Hi
this does sound like a memory problem and the solutions probably lies
in tweaking your system - what OS are you running? There've been
numerous email to the list about how to get e.g. cygwin to play nice.
It could very well be that your OS restricts the process size of a
single program to something much less than your physical limits.
Melodic requires less than 1GB for this data set - I just tried on
similarly sized data on my powerbook and all is fine.
cheers
christian
On 21 Dec 2006, at 00:20, Rich Crawford wrote:
> Hi All,
>
> My problem is that melodic stops running at the "Perform variance-
> normalisation ..." step. I was thinking it could be a memory issue
> because when running melodic on a smaller 4d file (120 time points
> vs. the
> full 180) I get an output of components just fine. However, givin I'm
> running with 3 GB of ram and 4 GB of virtual the 180 x 253417 file
> size
> doesn't seem that big relative to others I've seen on posts. CPU
> is a P4
> 3.6 GHz. Please find the full error report attached, and abbreviated
> report below, hope it's helpful.
>
> Many Thanks,
> Rich
>
> ######################################################################
> #### MELODIC INFORMATION
> ######################################################################
>
> Log files:
>
> Melodic Version ln(12) beta
>
> /usr/local/fsl/bin/melodic -i filtered_func_data -o
> c:/cygwin/images/fMRI_FFT/FFT017-
> 1_/ep2d_bold_8Ch/acq75_Analyze/swr/4d+++++.ica -v --nobet --
> bgthreshold=10 -
> -tr=2.5 -d 0 --mmthresh=0.5 --report
> ---------------------------------------------
>
> Melodic results will be in c:/cygwin/images/fMRI_FFT/FFT017-
> 1_/ep2d_bold_8Ch/acq75_Analyze/swr/4d+++++.ica
>
> Reading data file filtered_func_data ... done
> Create mask ... done
> Excluding voxels with constant value
> Data size : 180 x 253417
> Removing mean image ... done
> Estimating data covariance ... done
>
> /usr/local/fsl/bin/avwroi c:/cygwin/images/fMRI_FFT/FFT017-
> 1_/ep2d_bold_8Ch/acq75_Analyze/swr/4d example_func 90 1
>
> /usr/local/fsl/bin/betfunc c:/cygwin/images/fMRI_FFT/FFT017-
> 1_/ep2d_bold_8Ch/acq75_Analyze/swr/4d prefiltered_func_data_bet
>
> /usr/local/fsl/bin/ip prefiltered_func_data_bet filtered_func_data
> 10 -t
> 24.0 -1
> THRESHOLD 87.951200
>
> rm -rf prefiltered_func_data*
>
> /usr/local/fsl/bin/melodic -i filtered_func_data -o
> c:/cygwin/images/fMRI_FFT/FFT017-
> 1_/ep2d_bold_8Ch/acq75_Analyze/swr/4d+++++.ica -v --nobet --
> bgthreshold=10 -
> -tr=2.5 -d 0 --mmthresh=0.5 --report
>
> Melodic Version ln(12) beta
>
> Melodic results will be in c:/cygwin/images/fMRI_FFT/FFT017-
> 1_/ep2d_bold_8Ch/acq75_Analyze/swr/4d+++++.ica
>
> Reading data file filtered_func_data ... done
> Create mask ... done
> Excluding voxels with constant value
> Data size : 180 x 253417
> Removing mean image ... done
> Estimating data covariance ... done
> Perform variance-normalisation ...
> mkdir c:/cygwin/images/fMRI_FFT/FFT017-
> 1_/ep2d_bold_8Ch/acq75_Analyze/swr/4d_150/4d+.ica
>
>
> <fsl_ulzh0J.gz>
--
Christian F. Beckmann
Oxford University Centre for Functional
Magnetic Resonance Imaging of the Brain,
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
Email: [log in to unmask] - http://www.fmrib.ox.ac.uk/~beckmann/
Phone: +44(0)1865 222551 Fax: +44(0)1865 222717
|