Spurious test in gfortran.fortran-torture/execute/intrinsic_mod_ulo.f90

[Tobias Schlüter]

Paul Thomas wrote:
> Both rewrites of MOD that I am preparing do better than the existing, 
> from x/y ~ 10**17 up to 10**19.  Beyond this, the fast one gives wild 
> values, like the present function, whereas the "correct" one is about 
> 20% slower.
> 
> Which version would folk like to see?

Garbage in, garbage out?  Otherwise, conditionalizing on -ffast-math seems
appropriate, since even though you're not providing a "correct" result in the
slow case, it's at least predictable.

- Tobi