https://gcc.gnu.org/bugzilla/show_bug.cgi?id=128035
Hongtao Liu <liuhongt at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords|missed-optimization |
--- Comment #1 from Hongtao Liu <liuhongt at gcc dot gnu.org> ---
It's impacted by the macro COMPARE_MAX_PIECES which is by default
#define COMPARE_MAX_PIECES MOVE_MAX_PIECES
MOVE_MAX_PIECES is 16 bytes with sse2 and above but drop to UNITS_PER_WORD w/o
sse2.
Maybe we should define MOVE_MAX_PIECES to 2 * UNITS_PER_WORD w/o sse2 since x86
backend already supports {mov,cmp}{di,ti}_doubleword, so it should be
functional OK with that?
