Cache hierarchy
Two, and even three, levels of caches quite common now
L2 very large but since L1 filters many references, “local” hit rate might appear low (maybe 50%) (compulsory misses still happen)
L2 has in general longer cache blocks and might have more associativity
Multi-inclusion level property is implemented most of the time (with also L1 WT and L2 WB)
Prefetching at L2 level is an interesting challenge