在集群中 Low Latency 应用,The operating systems community has ignored network latency for too long. In the past, speed-of-light delays in wide area networks and unoptimized network hardware have made sub-100μs round-trip times impossible. However, in the
The table describes throughput and latency for processors with two FMA units, assuming all sources come from the FMA unit.
See FMA latency chapter in the optimization guide for more information.
Memory latencies are assuming Data Cache Unit (DCU) hit