[Bug tree-optimization/68654] [6 Regression] CoreMark Pro performance degradation

["rguenther at suse dot de" <gcc-bugzilla at gcc dot gnu dot org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68654

--- Comment #20 from rguenther at suse dot de <rguenther at suse dot de> ---
On Tue, 26 Jan 2016, law at redhat dot com wrote:

> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68654
> 
> --- Comment #19 from Jeffrey A. Law <law at redhat dot com> ---
> What's initially more important than a particular improvement or degradation of
> performance is bringing more stability to the coalescing results -- so that we
> can focus on issues with coalescing rather than getting distracted by the
> semi-random performance improvements/drops that we have now.

Well, you can simply insert a pass_release_ssa_names before pass_expand
then which will compact SSA names for you.  It keeps SSA name ordering
the same though so it probably doesn't gain anything.

Well, and then make the coalesce list stable by preserving the
coalesce candidate creation ordering (which works by walking BBs
and thus should be more stable in some sense)