https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93528
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords| |missed-optimization
Status|UNCONFIRMED |NEW
Last reconfirmed| |2020-02-04
CC| |rguenth at gcc dot gnu.org
Component|c++ |tree-optimization
Ever confirmed|0 |1
--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed. The small power-of-two sizes benefit from turing the memcpy calls
into copies with register temporaries.
The thing to notice is that we can "sink" the store to x.arr[0] across the
memcpy to p by rewriting it accordingly. "Fitting" passes might be
SRA or store-merging.
