--resource-profiler with MRIQC

Software Support
1

Summary of what happened:

We are running MRIQC with a singularity container on an HPC. MRIQC runs successfully, but we would like to profile the resources that MRIQC uses in order to optimize the cores and memory that we are asking for, like in this page. We used the flag --resource-monitor to do this, but are not sure if it worked. If it did, where would the corresponding gantt chart be? Or is there something else/additional that we should do to generate the gantt chart?

Command used (and if a helper script was used, a link to the helper script or the command generated):

I’m using a singularity --cleanenv command. I’m not pasting the full command to avoid paths and usernames, but the mriqc options were: mriqc_reports participant --verbose-reports --ica --fft-spikes-detector --resource-monitor --nprocs 4 --participant_label xxxx
mriqc_reports group --resource-monitor

Version:

22.0.6

Environment (Docker, Singularity, custom installation):

Singularity

Data formatted according to a validatable standard? Please provide the output of the validator:

Data have passed the bids validator.

2

Hi @chrysa,

If your HPC uses SLURM as your jobs scheduler, you can run seff $JOBID to see how much memory/run time was used compared to what was requested.

Best,
Steven

3

Thanks @Steven! We did use seff but that gives only an approximate value. What we would like is to see is which parts of the program use more memory and cores (and how many of them) like it’s shown the documentation page with the resource profiler.