[Bug c/82349] float INFINITY issue with division by zero in regression with compiler option '-ffast-math'

[marc.pres at gmx dot net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82349

--- Comment #4 from Marcello Presulli <marc.pres at gmx dot net> ---
Thats clear, but why are the 2 terms different each.
If so, they should output the same, because what would be the difference
between 1.0/denom and 1.0/0 in case of fast-math ?