https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121256
--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #4)
> Nope, it seems the initial recurrence vector is bogus:
>
> vect_cst__84 = {blue_lsm.10_58, blue_lsm.10_58, blue_lsm.10_58,
> blue_lsm.10_58, blue_lsm.10_58, blue_lsm.10_58, blue_lsm.10_58,
> blue_lsm.10_58};
> ..
> # vect_vec_recur_.17_85 = PHI <vect_bt_41.15_48(9), vect_cst__84(2)>
In fact I think we can at most handle a VF of 2?! Given the initial
value is composed solely of latch values. We're missing SLP handling
here as well (so on branches need to best limit to SLP_TREE_LANES == 1).
