Precision-doubled floating point types nearly as performant as hardware floats.
Author JeffreySarnoff
2 Stars
Updated Last
2 Years Ago
Started In
July 2019


Copyright © 2017-2020 by Jeffrey Sarnoff. This work is released under The MIT License.

XFloats exports two these two types: XFloat16, XFloat32.

  • These floating point types correspond to Float16 and Float32.
  • Use them to compute results with greater accuracy at very high speed.


           relative speeds
           X32∝F32  X64∝F64

    add:     1.00    1.08
    mul:     1.08    1.08
    fma:     0.93    1.08

other extended precison floating point types
  • The more accurate type for Float64 is Double64 (DoubleFloats.jl).
    • It calculates more slowly than these types do.

Used By Packages

No packages found.