challenges of implementing universal numbers into hardware, and the benefits they offer computing.
Supercomputing Frontiers 2015: The 101×10^2 Problem. Solution: Unums
We’ve almost reached the acceptance limit for floating point rounding errors. What’s the future? One potential solution was explained at Supercomputing Frontiers 2015.