Hi, The preprocessing job is the one that requires more memory ~8GB Then, the slices processing not much ~ 200MB or 300 MB each job and the postprocessing ~ 2GB
Moises On 31 January 2017 at 17:36, Gopalakrishnan, Karthik <gkart...@gatech.edu> wrote: > Thanks Niels and Moises! > > > > Niels, the documentation in the Github repo makes setting up the overlay > look pretty straight-forward. I did take a look at some of the code – it > looks like the fsl_sub script has been altered to use qsub for Torque. Do > you suggest deploying this overlay as-is? Or do you have any specific > change recommendations for these scripts? > > > > Moises, could you please tell me what kind of memory usage I should > anticipate when running this with HCP diffusion data after I set it up? > > > > Thanks a lot! > > > > Regards, > > Karthik > > > > *From:* Niels Bergsland [mailto:npbergsl...@bnac.net] > *Sent:* Tuesday, January 31, 2017 4:19 AM > *To:* Gopalakrishnan, Karthik <gkart...@gatech.edu> > *Cc:* Moises Hernandez Fernandez <mois...@fmrib.ox.ac.uk>; > hcp-users@humanconnectome.org > > *Subject:* Re: [HCP-Users] Running bedpostx on HCP diffusion data > > > > Hi Karthik, > You may find this useful for getting FSL scripts adapted for torque: > > > > https://github.com/octomike/fsl_overlay/ > > > > On Mon, Jan 30, 2017 at 11:30 PM, Gopalakrishnan, Karthik < > gkart...@gatech.edu> wrote: > > Hi Moises, > > > > Thanks for the reply! > > > > I also want to know a couple other things - is Sun Grid Engine (SGE) a > particular requirement to ensure bedpostx internally splits the slices > among all available CPU cores? Or will something like Torque and Maui also > be fine? The system administrator in my college says that we already have > Torque and Maui and so it’ll be much quicker to have that set up for me. > > > > If Torque and Maui will also be fine, could you also share the relevant > changes that we’d have to make to any files/anything else to successfully > and quickly run bedpostx for HCP diffusion data? And on a slightly > different note, what kind of memory usage should I anticipate when I run > this for HCP diffusion data? > > > > Regards, > > Karthik > > > > *From:* Moises Hernandez Fernandez [mailto:mois...@fmrib.ox.ac.uk] > *Sent:* Friday, January 27, 2017 3:37 PM > *To:* Gopalakrishnan, Karthik <gkart...@gatech.edu> > *Cc:* Glasser, Matthew <glass...@wustl.edu>; hcp-users@humanconnectome.org > > > *Subject:* Re: [HCP-Users] Running bedpostx on HCP diffusion data > > > > I mean information inside the mask, (nodif_brain_mask) > > > > Moises. > > > > On 27 January 2017 at 20:17, Gopalakrishnan, Karthik <gkart...@gatech.edu> > wrote: > > Thanks Moises and Matt! Just to clarify, doesn’t HCP diffusion data > (<subject>/T1w/Diffusion/data.nii.gz) have something like 145 slices, not > 90? > > > > Regards, > > Karthik > > > > *From:* Glasser, Matthew [mailto:glass...@wustl.edu] > *Sent:* Thursday, January 26, 2017 5:49 AM > *To:* Moises Hernandez Fernandez <mois...@fmrib.ox.ac.uk>; > Gopalakrishnan, Karthik <gkart...@gatech.edu> > *Cc:* hcp-users@humanconnectome.org > > > *Subject:* Re: [HCP-Users] Running bedpostx on HCP diffusion data > > > > One should use 3 fibers for HCP data however. > > > > Matt. > > > > *From: *<hcp-users-boun...@humanconnectome.org> on behalf of Moises > Hernandez Fernandez <mois...@fmrib.ox.ac.uk> > *Date: *Thursday, January 26, 2017 at 4:44 AM > *To: *"Gopalakrishnan, Karthik" <gkart...@gatech.edu> > *Cc: *"hcp-users@humanconnectome.org" <hcp-users@humanconnectome.org> > *Subject: *Re: [HCP-Users] Running bedpostx on HCP diffusion data > > > > Hi Karthik, > > > > It depends on your processor and on the parameters that you use. > > > > One large slice of HCP diffusion data takes about 4 hours on a single CPU > core (modern processors have several cores) using the default parameters > (2250 MCMC iterations, 2 fibres estimation). > > > > HCP data has about 90 slices with diffusion information, although some of > them are small and are processed in few minutes. > > > > If you have a queue system, such as SGE, you can send each slice to a > different CPU core. > > > > Moises. > > > > > > > > > > On 25 January 2017 at 23:29, Gopalakrishnan, Karthik <gkart...@gatech.edu> > wrote: > > Hi, > > > > I want to run bedpostx on HCP diffusion data – what is the estimated time > for successfully running it for a HCP subject in a regular computer without > a GPU? Also, what are the recommended hardware specs to run bedpostx on HCP > data? > > > > Thanks! > > Karthik > > _______________________________________________ > HCP-Users mailing list > HCP-Users@humanconnectome.org > http://lists.humanconnectome.org/mailman/listinfo/hcp-users > > > > _______________________________________________ > HCP-Users mailing list > HCP-Users@humanconnectome.org > http://lists.humanconnectome.org/mailman/listinfo/hcp-users > > > ------------------------------ > > The materials in this message are private and may contain Protected > Healthcare Information or other information of a sensitive nature. If you > are not the intended recipient, be advised that any unauthorized use, > disclosure, copying or the taking of any action in reliance on the contents > of this information is strictly prohibited. If you have received this email > in error, please immediately notify the sender via telephone or return mail. > > > > _______________________________________________ > HCP-Users mailing list > HCP-Users@humanconnectome.org > http://lists.humanconnectome.org/mailman/listinfo/hcp-users > > > > > > -- > > Niels Bergsland > Integration Director > Buffalo Neuroimaging Analysis Center > 100 High St. Buffalo NY 14203 > npbergsl...@bnac.net > _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users
