https://issues.dlang.org/show_bug.cgi?id=24125
--- Comment #3 from Steven Schveighoffer <schvei...@gmail.com> --- (In reply to Walter Bright from comment #2) > Looks like there are two problems. The first is: > > typedef float __m128 __attribute__ ((__vector_size__ (16), > __may_alias__)); > > This doesn't recognize __m128 as a vector. But this works: > > typedef float __attribute__ ((__vector_size__ (16), __may_alias__)) > __m128; I may have copied the header code wrong... I'll double check. That does look more reasonable. --