If nvme swap gets faster too with huge pages then testing a whole microservice swarm on a local dev machine might benefit from it when the currently idle services can be swapped more swiftly while something chews on data.