https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83800
--- Comment #4 from joseph at codesourcery dot com <joseph at codesourcery dot com> --- Functions bound to IEEE operations, such as sqrt and fma, should be correctly rounding for all IEEE floating-point types (so not IBM long double) supported by glibc. However, there's no generic ldbl-128 sqrtl / sqrtf128 in glibc; each architecture implements it either using soft-fp, or using hardware sqrt instructions where supported. There's a case that when built against new-enough glibc, quadmath.h should be mapping most function calls to call the libm *f128 functions instead (I suppose more work elsewhere might be needed to do that mapping for calls from Fortran code).
