Double precision floating point numbers are the most common in scientific computing. Sometimes software will mix in single precision for more speed. The molecular dynamics package GROMACS can use mixed single/double precision, for example:

http://manual.gromacs.org/documentation/2016/install-guide/i...