The Amsterdam-based Software Improvement Group (SIG) and the Hogeschool van Amsterdam (HvA) University of Applied Science have come together to create the Software Energy Footprint Lab (SEFL), opened in February this year.
The lab enables researchers to investigate questions such as:
How do different database management systems compare with each other in terms of energy consumption?
How do different programming languages compare with each other in terms of energy consumption?
How does asynchronous requests compare to synchronous requests in terms of energy consumption?
How does unsigned integer arithmetic operations compare to signed arithmetic operations with respect to energy consumption?
How accurate are software energy profiling tools?
SIG is a consultancy firm that provides quantitative assessments of risks related to corporate software systems. It started out as a spin-off of the Dutch National Research Institute for Mathematics and Computer Science, conducting research into software quality using source code analyses. According to the company, its analysis software can measure quality, reveal the underlying architecture and assess the risks related to a solution.
To measure the software energy footprint the new lab has computers specially rigged with sensors to measure the electric current into each of its components. While special programs or generic benchmarks run, the sensors report on how much current is flowing to each component. This brings together the electronics knowledge of the HvA students with the software expertise of SIG.
The SEFL is hoping that other researchers and practitioners will join in defining more research questions, designing better experiments or in discussions around this topic. The SEFL is open to anyone interested in software energy use.
There’s an increasing amount of focus on the energy used by software. It’s not just about the efficiency of the application itself, but also the extent to which it keeps disks spinning and other peripherals running for no good reason. The software can even prevent PCs and servers from going into a power-saving mode when not in use.
So assessing software on the energy it uses is important. It would be even better if the results for companies and solutions were publicly available so that potential purchasers could add energy used as a purchase criteria. We already have it for hardware (Energy Star, etc) so why not software? Solutions providers could quote low energy use as a product feature, or explain what the functionality is that means it uses more energy, and let purchasers decide.