Dear David,
I first thought that you had found the bug with explicit masks that
crept in the latest SPM12 update (r6906) but, no, this is different
problem here.
Would it be possible that several batches are trying to access the same
data concurrently? I know that there are some limitations with our
implementation of memory mapping on Windows platforms. Are you using
(large?) 4D files? While the error message is not very meaningful here,
it means that the input data of your GLM couldn't be accessed for
reading, and this can perhaps happen if another process is already
reading/writing the same file.
Best regards,
Guillaume.
On 09/11/16 18:47, David Hofmann wrote:
> Hi @all,
>
> I get the following error message when trying to run a GLM with SPM12
> version 6685
>
> PM12: spm_est_non_sphericity (v6015) 18:37:14 - 09/11/2016
> ========================================================================
> Chunk 1/27 :
> ...processingFailed 'Model estimation'
> Error using sub2ind (line 29)
> Size vector must have at least 2 elements.
> In file "C:\Program
> Files\MATLAB\R2014a\toolbox\matlab\elmat\sub2ind.m" (???), function
> "sub2ind" at line 29.
> In file "C:\User\SPM\spm12\spm12\@file_array\subsref.m" (v4136),
> function "multifile2mat" at line 153.
> In file "C:\User\SPM\spm12\spm12\@file_array\subsref.m" (v4136),
> function "subfun" at line 98.
> In file "C:\User\SPM\spm12\spm12\@file_array\subsref.m" (v4136),
> function "subsref" at line 65.
> In file "C:\User\SPM\spm12\spm12\@nifti\subsref.m" (v5759), function
> "rec" at line 221.
> In file "C:\User\SPM\spm12\spm12\@nifti\subsref.m" (v5759), function
> "subsref" at line 45.
> In file "C:\User\SPM\spm12\spm12\spm_data_read.m" (v6486), function
> "spm_data_read" at line 62.
> In file "C:\User\SPM\spm12\spm12\spm_est_non_sphericity.m" (v6015),
> function "spm_est_non_sphericity" at line 146.
> In file "C:\User\SPM\spm12\spm12\spm_spm.m" (v6678), function
> "spm_spm" at line 421.
> In file "C:\User\SPM\spm12\spm12\config\spm_run_fmri_est.m" (v5809),
> function "spm_run_fmri_est" at line 33.
> The following modules did not run:
> Failed: Model estimation
>
>
> Strangely, this error only appears when using parfor to run multiple
> matlabbatches, but not when running the batches separately with a normal
> for loop (so the batches are fine). Moreover, it only occurs when I use
> parfor with 15 or more workers, but not with 10 or less workers.
>
> I really have no idea what the problem could be, since running the
> batches one at a time works well
>
> Any help appreciated!
>
> greetings
>
> David
--
Guillaume Flandin, PhD
Wellcome Trust Centre for Neuroimaging
University College London
12 Queen Square
London WC1N 3BG
|