Sunday 09 March 2025
As software has become an integral part of our daily lives, it’s no surprise that the energy consumption of these programs has become a growing concern. With the increasing demand for powerful devices and cloud services, the power-hungry nature of software development has sparked a debate about its environmental impact.
Researchers have been working to shed light on this issue by investigating the energy consumption of automated test generation tools. In particular, they’ve focused on EvoSuite, a widely used tool for generating unit tests in Java programs. The study aimed to measure the energy consumption of different algorithms implemented in EvoSuite and compare it with manual testing.
The team used JoularJX, a software power monitoring tool, to track the energy consumption of EvoSuite during test generation. They found that the energy consumption varied significantly among the different algorithms, with some consuming up to 100 times more energy than others. The results showed that the cyclomatic complexity of the class under test was a key factor in determining energy consumption.
The study also compared the energy consumption of automatically generated tests with manually written tests. Surprisingly, they found that manually written tests were often more energy-intensive than their automated counterparts. This could be due to the complex nature of manual testing, which involves multiple iterations and debugging processes.
One of the most interesting findings was the identification of a small subset of high-energy-consuming methods within EvoSuite. These methods were related to low-level operations such as thread management and I/O operations. The researchers suggest that optimizing these methods could lead to significant energy savings in test generation.
The study highlights the importance of considering energy consumption in software development, particularly in automated testing. As developers strive for more efficient and sustainable practices, this research provides valuable insights into the energy footprint of their tools.
In addition to its practical implications, this study demonstrates the potential benefits of interdisciplinary collaboration between computer science and environmental studies. By combining expertise from both fields, researchers can tackle complex issues like software sustainability and contribute to a more environmentally conscious approach to software development.
Cite this article: “Measuring the Environmental Impact of Automated Test Generation Tools”, The Science Archive, 2025.
Software, Energy Consumption, Automated Testing, Evosuite, Java Programs, Power Monitoring, Cyclomatic Complexity, Manual Testing, Thread Management, I/O Operations.
Reference: Fitsum Kifetew, Davide Prandi, Angelo Susi, “On the Energy Consumption of Test Generation” (2025).
