You can already have sub 100ns latency on core-to-core communication on some common machines. And yes, thank if the magic of bus snooping the transfer can in pr

You can already have sub 100ns latency on core-to-core communication on some common machines. And yes, thank if the magic of bus snooping the transfer can in practice happen cache to cache, although counterintuitively it is not necessarily what you want for message queues.