https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68263
--- Comment #1 from Uroš Bizjak <ubizjak at gmail dot com> --- (In reply to H.J. Lu from comment #0) > Misaligned load/store are only handled for AVX, not SSE. This is because only AVX supports instructions with unaligned memory operands, and we have to provide a way to move these unaligned operands to a register. SSE doesn't support this functionality, so there is no need for generic unaligned SSE loads.
