Hi,
The amount of time is going to depend on several things - mainly the
size of the data set (both the number of voxels and number of time points)
and the size of physical memory and swap space. If you run top while
melodic is going, how much memory does it say it is using, and is it
causing a lot of swapping (virtual memory usage) ?
It isn't unusual for it to take several hours for either a large data set
or when there is substantial swapping involved.
As for the images - the example_func is a single 3D volume, whereas the
filtered_func_data is the whole 4D data set, hence you expect the latter
to be much bigger. With around 68 time points you'd expect the file
sizes to be in the ratio that you've got here.
All the best,
Mark
Jack Grinband wrote:
> Steve,
> I'm running MELODIC and it's taking several hours to finish on a SunBlade
> (900MHz). Is this typical or am I doing something wrong? One thing that seems
> strange is that example_func.img is around 800kb, while filtered_func_data.img,
> which I'm running through MELODIC, is 54Mb. The filtered_func_data.img was the
> output of a FEAT preprocessing. thanks,
>
> jack
>
>
> On Tue, 12 Nov 2002 17:20:48 +0000, Stephen Smith <[log in to unmask]>
> wrote:
>
>
>>Hi - there is nothing in FSL that explicitly parallelises either within or
>>across command-line calls. Apart from the obvious answer of starting
>>different separate tasks in parallel, I don't have a general suggestion
>>here. Which tool in particular were you interested in speeding up?
>>
>> Thanks, Steve.
>
|