Carbon - a lightweight tool for estimating the carbon emissions of HPC jobs
File(s)
Presentation
Author(s)
Coker, Jack
Type
Presentation
Abstract
This presentation introduces a lightweight tool for estimating the carbon emissions of HPC jobs. It requires only job resource usage from the workload scheduler and a minimal cluster configuration describing hardware power characteristics. From these inputs, energy consumption is estimated and converted to carbon emissions using region-specific intensity data. It does not depend on node-level monitoring or vendor-specific tools, making it straightforward to deploy across HPC systems. This non-intrusive tool helps researchers and support teams understand the environmental impact of computing, encouraging more sustainable practices in computational research.
This presentation was given at Computing Insight UK 2025, on 4/12/25.
This presentation was given at Computing Insight UK 2025, on 4/12/25.
Date Issued
2025-12-04
Citation
2025
Copyright Statement
© 2025 The Author(s).
Subjects
Carbon Emissions
High-performance Computing
Scientific Computing
Sustainability
Sustainable Computing
Coverage Spatial
Manchester Central, UK