A particular set of bits in each address is used to determine
the corresponding location in the cache. Why use that particular set
of bits?  Why not, for example, use the high-order (leftmost) bits in
the address to determine the cache location?