https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123204
--- Comment #3 from Victor Do Nascimento <victorldn at gcc dot gnu.org> --- Yep, working on the necessary cost model changes... We have a reduced testcase for this (albeit considerably smaller) slowdown in AArch64 that I'm using to look into costing. Just trying to make sure the same loop is also the culprit on x86_64. I'll create the Bugzilla for it later today once I've also looked into this x86 issue
