Anatomy of a Posit Number
johndcook.comShould say (2018)
Can people please stop posting about posits, every post about them dramatically misrepresents what they can represent while under present ieee754, and makes performance claims based on non like-for-like tests.
They all also seem to just point back to papers and articles written by the inventor.
> Can people please stop posting about posits,
The linked article is a decent explanation of how posits are defined in memory. While posits may be overhyped, technical articles are exactly what I come to HN for.
> every post about them dramatically misrepresents what they can represent while under present ieee754
This appears to be untrue as the linked article does not misrepresent Posits or IEEE 754 floats. It correctly points out the main reason to use Posits over floats being increased dynamic range for same number of bits. It doesn't mention the corresponding lack of precision, but that can comfortably be deduced from basic information theory.