Computing Appliances – Real-World Active Mode Benchmarking

Citation

  • Zivojnovic, V. and Mista, D. (2022). Computing Appliances – Real-World Active Mode Benchmarking. Canadian Standards Association, Toronto, ON.

Executive Summary

Standardized energy testing procedures, like those defined by the Energy Star program for computers, are key tools. They help the industry, energy efficiency advocates, and regulators to assess the energy use of computing appliances and plan improvements. With the evolution of technology and the emergence of mobile phones and tablets, typical idle power demand of key computer components has been reduced significantly. Therefore, the Energy Star tests’ key assumptions for computers no longer represent real-world use. It is like measuring the gas mileage of a car while coasting downhill. As such, it may mislead policymakers and consumers, and encourage manufacturers to invest in energy efficiency improvements that may not produce expected reductions in real-world energy use. This research project defines a new test procedure to assess the real-world energy consumption of computers in their active state. The aim is to enable the industry, energy efficiency advocates, and regulators to better assess and leverage computers’ energy savings potential cost-effectively in residential and commercial buildings. The new Real-World Active Mode Benchmark proposed here defines the information content and scenarios for executing a standard set of “light active” tasks to measure the active mode power demand of computers while performing common activities. The information content is web based and resides on a local or remote web server that is part of the execution environment. User activity is emulated by using scripted control of the web browsers. The test procedure defines the sequence of steps conducted by laboratory technicians, including the setup of the device under test, benchmark execution, and energy measurements. This Real-World Active Mode Benchmark can be the basis of an enhanced standardized test procedure for computing appliances representing real-world consumption better and meaningfully in a world where modern standby is broadly adopted in the market. The new test procedure is validated on a set of standard computer types and most popular browsers. The achieved results are reviewed for representativeness, repeatability, reproducibility, reasonability, and portability. In this project the content is stored locally, on a web server machine within the same local area network (LAN) as the computer tested. Next, to improve benchmark representativeness, the web server machine should be accessed over the wide area network (WAN), i.e., the internet. Additionally, larger scale testing involving other types of desktops and laptops running under other operating systems, like those of Apple, should be included. Modern computers have vastly improved idle mode power consumption, but disproportionally less than active mode consumption. Additionally, modern computers are 24/7 operating devices spending most of their duty cycle in active mode. Therefore, to avoid misguiding policymakers and the industry, it is necessary to retest modern computers and revise the Energy Start test procedure accordingly. The Active Mode Benchmark proposed here can guide future revisions of the Energy Star test procedure.