Summary taskset is used to affine an application to a specific set of CPU cores; usally taking into account NUMA architecture. CPU/memory affinity can improve performance of workloads that are performance-critical and/or that may be network-heavy. Affining closely related applications on the same NUMA node can improve memory access latencies, and newtork-heavy workloads can take advantage of affining applications by scheduling network I/O on the same NUMA node as the NIC card to reduce the latency between the CPU and PCIe.