Dear Kevin,
I was about to email you that I couldn't reproduce the crash you report
on that platform with large files but just before writing this email I
did a very last check on another random location of such a file and it
crashed in a similar way - which was quite a disappointment.
My hope is that it is a Mac only thing (but I'm 'happy' to be
contradicted) - would you have the possibility to try out the same thing
on the brand new 64bit Mac Intel Matlab R2009b:
http://www.mathworks.com/products/new_products/latest_features.html
This might require a recompilation of MEX files as explained here:
http://en.wikibooks.org/wiki/SPM/Installation_on_64bit_Mac_OS_(Intel)
If not, according to the kind of analysis you want to perform on your
data, as a temporary work-around, I would suggest that you split your
data in chunks (along trials) and Vladimir might be able to help there.
Best regards,
Guillaume.
Kevin Guise wrote:
> Hello,
>
> I believe I ran into a problem with file2mat.mexmaci. I'll try to describe
> what happened as accurately as possible.
>
> I used spm8 to to apply a time-frequency transformation to my data and
> generated a rather large file (~4.6 GB). When I attempt to view the
> transformed data or average it, matlab gave an error and forced me to quit
> (see below). I loaded the tf*.mat file into the workspace and looked at
> D.data.y, which was of size channel x freq x time x trial. I tried to
> extract individual trials (e.g. D.data.y(channel,:,:,i)) and found that
> matlab would crash without fail when I tried to extract the 107th trial. I
> tried this again using D.data.y(channel,1,1,i) and again matlab crashed when
> i reached 107. I was using Matlab 7.6.0 on macbook with a dual core 2 GHz
> processor and 2 GB of ram. I tried this again using a different (yet also
> very large) dataset in Matlab 7.8.0 on a mac pro with 2 x 2.8 GHz quad core
> processors and 2 Gb of ram and got a similar error (although I was able to
> extract data from up to the 117th trial). The error message from Matlab
> 7.6.0 is pasted below.
>
> Best
> Kevin
--
Guillaume Flandin, PhD
Wellcome Trust Centre for Neuroimaging
University College London
12 Queen Square
London WC1N 3BG
|