https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120892
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
See Also| |https://gcc.gnu.org/bugzill
| |a/show_bug.cgi?id=112402
CC| |pinskia at gcc dot gnu.org
Last reconfirmed| |2025-06-30
Ever confirmed|0 |1
--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
I was originally going to move path splitting to RTL after unrolling but there
was some performance regressions with that; I don't know if I have the patch
still though (see https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112402#c2 for
information on this idea and some of the history around path splitting which I
had collected a few years back). What I did to resolve the ifcvt issue was tune
the heurstics in path splitting instead. That obvious didn't solve the
unrolling issue. Maybe there is tuning to heurstics that can help if the trip
count is constant.
