PowerLetrics: Democratizing Energy Metrics for Linux
- Track: Energy: Accelerating the Transition through Open Source
- Room: H.2214
- Day: Sunday
- Start: 12:10
- End: 12:20
- Video only: h2214
- Chat: Join the conversation!
The environmental impact of software and hardware systems has become a pressing concern. While mainstream operating systems like macOS and Windows provide proprietary tools to analyze energy usage, Linux—the cornerstone of open-source computing—lacks comparable capabilities. To address this gap, we present PowerLetrics, an open-source framework enabling detailed, per-process energy footprint analysis on Linux systems.
Key Contributions:
-
Secure and Granular Power Metrics: By leveraging the /proc file system, PowerLetrics delivers real-time power consumption data on a per process/ container level without requiring elevated permissions, ensuring both usability and security.
-
Inspired by macOS's powermetrics: Our solution adapts the power estimation principles of macOS's powermetrics utility to Linux, offering an analogous capability tailored to open systems.
-
Customizable Energy Models: Through the Linux Energy Estimation Engine (L3E), PowerLetrics empowers users to tweak or design bespoke energy models, fostering a flexible and adaptable ecosystem.
-
Open-Source Accessibility: Available under an open-source license (AGPL), PowerLetrics is a community-driven initiative designed for scalability across diverse hardware platforms, from servers to mobile devices. By providing clear, actionable data on energy consumption, we empower the tech community to make informed decisions that favor environmental sustainability, aligning technological advancement with ecological responsibility.
With its modular architecture and minimal performance overhead, PowerLetrics integrates seamlessly into existing Linux environments, supporting both developers and system administrators in optimizing energy efficiency.
The project is accessible via GitHub (https://github.com/green-kernel), and we invite the open-source community to collaborate on enhancing its capabilities. Future directions include automated benchmarking tools for hardware-specific models and support for proprietary hardware components like GPUs.
Speakers
Didi Hoffmann |