In reply to mayurkubavat:
It measures performance :)
Most protocols or IP components have performance requirements such as: how many instructions, packets, or bytes need to be processed per second. There are many performance metrics that you can track. If you have a 10G Ethernet MAC, how do you know that you have created and run a testcase that proves your MAC really can process 10Gbps? The terminology existed long before OVM/UVM. I don’t know how EDA companies implement the code, but I track the DUT performance in the UVM Monitor and/or the UVM Scoreboard and collect/report the findings (via covergroups) after the run_phase completes (i.e. in final_phase()).