Intel's L3 is sharded out into 2mb chunks on a ring bus too so I would have expected more comparable numbers. Is Intel's QPI link and cache coherency implementi

Intel's L3 is sharded out into 2mb chunks on a ring bus too so I would have expected more comparable numbers. Is Intel's QPI link and cache coherency implemention that much better?