3 ms·
Excepting misalignment that causes a variable to straddle a cache line, modern x86 is generally extremely tolerant of misalignment, performance-wise (for GPR-si
by loeg 7y ago
Excepting misalignment that causes a variable to straddle a cache line, modern x86 is generally extremely tolerant of misalignment, performance-wise (for GPR-sized operations). Newer x86 is even pretty performant in the face of misaligned vector-sized accesses.