2 ms·

Superscalar execution has nothing to do with caches. You could do it on an architecture with no caches at all.