Posit floating point numbers: thin triangles and other tricks (2019)

Re:Posit floating point numbers: thin triangles and other tr... by adgjlsfhk1 on Thursday June 19, 2025 @03:37PM (#44321823)

I really want a standards body to standardize a positized version of floating point numbers. Everything about the encoding of posits (i.e. NaR, no -0.0, exponent encoding) is so much nicer. The tapered precision part of posits is IMO the least interesting part.

[ Parent | on HN ]

Re:Posit floating point numbers: thin triangles and other tr... by dnautics on Thursday June 19, 2025 @03:55PM (#44321988)

the twos complement thing is really elegant (-p == (posit -(int p))) in c notation

[ Parent | on HN ]

Re:Posit floating point numbers: thin triangles and other tr... by adgjlsfhk1 on Thursday June 19, 2025 @05:25PM (#44322655)

relatedly, having the same ordering as signed ints just makes everything so much easier. You can just use your integer radix sort to sort your floats without having to deal with sign and nans first!

[ Parent | on HN ]

Re:Posit floating point numbers: thin triangles and other tr... by immibis on Thursday June 19, 2025 @05:49PM (#44322830)

So do floats if you don't care about NaN

[ Parent | on HN ]

Re:Posit floating point numbers: thin triangles and other tr... by tomsmeding on Thursday June 19, 2025 @07:09PM (#44323355)

They do not; if you sort as unsigned ints then negative floats sort above positive ones, and if you sort as signed ints then the negative range is reversed. IEEE floats are sign-magnitude, not twos-complement.

[ Parent | on HN ]

Re:Posit floating point numbers: thin triangles and other tr... by andrepd on Thursday June 19, 2025 @08:45PM (#44323834)

Add "simple and consistent rounding" to that list!

There's a tentative standard btw: https://posithub.org/docs/posit_standard-2.pdf

[ Parent | on HN ]