https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80155
--- Comment #8 from Thomas Preud'homme <thopre01 at gcc dot gnu.org> --- (In reply to rguent...@suse.de from comment #6 and #7) > > The testcase shows hardly profitable PRE which the patches should > disable (I didn't verify the patches fix the testcase!) The patch does not help for the testcase either (at least the Cortex-M7 one). > > If you're benchmarking at -O2 can you add -fpredictive-commoning? As > said, the patch removes important loop optimizations from PRE. Sorry, it does not make any difference (either for the benchmark or the testcase).