Neuroimaging Pipeline Carbon Tracker Toolboxes

OHBM SEA-SIG > Neuroimaging Pipeline Carbon Tracker Toolboxes

Neuroimaging Pipeline Carbon Tracker Toolboxes

How to track the carbon footprint of your computational pipeline

There are many tools available to help facilitate green computing, in neuroimaging research and beyond. Below, we provide a list of tools that can help you both measure and reduce your footprint, some of which have been developed by SEA-SIG members!

Measuring impact

  • Green Algorithms – This online calculator allows users to extract estimates of carbon emissions for a given computing job, without requiring any software installation. Users will need access to information including the run time of a job, and the type of processor used
  • GA4HPC – The ‘Green Algorithms 4 High Performance Computing’ tool is a SLURM-based dashboard which makes use of log files to estimate the carbon footprint of jobs submitted via an HPC cluster
  • Calc_Carbon – A server-side carbon tracking tool developed by SEA-SIG members, built for use on a Sun Grid Engine (SGE) system. Again, this tool makes use of HPC logs to estimate carbon emissions
  • CodeCarbon – An ‘embedded’ carbon tracking tool which can be coded directly into other software packages. For instance, CodeCarbon can be accessed while running fMRIPrep, using the –track-carbon and –country-code flags.
  • Carbontracker – Another embedded carbon tracking tool, built to predict the carbon footprint of deep learning models
  • Custom BIDS-app tracker – This tool uses a wrapper module that tracks any containerized BIDS-app pipeline. See this tutorial for tracking MRIQC compute costs. Give this a try to monitor the carbon footprint of your own containerized code! 

Reducing impact

  • fMRIPrepCleanup – Storing ‘junk’ files unnecessarily can contribute to energy needed to back up data on servers, and can increase the rate at which new servers need to be manufactured and procured. This tool automatically identifies data generated by fMRIPrep that is not of use to the user, and deletes it. This tool should be used with caution after reading the available documentation.
  • SPMCleanup – This tool performs the same function as fMRIPrepCleanup, but for preprocessing and analysis files generated by SPM.
  • Climate Aware Task Scheduler (CATS) – This scheduling tool uses live carbon intensity data for the UK to schedule computing tasks to run at times of low carbon intensity. In doing so, researchers can reduce the carbon footprint of their computing without using any less energy.