Print

Print
Hi - you should just be able to change the setups in $FSLDIR/bin/fsl_sub (and/or your SGE setup) so that the estimated timings for FSL programs match to your different queues (with their own timing limits) better - this should be straightforward for your SGE admin person to sort out.

Cheers.



On 22 May 2011, at 23:28, Benjamin Matthew Deen wrote:

Hi,

I've been running into issues running certain FEAT analyses on the Sun Grid Engine.  In particular, I'm doing preprocessing of relatively high-resolution (2x2x2mm) functional data.  The analysis work fine if I don't use SGE, but on the grid engine the analyses are cut off before producing filtered_func_data (no errors are produced in the log, although there is an empty error file).  How far the analyses get seems to depend on various factors, such as the number of preprocessing steps and current load on the grid.  I believe the problem is that fsl_sub isn't requesting enough time when submitting the preproc job.  However, I'm not sure how to change this without going in and editing FSL scripts (such as fsl_exec.tcl), which is not really an option as this copy of FSL is being used by a number of researchers.  Any idea how I might get around this?  Thanks for any help,

Ben


---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------