https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88751
--- Comment #5 from Vladimir Makarov <vmakarov at gcc dot gnu.org> --- (In reply to Andreas Krebbel from comment #4) > (In reply to Babneet Singh from comment #3) > > Hi Andreas and Richard: What's the status for this issue? Which approach > > will be used to resolve this issue? > > I would like to have Vladimir comment on this first, since he wrote the code > and definitely knows this stuff best. > It is very hard to define IRA algorithm complexity (there are too many factors to consider, e.g. live range lengths) but it is definitely not linear of number used pseudos. The current IRA heuristic is very old one. As I remember It was a very quick fix for a huge automatically generated function. I believe the proposed patch is a better solution. It takes pseudo sparseness into account. So, Andreas, please go ahead and commit the patch. It would be nice to introduce a parameter too. But it can be done in a separate patch. Thank you for the patch, Andreas. > Richard, would it be ok with you to raise the prio? OpenJ9 is a pretty > important workload I think.
