https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115932
--- Comment #2 from Hans-Peter Nilsson <hp at gcc dot gnu.org> --- (In reply to Andrew Pinski from comment #1) > I wonder if reapplying the fix for PR 116028 will improve the situtation > here. No. Baseline: r15-7318-gdd6247cb8fc1 coremark cycles: 4897981 (Re-)applying 3c67a0fa1dd39a3378d and passing -flate-combine-instructions: 5033977 Re-applying that commit made very little change both with and without -flate-combine-instructions.