Energy-Aware Software Development in Operating Systems : An Analysis of OS-LevelOptimization Techniques

Read full text Summarise with VIRAT
Authors: Samuel Sverker
Cited by: 0
References: 11

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Energy-aware software development is a growing concern, driven by the need to reduce resource consumption and by worries that AI tools trained on large code bases may learn inefficient coding practices. Operating systems play a central role in how hardware resources are managed and powered, yet we still have limited empirical knowledge of how OS developers actually implement energy-saving changes in practice. This thesis examines energy-aware commits in open-source operating systems and provides an analysis of the optimization techniques used at the OS level. Building on a study by Moura et al. of energy-aware commits inopen-source projects, this work modifies their approach and then focuses down to operating systems. We first train a transformer-based text classifier on manually labelled data and use it to identify likely energy-saving commits. We then focus on the commits that modify operating-system code and perform a qualitative analysis of their full context, using open and axial coding to derive and refine themes. Our results show that energy-aware changes are heavily concentrated in kernels, and predominantly written in C. Developers most commonly reduce energy consumption by improving idle behavior and sleep states, avoiding unnecessary wake-ups (for instance through batching or coalescing work), tuning process scheduling and load balancing. Compared to Moura et al., our replication shows a much stronger concentration of energy-aware changes in operating-system code (87% vs. 50%) while retaining a similar dominance of C as the implementation language (76% vs. 81%), underscoring the central role of operating-systems in energy optimization. Overall, the thesis maps out recurring OS-level optimization techniques for energy savings and refines prior work by highlighting the central role of sleep states and scheduling in practical energy-aware operating-system development.

Fields of Study & Subjects

Green IT and Sustainability Energy Efficiency in Computing Parallel Computing and Optimization Techniques

Publication details

Journal / source
Uppsala University Publications (Uppsala University)
Publication date
1 January 2026
Language
EN
Document type
Thesis