https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124580
--- Comment #4 from Jeffrey A. Law <law at gcc dot gnu.org> --- Ha. Our comments crossed in the ether last night and I didn't see the bugzilla mid-air collision note until just now. The requested optimization is invalid inless we can prove that the scores array can't overlap with the input array. Robin and I looked at this a while back :-) The only sensible path forward for the optimization would be to version the code based on the aliasing properties at runtime. That felt like pie in the sky to me.
