Duke-UNC Brain Imaging and Analysis Center
BIAC Forums | Profile | Register | Active Topics | Members | Search | FAQ
Username:
Password:
Save Password   Forgot your Password?
 All Forums
 Support Forums
 Cluster Support
 Installing/updating packages
 New Topic  Reply to Topic
 Printer Friendly
Author  Topic Next Topic  

mle25
Starting Member

2 Posts

Posted - Oct 05 2016 :  5:55:28 PM  Show Profile  Reply with Quote
Currently the afni version installed on Hugin is from 2011. A lot has changed since this version and I would like to update it for processing use in batch scripts and interactive nodes. What is the best way for doing this without the required sudo access needed?

petty
BIAC Staff

USA
449 Posts

Posted - Oct 05 2016 :  9:01:54 PM  Show Profile  Reply with Quote

Generally, you'd just need to ask for it to be updated by sending an email to biac-help

Otherwise, if there are packages you just wanted to test, you could download your own version to an experiment folder and execute from there. If they were useful packages that others might want, we can install it to the OS
Go to Top of Page

petty
BIAC Staff

USA
449 Posts

Posted - Oct 05 2016 :  9:25:32 PM  Show Profile  Reply with Quote
Also, i updated to the latest version.

the older versions are still in /usr/local/packages with the date/version included in the afni folder name
Go to Top of Page

mle25
Starting Member

2 Posts

Posted - Oct 06 2016 :  4:19:02 PM  Show Profile  Reply with Quote
Thank you.

Another related question, Some of the afni commands and other packages allow for parallel computing to speed up a large job to run on multiple processors. I have tried to find an option like "qrsh -l h_vmem=10G bash -li" for requesting extra cpus so that I can run a parallel job without overloading a node. Is there a way to do this?
Go to Top of Page

petty
BIAC Staff

USA
449 Posts

Posted - Oct 06 2016 :  4:33:38 PM  Show Profile  Reply with Quote
the "-l h_vmem" represents a specific request ... it works because virtual memory is set-up as a consumable. the only other configured consumable is "slots" .. .meaning how many slots your job occupies.

this is only going to be relevant for submitted jobs ( qsub ), not interactive jobs ( qrsh ).

If you had a job you could submit, that you knew would use 10 CPUs, you could submit that in a script with something like "qsub -v EXPERIMENT=This.01 -l h_vmem=10G,slots=10 script.sh"

Alternatively, if its something that you can parallelize yourself ( i.e.: iterations, loops, etc ) ... you could just submit 10 jobs, one for each iteration, which would only occupy 1 slot. You're much better off creating scripts, submitting those at batch jobs, than just interactively running commands.

here's a good overview:
http://bioinformatics.mdc-berlin.de/intro2UnixandSGE/sun_grid_engine_for_beginners/README.html

Edited by - petty on Oct 06 2016 4:45:52 PM
Go to Top of Page

petty
BIAC Staff

USA
449 Posts

Posted - Oct 06 2016 :  4:43:18 PM  Show Profile  Reply with Quote
Also, there are two parallel environments set-up ( mpi and ompi ), so there's also the possibility that afni install, which is openmp, already knows how to utilize that. I've, personally, never done anything with the parallel environments
Go to Top of Page
   Topic Next Topic  
 New Topic  Reply to Topic
 Printer Friendly
Jump To:
BIAC Forums © 2000-2010 Brain Imaging and Analysis Center Go To Top Of Page
This page was generated in 0.06 seconds. Snitz Forums 2000