Hi FreeSurfer users, is it possible to run FreeSurfer on parallel processors? We would like to run mri_glmfit-sim in parallel to reduce the run time. Have you any experience running FreeSurfer in parallel, or would it be relatively easy to reconfigure?
Any thoughts would be greatly appreciated.
Thanks in advance,
Brendan Ansell Melbourne University
mri_glmfit-sim can be run in parallel. See the --help. Also, I've run a big batch of simulations and put the results here ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/mult-comp-cor.tar.gz
this is for whole hemisphere on fsaverage. If that is what you are doing, then it should be fine. You just need to figure out which file is appropriate (see the README file in the directory it produces).
doug
Brendan Robert Ansell wrote:
Hi FreeSurfer users, is it possible to run FreeSurfer on parallel processors? We would like to run mri_glmfit-sim in parallel to reduce the run time. Have you any experience running FreeSurfer in parallel, or would it be relatively easy to reconfigure?
Any thoughts would be greatly appreciated.
Thanks in advance,
Brendan Ansell Melbourne University
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Dear FreeSurfer users,
I have attempted to run mc-full for 20 iterations on 18 cores to test the time per iteration. The process is taking more than 14 hours.
It times out at Poll 4725 job 1
How long does each iteration take for mc-full?
We have 8 hemispheres to run, which will total 80,000 iterations.
How many CPUs are required to run mc-full (threshold: 2, iterations: 10000) in a reasonable time period, say 2 weeks?
Any information on the steps involved in the completion of the polling process and subsequent steps would be appreciated.
Thanks in advance,
Brendan Ansell
MNC, Melbourne University
Hi Brendan, do you really need the full simulation? If you can use the mc-z, then we have a new version of the simulation script that will run in a few minutes using pre-cached simulation tables. You can use this as long as you use a threshold that is either 1.3, 2.0, 2.3, 3.0, 3.3, or 4. This will be standard with version 5. Until we have version 5 is ready, you can download and install the new simulation script and cached tables with:
cd $FREESURFER_HOME/bin mv mri_glmfit-sim mri_glmfit-sim.backup
Get new version from here ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/mri_glmfit-sim
cd $FREESURFER_HOME/average Get this tar file ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/mult-comp-cor.tar.gz tar xvfz mult-comp-cor.tar.gz
See mri_glmfit-sess --help for more info.
doug
Brendan Robert Ansell wrote:
Dear FreeSurfer users,
I have attempted to run mc-full for 20 iterations on 18 cores to test the time per iteration. The process is taking more than 14 hours.
It times out at Poll 4725 job 1
How long does each iteration take for mc-full?
We have 8 hemispheres to run, which will total 80,000 iterations.
How many CPUs are required to run mc-full (threshold: 2, iterations: 10000) in a reasonable time period, say 2 weeks?
Any information on the steps involved in the completion of the polling process and subsequent steps would be appreciated.
Thanks in advance,
Brendan Ansell
MNC, Melbourne University
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Brendan Robert Ansell -
Douglas N Greve