Print

Print


Hi Joseph

What is the resolution of your data and how many directions do you have? And how do you call bedpostx?

Bedpostx is not multi-threaded. It can only use multiple cores via SGE or similar cluster scheduler. If you do not have that, every slice will be processed serially on a single core, so for high resolution data and depending on how many MCMC iterations/model type you have asked this can take very long. Which is why we developed a GPU version, as Moises pointed out.

Cheers
Stam 

On 10 Jul 2018, at 20:29, Krieger, Donald N. <[log in to unmask]> wrote:

Hopefully one of the FSL people will weight in on this.
I'm interested to hear what they say.

When I saw something comparable, I just assumed there was something wrong with the scan which was tripping FSL up.

Best - Don
 

From: FSL - FMRIB's Software Library <[log in to unmask]> on behalf of VA Research <[log in to unmask]>
Sent: Tuesday, July 10, 2018 1:58:25 PM
To: [log in to unmask]
Subject: Re: [FSL] BEDPOSTX - Still Running After 5 Days
 
Every slice seems to be taking 3+ hours to complete.

I have plenty of cores(12) and ram (64gb) available.

looking the activity monitor  xfibres 156MB of ram

On Tue, Jul 10, 2018 at 3:24 AM, Krieger, Donald N. <[log in to unmask]> wrote:
Hi Joseph,
 
If it’s still running take a look at how much memory it is using, both resident and virtual.
I have seen bedpostx jobs run interminably on a single core when I had more than two jobs running on a multi-core device which shares 8 GBytes of memory.
I a lot of virtual memory is in use, the job is likely going to the disk often for memory space which slows things down by more than an order of magnitude.
I also have had jobs which ran interminably for no reason that I could identify but I presumed that there was a problem with the scan quality which was causing the problem.
 
I hope this is helpful.
 
Best – Don 
 
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Moises Hernandez
Sent: Monday, July 9, 2018 10:45 PM
To: [log in to unmask]
Subject: Re: [FSL] BEDPOSTX - Still Running After 5 Days
 
Hi Joseph,
 
It is possible that something did not work correctly,
but bedpostx can take 1 or 2 days if you use a single CPU core,
or even more if your data is high resolution.
 
Do you get any message at all?
 
An alternative is to use an NVIDIA gpu:
 
Moises.
 
On 9 July 2018 at 21:39, VA Research <n.n.di.methyl.tryptamine1@gmail.com> wrote:
Hello Experts, 
 
Any ideas, why it would be taking this long to complete one DTI image?
 
Thanks so much,
 
Joseph Veliz
 

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
 
 

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1


To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1



To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1


To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1



To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1