https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124659
--- Comment #3 from Chen Yuanwei <522024330006 at smail dot nju.edu.cn> --- Thanks for the explanation regarding the memory layout of _mm_set_epi32！ However, I'm curious about the specific optimization transformation occurring at -O2. Is the discrepancy in output (getting 01 instead of an immediate break) a direct consequence of Alias Analysis？Or is it correct to say that because this is UB, the compiler is no longer guaranteeing sequential consistency for these memory accesses? I'd appreciate a brief confirmation if this is considered a formal Undefined Behavior due to strict aliasing violations.
