"In certain memory bound benchmarks, NUMA systems perform better
than SMP systems because the available aggregate memory bandwidth
actually increases as more nodes are added to the system. This is
because processors can access local memory in parallel, with no
contention for a common bus."
Some of the products that appear on this site are from companies from which QuinStreet receives compensation. This compensation may impact how and where products appear on this site including, for example, the order in which they appear. QuinStreet does not include all companies or all types of products available in the marketplace.