https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114480
--- Comment #37 from Richard Biener <rguenth at gcc dot gnu.org> --- (In reply to Jason Merrill from comment #36) > (In reply to Richard Biener from comment #34) > > I don't have any more patches or analysis in this bug, there are now > > separate bugs for found issues. > > So should this one be closed? I can't find the bug I remembered to open for the RA issue, so lets keep this open for this part. Current numbers for GCC 15.2 are the following - the memory usage of EH lowering/optimizing remains and -O0 suffers from lack of optimizing the number of pseudos and thus RA exploding. -O0: 195s, 4GB integrated RA : 103.21 ( 52%) LRA create live ranges : 53.38 ( 27%) -Og: 101s, 11GB tree eh : 11.49 ( 11%) 17643M ( 92%) tree SSA rewrite : 15.89 ( 15%) tree SSA incremental : 34.40 ( 33%) -O1: 148s, 11GB alias stmt walking : 11.15 ( 7%) tree eh : 11.87 ( 8%) 17643M ( 91%) tree PTA : 17.92 ( 12%) tree SSA rewrite : 15.76 ( 10%) tree SSA incremental : 35.74 ( 23%) expand vars : 10.38 ( 7%) -O2: 263s, 11GB ipa icf : 30.68 ( 11%) alias stmt walking : 21.39 ( 8%) tree eh : 11.50 ( 4%) 17643M ( 90%) tree PTA : 28.93 ( 11%) tree SSA rewrite : 15.50 ( 6%) tree SSA incremental : 34.77 ( 13%) expand vars : 52.85 ( 20%)
