MRIQC running on CamCAN for 24+ hours

NBCLab / power-replication

A replication and extension of Power et al. (2018)

https://www.overleaf.com/read/swgjxcjqytxg

Apache License 2.0

2 stars 0 forks source link

MRIQC running on CamCAN for 24+ hours #7

Closed tsalo closed 3 years ago

tsalo commented 3 years ago

Each subject is currently still running MRIQC, 24 hours after submission. It should only take ~2 hours per subject, so I'm not sure what's going on. The only thing I can think of is that maybe pybids is having trouble with the size of the dataset, since there are 600+ subjects?

tsalo commented 3 years ago

Currently trying an approach where each subject's data and the BIDS top-level files are copied to a new mini-dataset before running MRIQC. I will see if this speeds things up. 🤞

emdupre commented 3 years ago

:crossed_fingers: Good luck !! Do the logs reveal if it's stuck on any particular step ? Or is it just slow generally ?

tsalo commented 3 years ago

I checked the stderr and stdout but didn't look at the mriqc logs. The single-subject dataset version runs fairly well, so I think it just came from the size of the full dataset.

tsalo commented 3 years ago

Things seem to be progressing nicely. 14 failed on one or more runs/images, so I'm re-running them now.

Ultimately, most finished in ~2-2.5 hours.