Run-Time Energy Consumption Estimation Based on Workload in Server Systems
Source: University of Louisiana
This paper proposes to develop a system-wide energy consumption model for servers by making use of hardware performance counters and experimental measurements. The paper develops a real-time energy prediction model that relates server energy consumption to its overall thermal envelope. While previous studies have attempted system-wide modeling of server power consumption through subsystem models, the approach is different in that it uses a small set of tightly correlated parameters to create a model relating system energy input to subsystem energy consumption. The paper develops a linear regression model that relates processor power, bus activity, and system ambient temperatures into real-time predictions of the power consumption of long jobs and as result controlling their thermal impact.