So, I believe I have isolated the problem. The failure happens during the mri_concat phase of make_average_volumes. The seg-subj#.mgh temp files that are being concatenated are 65 MB each and I can successfully run 59 subjects, but for 60 subjects and up, mri_concat exits immediately with the error message as written below. I have 4GB of memory, and 65*59 is less than 4 GB so that works; add any more subjects and the process is shut down. It appears that mri_concat checks to see how much memory will be needed to hold all those .mgh files in active use, then checks the machine to see how much physical memory is available. I can easily set up a larger swap file to handle the full 175*65 that is needed (it will be insanely slow, but that's better than nothing). However, I can't get FreeSurfer to utilize the swap space I already have available.

I watched the whole make_average_volumes process play out on my resource monitor (I imagine that's what watching NASCAR is like--Come on, just crash already!) and while it came very close to using all 4GB--leaving a paltry 19 MB free at most points--it didn't touch the swap at any point during the run. This is not the case with Matlab and other programs, which will greedily eat up physical memory then seek out the swap. While I appreciate that the FreeSurfer programs are set up to be judicious consumers of resources, is there some sort of environmental variable in FreeSurfer or line of code in mri_concat that I can modify that would allow the programs to reach out and use the swap space?

Thanks,
Jim
---------
Jim Porter
Graduate Student
Clinical Science & Psychopathology Research
University of Minnesota


James N. Porter wrote:
Juergen--Have you had successful resolution to this problem?

Nick--I'm attempting to average 175 subjects, but make_average_volume returns the same error as Juergen reports.
    > Allocing output
    > MRIalloc(256,256,256): could not allocate 262144 bytes for 15547th slice

I'm running freesurfer-Linux-rh9-stable-pub-v4.0.1 on SUSE Linux 9.3 2.6.11.4-21.17-smp x86_64, with a pair of Pentium 3.20 GHz processors, 4 GB of physical memory and 8 GB of swap memory. My limits are also quite liberal.
    > cputime         unlimited
    > filesize        unlimited
    > datasize        unlimited
    > stacksize       unlimited
    > coredumpsize    0 kbytes
    > memoryuse       unlimited
    > vmemoryuse      unlimited
    > descriptors     1024
    > memorylocked    32 kbytes
    > maxproc         28660

I've tried to also use the alternative script posted by Doug Greve (http://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg06641.html). That also returns the same error.

Any suggestions?

Many thanks,
---------
Jim Porter
Graduate Student
Clinical Science & Psychopathology Research
University of Minnesota


Nick Schmansky wrote: 
Juergen,

Edit the 'make_average_subject' script removing (or commenting-out) the
line 'limit descriptors unlimited'.  You may have to do the same for
'make_average_surface' and 'make_average_volume'.  Although in v4.0,
these should already be commented-out.  

Nick

On Mon, 2008-01-28 at 16:29 +0100, Juergen Haenggi wrote:
  
Dear FS experts

I tried to run make_average_subject in the past with 92 subjects and got the
following error: Cannot allocate memory. This is now on the to-do list of
the FS website. Then I bought some additional RAM and changed from OS X
tiger to leopard. Now I tried it again and get this error:

[Macintosh-3:/Applications/freesurfer/subjects] juergenhaenggi% limit
descriptors 4096
[Macintosh-3:/Applications/freesurfer/subjects] juergenhaenggi% limit
cputime      unlimited
filesize     unlimited
datasize     6144 kbytes
stacksize    8192 kbytes
coredumpsize 0 kbytes
memoryuse    unlimited
descriptors  4096 
memorylocked unlimited
maxproc      266 
[Macintosh-3:/Applications/freesurfer/subjects] juergenhaenggi%
make_average_subject --subjects SNPs_Vp*
limit: descriptors: Can't remove limit (Invalid argument)
[Macintosh-3:/Applications/freesurfer/subjects] juergenhaenggi%

I set the descriptor limit to 4096, but the make_average_subject command
will not start anymore.

Any idea what is wrong here?

Thanks in advance
Best regards
Juergen

-----------------------------------------------------------
Juergen Haenggi
Ph.D. (Dr. des.)
Division of Neuropsychology
Institute of Psychology
University of Zurich
Binzmuehlestrasse 14, PO Box 25
8050 Zurich, Switzerland
0041 44 635 73 97 (phone office)
0041 76 445 86 84 (phone mobile)
0041 44 635 74 09 (fax office)
BIN 4.D.04 (office room number)
j.haenggi@psychologie.uzh.ch (email)
http://www.psychologie.uzh.ch/neuropsy/ (website)
http://www.juergenhaenggi.ch (private website)
-----------------------------------------------------------


_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer