[Mrtrix-discussion] tckgen stuck

[Robert Smith]

Hi Basile

Thanks for the very informative and helpful report; it made the error much
easier to locate and fix.

The offending commit
<https://github.com/jdtournier/mrtrix3/commit/c8e3955811fc05f3406d353028dbf1a5>
actually relates to code deep within the multi-threading framework.
In every other use case this change will have had no effect whatsoever, but
the dynamic seeding framework is far from a 'typical' use case (it's a thing
of beauty
<https://github.com/jdtournier/mrtrix3/blob/master/src/dwi/tractography/tracking/exec.h#L78>,
actually). Following the change in the multi-threading code, the threads
didn't know when it was time to terminate, and so the command hung at 99%.

I've just pushed a code commit that fixes the problem.

A couple of other quickies since I never get tired of hearing myself type:

   - GitHub (which hosts the MRtrix3 code) has an issue tracking system
   <https://github.com/jdtournier/mrtrix3/issues?state=open>. If you ever
   experience behaviour in the new release that you are confident is a bug, or
   have a great idea for a feature, people are more than welcome to post there
   if they choose; it also means that you get credit for anything you report /
   suggest.
   - I couldn't help noticing the use of compressed NIfTI images... Given I
   went to the effort of coding it up, I might as well mention that the new
   version supports .mif.gz images. Solving all your image compression needs
   without the constraints of a fixed-width header and ambiguous transform :-P

Good <your relevant timezone> everyone

Rob


--

*Robert Smith, Ph.D*
Research Officer, Imaging Division

The Florey Institute of Neuroscience and Mental Health
Melbourne Brain Centre - Austin Campus
245 Burgundy Street
Heidelberg Vic 3084
Ph: +61 3 9035 7128
Fax: +61 3 9035 7301
www.florey.edu.au


On Wed, Jun 25, 2014 at 5:51 PM, basile pinsard <basile.pinsard at gmail.com>
wrote:

>  Hi MRtrix experts,
>
>  running tckgen on a cluster and the nodes are centos 6.5, with this
> specific set of option:
>
>  $ tckgen -force -act 5tt.nii.gz -info -maxlength 250 -nthreads 12
> -number 10000 -seed_dynamic fod.nii.gz -step 0.2 fod.nii.gz track.tck
> ....
> tckgen [INFO]: step size = 0.2 mm
> tckgen [INFO]: maximum deviation angle = 14.4 deg
> tckgen [INFO]: minimum radius of curvature = 1.25 mm
> tckgen [INFO]: iFOD2 internal step size = 0.0666667 mm
> tckgen [INFO]: opening image "fod.nii.gz"...
> tckgen [INFO]: rejection sampling will use 7 directions with a ratio of
> 1.0728 (predicted number of samples per step = 8.0391)
>    61056 generated,     9984 selected    [ 99%]
>
>  it runs until it almost reachs the total count of fibers requested then
> get stuck there using all the cpu resources allocated with nthreads, it
> only happens when using seed_dynamic.
> A slightly earlier version of mrtrix3 (
> 52617b9f3a4d58f50d0e802d2609bc7fb503da21 ) does not have this bug. Looking
> at the diff shows mainly changes related to strides but the fod have been
> generated with the latest version.
>
>  thanks for you help.
>
>  cheers
>
> --
>  Basile Pinsard
>
> *PhD candidate *
> Laboratoire d'Imagerie Biomédicale, UMR S 1146 / UMR 7371, Sorbonne
> Universités, UPMC, INSERM, CNRS
> Unité de Neuroimagerie Fonctionnelle, CRIUGM, Université de Montréal
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.nitrc.org/pipermail/mrtrix-discussion/attachments/20140625/aaead2e9/attachment.html>