Normally in fortran, setting up a variable as "real" makes it a 32-bit float, while "double precision" or "real*8" makes it a 64-bit float.
-r8 is a flag used by many compilers to force both "real" and "real*8" variables to be 64-bit floats. gfortran has this option too, but the flag is called "-fdefault-real-8" instead. It may also promote "double precision" variables to 128-bit, if possible on your system.
Last edited by suicidaleggroll; 05-28-2013 at 10:25 AM.