Floating-point number precisions according to IEEE-754.
Single precision: 64-bit.
Double extended precision: 80-bit.
See Implementation
Floating-point number precisions according to IEEE-754.