Single page Print

Memory subsystem performance

My six cores each have very fast L1 and L2 caches attached, so my cache bandwidth at the smaller block sizes is, uh, prodigious. Yeah, yeah, Sandy B's fancy internal ring bus gives her an edge at the 4MB size. But look, I'm back in the saddle again at 16MB thanks to my massive 12MB L3 cache. There's really no comparison.

My memory bandwidth and latency are among the best for PC processors, too, just as expected. Sure, the Core i7-950 scores a little higher in Stream, and the Core i7-875K has the lowest latency measured, but those are fluke performances from aging 45-nm chips with no real future. I'm your best bet all around.