https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114635
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords| |openmp
Blocks| |53947
CC| |rguenth at gcc dot gnu.org,
| |rsandifo at gcc dot gnu.org
--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
We have
(compute_affine_dependence
ref_a: D.5604[_33], stmt_a: _65 = D.5604[_33];
ref_b: D.5604[_33], stmt_b: D.5604[_33] = _ifc__144;
) -> dependence analysis failed
but of course safelen should apply here, but we likely hit
if (max_vf != MAX_VECTORIZATION_FACTOR
&& maybe_lt (max_vf, min_vf))
return opt_result::failure_at (vect_location, "bad data dependence.\n");
we have max_vf == 16 but min_vf is {4, 4} with E_VNx4SFmode. Maybe
safelen (and thus max_vf) needs to be poly-int? Maybe somehow safelen
needs to be ensured by additional masking with SVE?
Referenced Bugs:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=53947
[Bug 53947] [meta-bug] vectorizer missed-optimizations
