https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115825
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |hubicka at gcc dot gnu.org,
| |rguenth at gcc dot gnu.org
--- Comment #6 from Richard Biener <rguenth at gcc dot gnu.org> ---
We could try to record the size portion that is known to _not_ simplify
(like volatile stores) and do (unr_insns - (unr_insns - size.fixed) * 1 / 3)
to not over-estimate the 2/3 shrinking in cases like this. Or count
the insn part that is a candidate for later simplification/CSE - that's
mostly all IV dependent stmts (that are not volatile) and that didn't end up
estimated constant.
