[Camino-users] sfpeaks

[Daniel Alexander]

There are some guidelines on setting those parameters in the sfpeaks man page.  Almost certainly you can reduce the run time for q-ball, because the ODFs are very smooth.

Danny

On 14 Dec 2011, at 20:19, Yan Jin wrote:

> Hi Mara and Kiran,
> 
> Thank you so much for your kindly replies! Okay, the default value for
> -density is 1000. When I reduced it to 100, I got it done within an
> hour or so with the pig brain data (I believe that all the commands I
> typed were right since I just copied and pasted from the website.).
> But three hours is not possible for my computer with the default
> setting. Our data set is 128x128x55. I used density=300. I left it run
> and went to bed last night. When I got up this morning, it was done.
> So I think that it got done within 10 hours or so. My computer has a
> single QuadCore CPU (Intel  i7 @ 3.40 GHz with 16 GB RAM). My
> questions are:
> 
> 1. Based on your experience, what typical value do you use for
> -density? Is 1000 necessary?
> 
> 2. Also, there are some other parameters that you can tune, such as
> -numpds, -searchradius, -pointset. Do they affect the running time as
> much as -density do? Do pointset overrides density? What is the
> default value for -pointset?
> 
> Thanks,
> Yan
> 
> On Wed, Dec 14, 2011 at 8:43 AM, Cercignani, Mara
> <m.cercignani at ucl.ac.uk> wrote:
>> hello,
>> 
>> That depends a lot on what type of computer you're using...anyway I can tell you that in my experience (and on the computer I use which has a double QuadCore processor E5630)  a dataset of 96x96x60 takes about 20 hours - so I guess the pig brain data, which is 128x128x10, should take about 1/4 of that time, i.e. 5 hours.
>> 
>> Mara
>> 
> On Wed, Dec 14, 2011 at 1:38 AM, Seunarine, Kiran <k.seunarine at ucl.ac.uk> wrote:
>> Hi Yan,
>> 
>> I would expect sfpeaks to finish within about 1-3 hours (certainly no longer than a day).  As far as I can see, the sfpeaks command you used looks correct.  Please send the list of commands you used to generate pigBrainODFs_SH4_PDs.Bdouble and I'll try and work out why this step is taking so long.
>> 
>> Best regards,
>> 
>> Kiran
>>> ________________________________________
>> From: camino-users-bounces at www.nitrc.org [camino-users-bounces at www.nitrc.org] on behalf of Yan Jin [yjinz at umich.edu]
>> Sent: 14 December 2011 01:53
>> To: camino-users at www.nitrc.org
>> Subject: [Camino-users] sfpeaks
>> 
>> Hi,
>> 
>> I am new to Camino... I wanted to use sfpeaks to process our q-ball
>> imaging data; however, it seems to take extremely long time to run it.
>> In fact, I have never seen it finished. I used the example in
>> 
>> http://web4.cs.ucl.ac.uk/research/medic/camino/pmwiki/pmwiki.php?n=Tutorials.QBallTutorial
>> 
>> and type:
>> 
>> sfpeaks -inputmodel sh -order 4 -numpds 3 < pigBrainODFs_SH4.Bdouble >
>> pigBrainODFs_SH4_PDs.Bdouble
>> 
>> It runs forever... Does anyone know how long it takes to run such a
>> simple data set? I downloaded the latest camino version r1018.
>> 
>> Thanks,
>> Yan
>> _______________________________________________
>> Camino-users mailing list
>> Camino-users at www.nitrc.org
>> http://www.nitrc.org/mailman/listinfo/camino-users
>> 
>> 
> _______________________________________________
> Camino-users mailing list
> Camino-users at www.nitrc.org
> http://www.nitrc.org/mailman/listinfo/camino-users