Hi Xiaochu,
unfortunately, unless your lab has got a SGE system,
that will not work in parallel on your server. Either
you go for serial, then you will have to wait for a
few days or alternatively, if you're familiar with
scripting, you could also split your data in 16 bits
and look in the fslvbm3 script what you should do to
make it work. I would definitely recommend the first
option!
Cheers,
Gwen
--- "Zhang, Xiaochu (NIH/NIDA) [F]"
<[log in to unmask]> a crit :
> Thank you very much! In our lab, there is a server
> with 16 CPUs (the brand is HP). I always open a
> terminal in my computer and do some analysis.
> However, my computer was rebooted suddenly for some
> reason. Therefore, I have to re-open a new terminal
> and re-type the "fslvbm_3" again. Is your means that
> the "fslvbm_3" was restart? And the time was before
> reboot is no useful any more?
> Thanks, again!
> xiaochu
> -----Original Message-----
> From: Gwenaëlle DOUAUD
> [mailto:[log in to unmask]]
> Sent: 2007年10月11日 12:30
> To: [log in to unmask]
> Subject: [FSL] RE : Re: [FSL] RE : Re: [FSL] RE :
> [FSL] hi and for VBM problem.
>
> If your lab has indeed a cluster (see your IT guys),
> you could always restart it, it would only overwrite
> what has been already created. I guess you can stop
> it
> by a simple ctrl c in the shell where you've typed
> the
> fslvbm3 command. But I would suggest you wait a bit,
> as it should not be too long to give you the results
> now.
>
> Gwenaelle
>
> --- "Zhang, Xiaochu (NIH/NIDA) [F]"
> <[log in to unmask]> a écrit :
>
> > Thank you very much!
> > So my question is whether I am able to interrupt
> the
> > program when it is running (For it took a long
> time,
> > sometimes it can not be avoid. ) and restart it
> > again.
> > Does it influence the final result?
> > Thanks, again!
> >
> > -----Original Message-----
> > From: Gwenaëlle DOUAUD
> > [mailto:[log in to unmask]]
> > Sent: 2007年10月11日 10:36
> > To: [log in to unmask]
> > Subject: [FSL] RE : Re: [FSL] RE : [FSL] hi and
> for
> > VBM problem.
> >
> > Hi,
> >
> > sorry, I should have been more specific... It can
> be
> > run in parallel if your lab has got a SGE system
> or
> > something like that (sorry I'm not very good at
> > these
> > things) and then you send your job to a bunch of
> > computers involved in the cluster. It should not
> > take
> > much longer anyway.
> > Hopefully, we should get a better and much quicker
> > non-linear registration (FNIRT) soonish and that
> > will
> > be included in the next FSL4.1 patch.
> >
> > Cheers,
> > Gwenaelle
> >
> >
> > --- "Zhang, Xiaochu (NIH/NIDA) [F]"
> > <[log in to unmask]> a écrit :
> >
> > > Thank you very much for speedy response!
> > > I think the first question has been resolved.
> For
> > 2
> > > hours per subject, it will take about 100 hours
> > > (about 4 days) to finish the whole 48 subjects.
> I
> > > need wait two days more.
> > > For you said it can be run in parallel and the
> > > computer in our lab have multiple CPUs. So, can
> I
> > > open multiple terminals and run the script in
> each
> > > one to shorten the whole time?
> > > Thanks again!
> > >
> > > -----Original Message-----
> > > From: Gwenaëlle DOUAUD
> > > [mailto:[log in to unmask]]
> > > Sent:
> 2007å¹´10Ã¦Å“Ë 11æ—¥
> 9:39
> > > To: [log in to unmask]
> > > Subject: [FSL] RE : [FSL] hi and for VBM
> problem.
> > >
> > > Hi,
> > >
> > > I'm not sure about your first question. The
> > > non-linear
> > > registration can take a while, let's say at
> least
> > 2
> > > hours per subject.
> > > We have done the things so that you can run the
> > jobs
> > > in parallel if your lab has got a cluster to do
> > so.
> > > If
> > > this is not the case, yes, it can be long, and
> > much
> > > longer than fslvbm2 for which you have hopefully
> > > chosen the recommended affine option with -a.
> > > Just check that your 4D image GM_mod_merg.nii.gz
> > in
> > > your stats directory has been created by the
> most
> > > recent fslvbm3 you ran, and if so, just abort
> the
> > > first fslvbm3 command.
> > >
> > > Hope this helps,
> > > Gwenaelle
> > >
> > >
> > > --- "Zhang, Xiaochu (NIH/NIDA) [F]"
> > > <[log in to unmask]> a écrit :
> > >
> > > > Hi,
> > > >
> > > > I'm working on the VBM data. My sample
> included
> > 48
> > > > subjects (24 patients
> > > > and 24 controls). I have two questions about
> the
> > > > "fslvbm_3_proc".
> > > >
> > > > The first one, I ran it 2 days ago. However,
> up
> > to
> > > > now, it have NOT
> > > > finished. I checked the program is not died.
> > > > However, when I run
> > > > "fslvbm_1" and "fslvbm_2", it took no more
> than
> > 4
> > > > hours. Is it normal
> > > > phenomenon?
> > > >
> > > > The second one, about 1 day ago, my computer
> was
> > > > rebooted by power
> > > > problem. After reboot, I re-run the
> "fslvbm_3".
> > It
> > > > looks like OK. Need I
> > > > to deleted all files and restart from
> > "fslvbm_1"?
> > >
> > > >
> > > > Thank you very much!
> > > >
> > > > Xiaochu Zhang Ph.D
> > > >
> > > > Visiting Research Fellow
> > > >
> > > > NIH/NIDA-IRP
> > > >
> > > > 5500 Nathan Shock Drive
> > > >
> > > > Baltimore MD 21224
> > > >
> > > >
> > > >
> > > > Tel: (410) - 550 - 1440 ext. 434
> > > >
> > > >
> > > >
> > > >
> > >
> > >
> > >
> > >
> > >
> >
>
_____________________________________________________________________________
> > >
> > > Ne gardez plus qu'une seule adresse mail !
> Copiez
> > > vos mails vers Yahoo! Mail
> > >
> >
> >
> >
> >
> >
>
_____________________________________________________________________________
> >
> > Ne gardez plus qu'une seule adresse mail ! Copiez
> > vos mails vers Yahoo! Mail
> >
>
>
>
>
>
_____________________________________________________________________________
>
> Ne gardez plus qu'une seule adresse mail ! Copiez
> vos mails vers Yahoo! Mail
>
_____________________________________________________________________________
Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail
|