Dear June,
The comments from Will in the email exchange you refer to are also valid
here. What analyses (contrasts) are you interested in? It is very likely
that you don't have to estimate a model with so many variance
components. Could you share (in a .mat file) the variables 'I' and
'factor' that enter spm_get_vc.m.
Best regards,
Guillaume.
On 03/11/2019 04:36, June Kang wrote:
> Hello,
>
> I am running a very large (>1260 beta images as input, and 2 x 3 x
> 10(repetition) ) in 2nd Level analysis.
> It seems the process stop before save .mat files, but I found below issue.
>
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1003&L=SPM&D=0&P=540608 <https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.jiscmail.ac.uk%2Fcgi-bin%2Fwebadmin%3FA2%3Dind1003%26L%3DSPM%26D%3D0%26P%3D540608&data=02%7C01%7C%7Cc90caffa6063450ece9908d76018d3f9%7C1faf88fea9984c5b93c9210a11d9a5c2%7C0%7C1%7C637083532068181255&sdata=YMG5oZB423Rr0afN8v7QOEeXzq8n%2BgXwTzrOalWSP94%3D&reserved=0>
>
> Unfortunately, disabling below part make the estimation not possible.
>
> %Vi = Vi(~az);
>
> %dupl = false(size(Vi));
>
> %for cVi = 1:numel(Vi)
> % if ~dupl(cVi)
> % for cVi1 = (cVi+1):numel(Vi)
> % dupl(cVi1) = dupl(cVi1)||full(all(Vi{cVi}(:) == %Vi{cVi1}(:)));
> % end
> % end;
> %end;
> %Vi = Vi(~dupl);
>
> So, I put the tic, toc function and let the code display every cVi.
> It shows me that the process will calculate 8580 replication that takes
> 12 to 22 secs per run.
> Interestingly, the CPU usage is <5% for this calculation.
>
> Of course the design is strange, but at least it didn't stop but need time.
> I am wondering there might be some way to improve the efficiency of this
> calculation.
> Anybody faced this issue and any experience of mod this code?
>
> Bests,
> J.
--
Guillaume Flandin, PhD
Wellcome Centre for Human Neuroimaging
UCL Queen Square Institute of Neurology
London WC1N 3BG
|