At 5/30/2006 03:25 PM, Doug Greve wrote:
It was running out of memory. I've copied a new version into that ftp site. It might do a little better with the memory managment.
First off -- thanks for the quick response with new versions of the program to try, and obviously the addition of the diagnostic output has really helped.
Response with this version was basically the same as before... "Cannot allocate memory" (log excerpt below).
FWIW here's more info:
Linux box running FC4. "Only" half gig of memory, plus one gig swap space.
While running these last couple of times I watched memory allocation in KInfoCenter's Memory monitor pane.
With no FS script running:
Physical memory: 432 Meg free Swap: 933 Meg free.
With mri_remove_neck running: Physical memory: fluctuates between 5 Meg and 40 Meg free. Swap memory: Generally about 500 Meg free.
So, clearly mri_remove_neck sucks up a lot of memory, and one path-of-least-resistance is a trip to Fry's for some more memory :-).
Any other suggestions are welcome, of course.
Graham
####################################### #-------------------------------------- #@# Remove Neck Tue May 30 14:39:56 PDT 2006
mri_remove_neck -radius 25 nu.mgz transforms/talairach.m3z /brain-bin/fs_20060528/centos4/freesurfer/average/RB_all_2006-02-15.gca nu_noneck.mgz
erasing everything more than 25 mm from possible brain reading atlas '/brain-bin/fs_20060528/centos4/freesurfer/average/RB_all_2006-02-15.gca'... setting gca type = Normal gca type reading input volume 'nu.mgz'... reading transform 'transforms/talairach.m3z'... gcam->type = vox reading labels out of gcam file... removing structures at least 25 mm from brain... 11495184 nonbrain voxels erased writing output to nu_noneck.mgz... mghWrite(nu_noneck.mgz, -1): could not open file ERROR: opening pipe gzip -f -c > nu_noneck.mgz, errno = 12 Cannot allocate memory Linux gwlinux.ucsd.edu 2.6.11-1.1369_FC4 #1 Thu Jun 2 22:55:56 EDT 2005 i686 i686 i386 GNU/Linux
recon-all exited with ERRORS at Tue May 30 14:54:52 PDT 2006