While a lot of the applications in the world run on clusters of systems with a relatively modest amount of compute and memory compared to NUMA shared
