Module extended_float

Extended precision floating-point type.

Also contains helpers to convert to and from native rust floats. This representation stores the mantissa as a 64-bit unsigned integer, and the exponent as a 32-bit unsigned integer, allowed ~80 bits of precision (only 16 bits of the 32-bit integer are used, u32 is used for performance). Since there is no storage for the sign bit, this only works for positive floats.

Structs

• ExtendedFloat
  Extended precision floating-point type.