https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121993

--- Comment #2 from Filip Kastl <pheeck at gcc dot gnu.org> ---
Created attachment 62431
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=62431&action=edit
spec config file to reproduce the slowdown with

(In reply to cuilili from comment #1)
> I don't have znver2 machine, so we tried to reproduce the 519 regression on
> znver3 machine using the "-O2 -flto" options (both with and without the
> commit). The results showed that the generated binaries were identical.
> Could you please help confirm this ?Thanks.

That's strange.  I'm definitely getting different binaries when I build
519.lbm_r with r16-3485 and r16-3484.  Here's the SPEC config that I use.

Reply via email to