Fmriprep taking longer than expected to complete

fmriprep
#1

Hello,

I’m running version 1.5.0rc2 on a batch of subjects, and am noticing that my jobs are taking ~21 hours to complete, even though I ran freesurfer’s recon-all beforehand and that data is in $bids_root_dir/derivatives/freesurfer. My command is:

unset PYTHONPATH; singularity run -B /N/dcwan/projects/irf/templateflow:/opt/templateflow /N/dcwan/projects/irf/containers/fmriprep-1.5.0rc2.simg \
		$bids_root_dir $bids_root_dir/derivatives						\
		participant														\
	    --skip-bids-validation 											\
		--participant-label $s 											\
		--fs-license-file $FREESURFER_HOME/license.txt 					\
		--ignore slicetiming 											\
		--use-aroma 													\
		--aroma-melodic-dimensionality 100 								\
		--output-spaces T1w fsnative fsaverage MNI152NLin6Asym:res-2	\
		--nthread 8												\
		--fd-spike-threshold 0.2										\
		--dvars-spike-threshold 5 										\
		--stop-on-first-crash 											\
		--resource-monitor 												\
		--low-mem 														\
		--mem_mb 45000 												\
		--use-plugin $bids_root_dir/fmriprep_plugin_${s}.yml 			\
		--verbose 														\
		-w $bids_root_dir/derivatives

I realize that requesting multiple output-spaces and aroma increases time, but I was wondering if this amount of time to completion seems reasonable, especially given that freesurfer was run prior?

Furthermore, in the logs I see things like this, so I’m unsure if freesurfer is being run again for some reason.

190826-16:23:34,904 nipype.interface INFO:
	 recon-all complete : Not running
190826-16:23:35,331 nipype.workflow INFO:
	 [LegacyMultiProc] Running 2 tasks, and 0 jobs ready. Free memory (GB): 43.63/48.83, Free processors: 0/8.
                     Currently running:
                       * fmriprep_wf.single_subject_1000_wf.anat_preproc_wf.surface_recon_wf.autorecon1

Thanks for the help!