Is it a good idea to re-benchmark occasionally? Maybe as difficulty changes or other things, an occasional re-benchmark will make estimates and related auto-switching more accurate? Or is that simply not necessary if hardware has not changed?