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
 Image exception:#99::out of memory
 New Topic  Reply to Topic
 Printer Friendly
Author Previous Topic Topic Next Topic  

Adria
New Member

19 Posts

Posted - Feb 07 2012 :  4:34:29 PM  Show Profile  Reply with Quote
Hi,

I am running step 2 (fslvbm_2_template) on my data and I keep on getting this error in my output file for some images:
+ /usr/local/packages/fsl-4.1.8/bin/fast -R 0.3 -H 0.1 PATB_FSE480_struc_brain
Image Exception : #99 :: Out of memory
terminate called after throwing an instance of 'RBD_COMMON::BaseException'
fslvbm2a: line 215: 22480 Aborted /usr/local/packages/fsl-4.1.8/bin/fast -R 0.3 -H 0.1 PATB_FSE480_struc_brain
+ /usr/local/packages/fsl-4.1.8/bin/immv PATB_FSE480_struc_brain_pve_1 PATB_FSE480_struc_GM


and then later in the file:

** ERROR (nifti_image_read): failed to find header file for 'PATB_FSE480_struc_GM'
** ERROR: nifti_image_open(PATB_FSE480_struc_GM): bad header info
ERROR: failed to open file PATB_FSE480_struc_GM
ERROR: Could not open image PATB_FSE480_struc_GM
Image Exception : #22 :: Failed to read volume PATB_FSE480_struc_GM
terminate called after throwing an instance of 'RBD_COMMON::BaseException'
/usr/local/packages/fsl-4.1.8/bin/fsl_reg: line 124: 21260 Aborted ${FSLDIR}/bin/flirt -ref $REFERENCE -in $INPUT $INMASK -omat ${I2R}.mat $flirtopts
** ERROR (nifti_image_read): failed to find header file for 'PATB_FSE480_struc_GM'
** ERROR: nifti_image_open(PATB_FSE480_struc_GM): bad header info
ERROR: failed to open file PATB_FSE480_struc_GM
ERROR: Could not open image PATB_FSE480_struc_GM
Image Exception : #22 :: Failed to read volume PATB_FSE480_struc_GM
terminate called after throwing an instance of 'RBD_COMMON::BaseException'

What could be the problem?

petty
BIAC Staff

USA
453 Posts

Posted - Feb 07 2012 :  5:21:41 PM  Show Profile  Reply with Quote
1st one seems to indicate there isn't enough memory (RAM) to write the data

2nd one indicates that the image being read is incomplete.
Go to Top of Page

Adria
New Member

19 Posts

Posted - Feb 10 2012 :  10:12:38 AM  Show Profile  Reply with Quote
Is the memory issue from the cluster or the program?
Go to Top of Page

syam.gadde
BIAC Staff

USA
421 Posts

Posted - Feb 10 2012 :  10:17:10 AM  Show Profile  Reply with Quote
You will need to request more memory for these types of jobs. So when you submit you will need to add the following option to qsub:

-l h_vmem=6G

where 6G means 6 Gigabytes of memory. Default for our cluster is 4G. If that doesn't work, you can try higher numbers.
Go to Top of Page

Adria
New Member

19 Posts

Posted - Feb 10 2012 :  10:28:57 AM  Show Profile  Reply with Quote
Thank you! So I can tag that at the end of the command "qsub -v EXPERIMENT= etc."?
Go to Top of Page

syam.gadde
BIAC Staff

USA
421 Posts

Posted - Feb 10 2012 :  10:40:04 AM  Show Profile  Reply with Quote
Yes, at the end or beginning of the qsub command is fine:
qsub -l h_vmem=6G ...other options...
Go to Top of Page

Adria
New Member

19 Posts

Posted - Feb 24 2012 :  10:03:06 AM  Show Profile  Reply with Quote
What is the maximum amount of memory I can submit a qsub with?
Go to Top of Page

syam.gadde
BIAC Staff

USA
421 Posts

Posted - Feb 24 2012 :  10:10:12 AM  Show Profile  Reply with Quote
There is no theoretical maximum, though practically, the nodes in the cluster have only a finite amount of memory. Some nodes have a maximum of 30 gigabytes or so available to cluster jobs. Others have more. Most users will never need more than 30 gigabytes. Generally, trial and error will show you the minimum you need to request. If you request more than you need, that's OK, but it also potentially increases the amount of time your job will need to wait for a node with enough resources to allocate to your job.
Go to Top of Page

petty
BIAC Staff

USA
453 Posts

Posted - Feb 24 2012 :  11:29:09 AM  Show Profile  Reply with Quote
the maximum is 92G ... there are currently only two nodes that can handle this.

the max is dictated by how much physical RAM is installed in each node, since the RAM cannot be shared across the cluster.

Edited by - petty on Feb 24 2012 11:31:18 AM
Go to Top of Page
  Previous Topic 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 0.48 seconds. Snitz Forums 2000