Fortran has been using `int8`, `int16`, `int32`, `int64`, and `real32`, `real64`, `real128` kinds officially for at least 2 decades. `double precision` has been long declared obsolescent, although still supported by many compilers. Regarding types and kind, (modern) Fortran is tremendously more accurate and explicit about the kinds of types an values.
(b) Many important machines had word sizes that were not a multiple of 8.