[Bug c++/69564] [5/6 Regression] lto and/or C++ make scimark2 LU slower

[jason at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69564

Jason Merrill <jason at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |NEW
           Assignee|jason at gcc dot gnu.org           |unassigned at gcc dot gnu.org

--- Comment #9 from Jason Merrill <jason at gcc dot gnu.org> ---
(In reply to Jason Merrill from comment #7)
> But if I add loop inversion back into the C++ front end
> so that the .optimized output is indistinguishable, that resolves the
> difference without LTO

Actually, no, the loop inversion patch in comment 8 doesn't improve
performance; even though the .optimized dump is the same between C and C++, the
C++ build is still slower by roughly the same amount.  I guess the RTL
optimizers must be doing something different?  Mysterious.  Unassigning myself.