Green Algorithms

Green Algorithms estimates the carbon footprint of computational tasks to quantify greenhouse gas (GHG) emissions from high-performance and large-scale computing.

Key Features:

• Standardized framework: Provides a methodological framework to estimate carbon emissions from computational tasks, enabling consistency across studies and comparative analyses.
• Hardware compatibility: Accounts for diverse hardware configurations when assessing energy use and emissions to support applicability across computing environments.
• Contextual GHG metrics: Produces metrics that contextualize computed emissions relative to broader ecological and societal benchmarks.

Scientific Applications:

• Particle physics simulations: Quantifies carbon emissions associated with computational algorithms used in particle physics simulations.
• Weather forecasting models: Estimates the GHG emissions of weather forecasting computational workloads.
• Natural language processing tasks: Calculates emissions from natural language processing model training and inference.

Methodology:

Considers energy consumption rates and specific hardware configurations, aggregates these data, and calculates the greenhouse gas emissions attributable to computational tasks.