[hcs]

Memory access time is a constant multiplier, whether it hits or misses the cache. There can be a big difference in that constant depending on the access pattern, but it is still considered a constant factor.

Yes, random access (missing the cache every time) may be 100x slower than sequential (hitting the cache almost always), but if you're iterating through an array twice as large, it will still be 100x slower.

[JohnLeTigre]

I agree, on the same machine, the slow-downs are constant for each level of cache and for ram access.