https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127216
Hongtao Liu <liuhongt at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |liuhongt at gcc dot gnu.org
--- Comment #1 from Hongtao Liu <liuhongt at gcc dot gnu.org> ---
The fallback is for the backends which don't support avg_floor, looks like
riscv backend already have autovec.md:(define_expand "avg<mode>3_floor", maybe
it does not cover the related mode in the regressed testcase?
originally, when fallback is false, the vectorizer will manually do sign extend
, plus and then >> 1, does some riscv backend optimization rely on that to
generate widen mode avg_floor?
