“Posit” extension to RISC-V ISA is a possible alternative to floating point

I was reading through the RISC-V ISA spec―specifically the section on floats―and thought "I wonder if RISC-V and its brand-new ISA are a good time to introduce that 'posit' thing I read about a couple years ago."

Then I looked it up, and, of course, a bunch of smart people had arrived at that conclusion before me. Thought it might be worth sharing here.

More technical details can be found in the PERI paper [0].

    [0]: https://arxiv.org/abs/1908.01466