Getting requests up through the cache hierarchy to the DRAM controller, and data back down to the requesting core's load/store units is also a non-trivial part

Getting requests up through the cache hierarchy to the DRAM controller, and data back down to the requesting core's load/store units is also a non-trivial part of this total latency.