[Bug ipa/92816] [10 Regression] 35% performance drop for 433.milc with -O2 -flto on znver1 since r278879

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92816

--- Comment #3 from Martin Liška <marxin at gcc dot gnu.org> ---
(In reply to Jan Hubicka from comment #2)
> It does not reproduce for me on bdver1 and also it does not seem to
> reproduce at skylake (with generic tuning)
> https://lnt.opensuse.org/db_default/v4/SPEC/graph?highlight_run=8405&plot.
> 70=227.70.0
> 
> So we need to profile it on zen...

Yes. Feel free to grab a free Zen machine (a non-LNT machine) we have at SUSE.