https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103000
--- Comment #2 from Tamar Christina <tnfchris at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #1)
> Looks like these are missing { target { vect_complex_add_float } } in there.
>
No, that's not right, `vect_complex_add_float` is for when the target supports
code-gen for the the sequence. But detection should pass for any target that
can SLP vectorize the test case at all.
>From a quick look at the failures on -m32 it looks like SLP vectorization fails
completely but these targets declare vect_float so that's not enough to stop
it.
Will need to check monday but likely missing some option somewhere. And
https://godbolt.org/z/Pxs7dfj1a shows it detecting them, so something is up
with the options being passed.
Will look monday.
