http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51616
--- Comment #5 from Steve Kargl <sgk at troutmask dot apl.washington.edu> 2011-12-19 02:21:34 UTC --- On Mon, Dec 19, 2011 at 01:45:16AM +0000, danglin at gcc dot gnu.org wrote: > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51616 > > --- Comment #4 from John David Anglin <danglin at gcc dot gnu.org> 2011-12-19 > 01:45:16 UTC --- > Possibly, you referred to the wrong PR. PR 51057 is about the precision > of IBM long double. > Read the entire PR. You'll eventually find the thread starting at http://gcc.gnu.org/ml/fortran/2011-11/msg00051.html hpux11 appears to be yet another OS that has sufficient quad support that gfortran detects a REAL(16) type, but the OS lacks the basic libm functions.