https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116855
Richard Biener <rguenth at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Ever confirmed|0 |1 Keywords| |wrong-code CC| |rguenth at gcc dot gnu.org Summary|Unsafe early-break |[14/15 Regression] Unsafe |vectorization |early-break vectorization Version|unknown |14.2.1 Target Milestone|--- |14.3 Status|UNCONFIRMED |NEW Last reconfirmed| |2024-09-26 Blocks| |53947 --- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> --- We're using ref_within_array_bound and I think that's OK in the case we're using loop masking - but when not using loop masking and in particular also masking the load, relying on it is broken - we'd have to amend the API to pass in a maximum number of space for excess elements accessed. Relying on alignment of course works as well. Thus confirmed, also happens on x86 with -msse4.1 Referenced Bugs: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=53947 [Bug 53947] [meta-bug] vectorizer missed-optimizations