Related Work (Frameworks, Tools, etc)

[lfwa]

Just wanted to add an overview of related work:

• Scaphandre: Track energy consumption of running processes (mostly CPU-based)
• Boagent: Actively measure environmental impact of running applications (tries to include more life cycle phases, relies on Scaphandre)
• CodeCarbon: Track carbon footprint from any kind of compute, not just ML, (also has a pretty dashboard)
• experiment-impact-tracker: Track emissions of compute, focused on ML (no longer updated)
• ML CO2 Calculator: Calculate emissions from ML by inputting hardware used, hours used, provider, and region.

Feel free to expand this list.

[moonstarsky]

谢谢，您的邮件已收到~

[lfwa]

Update with newer works:

• Website Carbon Calculator: Calculates the carbon emission by the network traffic from a given URL
• Google Cloud Region Picker: ool to help you pick a Google Cloud region considering carbon footprint, price and latency
• Carbonifer: Command Line Tool to control carbon emission of your cloud infrastructure
• Tracarbon: Tracks your device's energy consumption and calculates your carbon emissions using your location (depends on CO2signal from ElectricityMaps and therefore requires users to sign up for an API key)
• Cloud Carbon Footprint: Estimates the energy (kilowatt hours) and carbon emissions (metric tons CO2e) of public cloud provider utilization
• CATS: Climate-Aware Task Scheduler. It schedules cluster jobs to minimize predicted carbon intensity of running the process
• Green Algorithms: Web-tool and framework for clusters (see also Green Algorithms for High Performance Computing)