https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103116
--- Comment #6 from rguenther at suse dot de <rguenther at suse dot de> --- On Wed, 4 May 2022, rsandifo at gcc dot gnu.org wrote: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103116 > > --- Comment #5 from rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> > --- > (In reply to Richard Biener from comment #3) > > We could make peeling for gaps handle this by making it not a flag but > > indicate the number of vector(!?) iterations we need to peel. > I think it'd be better to keep it in scalar iterations if possible. > For example, we still need to peel for gaps when using masked LD2s > if only the first vector result is needed. But we do only need to > peel 1 scalar iteration, rather than a whole vector's worth. > > But yeah, agree that moving from 0-or-1 to a general number sounds good. Yep. I'll keep that for GCC 13, the posted fix should be what we can reasonably easy backport.