https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123449
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2026-01-07
Ever confirmed|0 |1
--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
This is an effective regression for -march=znver5. I'll note that w/o partial
vectors we get
t.c:3:21: optimized: epilogue loop vectorized using 32 byte vectors and unroll
factor 32
t.c:3:21: optimized: epilogue loop vectorized using 16 byte vectors and unroll
factor 16
t.c:3:21: optimized: epilogue loop vectorized using 8 byte vectors and unroll
factor 8
and when versioning we still do not add to the versioning condition that
the number of iterations is >= 8 which is a similar missed optimization.
But we fall to the scalar epilog after 3 checks on niter.
