If the suggestion of Dr. Sunaert does not work then the problem will lie
with the number of files your SGI will allow you to open. I don't know
how to check on an SGI but its probably similar to Sun Solaris machines.
If you type 'limit' (in Solaris) you get a list of limitations on the system
resources currently avalaible to the shell you're in (eg. filesize, stacksize,
etc). When you type 'unlimit' (in Solaris) you increase the size on some, if
not most, of these. The one we seem to be running into problems with is
'descriptors.' Descriptors are the number of file descriptors one is allowed
to open at any one time, and consequently the number of open files thatt are
allowed. If ya don't got a BIG enough descriptors size ya can't open all the
darn image files! Our default is 64, and after 'unlimit' is 1024. This was
still not enough so I got our system admin to change the deafulat to 1024 and
the upper limit to 4096. In Solaris the changes were made in the
/etc/system file. No problem now!
THis solution was given by Martin Stumpf back in Nov. on the mailbase, and is
archived on the SPM webpage.
My thanks to him for his help in solving the problem for us.
Sincerley,
Chris Thomas
Laboratory for Functional Magnetic Resonance Research,
Robarts Research Institute,
London, Ontario, Canada.
>>after fighting with SPM and receiving hints from different
>people (thank you all !), I am still stuck when trying to do a group
>analysis on six subjects, 390 images (normalized with defaults) each.
>Every of the individuals runs through fine by itsself but not
>when in combination with a total group size > 4.
>My procssing includes scaling, temporal smoothing and two confounding
>variables. The same happens regardless of use of 'spmlarge' or the default
>spm 96b. Here are the error messages:
>
>>> ??? Cant map image file (/stim1/NEUMA_exec/sn5010041.img).
>
>Error in ==> /data/sw/spm96/spm_map_vol.mexsg
>Error in ==> /data/sw/spm96/spm_map.m
>On line 33 ==> V = spm_map_vol(Filename,ID);
>
>Error in ==> //nhome/erhard/bin/SPM/spm_fmri_spm_ui.m
>On line 428 ==> for i = 1:q; V(:,i) = spm_map(P(i,:)); end
>
>??? Error while evaluating callback string.
>
>
>These error messages are very reproducible, the processing dies on the
>same image, even though the image is fine by itsself, as individual
>processing proves.
>
>I use a SGI O2 R10000 195 MHz with IRIX 6.3 and Main memory size: 576
>Mbytes. Swap space is 1.27 GB + 180MB.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|