https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124174

--- Comment #2 from Alex Coplan <acoplan at gcc dot gnu.org> ---
(In reply to Richard Earnshaw from comment #1)
> >it is wrong because a NEON vadd should be well-defined on overflow
> 
> As we discussed internally, I disagree with this claim, so I'm not convinced
> this is actually a bug.

As per
https://github.com/ARM-software/acle/blob/main/neon_intrinsics/advsimd.md,
vaddq_s32 is specified to have the behaviour of the NEON instruction:

ADD Vd.4S,Vn.4S,Vm.4S

which is well-defined on overflow.  Likewise for the other types, of course.

Reply via email to