https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114010
--- Comment #8 from Richard Biener <rguenth at gcc dot gnu.org> --- It's expected that SSA naming can affect code generation, there's no "declaration order" as for VAR_DECLs we could ultimatively fall back to. Ideally algorithms should not depend on particular ordering so it might be good to investigate particular bad cases. For that I'd even suggest to add a debug mechanism to "shuffle" SSA names somewhat randomly (but note statements should be re-canonicalized afterwards). There are some cases with SLP vectorization of COND_EXPRs that are problematic where both the false/true arms and the condition can be "canonicalized" (yes, the inside the vectorizer COND_EXPRs with embedded conditions are still a thing - thanks to pattern recog). I don't know of any others.
