[Gate-users] Problem with GPU PET example

Liliana Caldeira llcaldeira at gmail.com
Wed Jan 11 17:15:53 CET 2017


Hi Mirco,

The GATE GPU version does not save the source position, but only the
position when it leaves the voxelized phantom, thus why you only see the
faces of the phantom.
If you are really interested in the source positions, then you can change
the GPU code to output that information to a file (although it might be
very large txt file).
In the file GateEmisTomo.cc:

kernel_voxelized_source_b2b<<<grid, threads>>>(gamma1_d, gamma2_d,
activities_d, E,
                                                   phantom_d.size_in_vox,
phantom_d.voxel_size);
    cudaThreadSynchronize();
    printf(" ::   Generation ok\n");
PRINT INFO HERE: gamma1_d.eventID gamma1_d.px gamma1_d.py gamma1_d.pz

Cheers,
Liliana
Forschungszentrum Jülich GmbH


On 10 January 2017 at 14:11, Mirco Heß <mirco.hess at uni-muenster.de> wrote:

> Hi all,
>
> I am trying to run the GPU PET example but am facing some problems. When
> looking at the output, it seems like the XYZ positions of the source are
> accumulated on the faces of the provided phantom/source volume, i.e. the
> bounding box of the 3D image. This is only true for the GPU-based
> simulation. When running the CPU-based simulation everythig seems to be
> fine. I have attached two screenshots to visualize the problem and the
> difference between the two cases. The problem is present in both the ASCII
> and ROOT output.
>
> I already had this problem about two years ago with Gate 7.0 and was not
> able to resolve it.
> Now I am running Gate 7.2 with GPU support activated, on Ubuntu 14.04
> with updated NVIDIA and CUDA drivers.
>
>
> Has anybody been able to successfully run this example?
>  (gate_v7.2/examples/gpumacros/pet/main-gpu.mac)
>
> Any advice and pointers to resolve this issue would be highly appreciated.
> Thanks a lot in advance!
>
>
> Cheers,
> Mirco
>
>
> --
> *Mirco Heß*
> PhD Candidate
> European Institute for Molecular Imaging (EIMI)
>
> [image: _European Institute for Molecular Imaging - EIMI_]
> European Institute for Molecular Imaging - EIMI
> Waldeyerstr. 15
> 48149 Münster
> http://www.uni-muenster.de/eimi
>
> +49 (0)251 83 - 49318 <+49%20251%208349318>
> +49 (0)251 83 - 49313 <+49%20251%208349313> (Fax)
> eMail: mirco.hess at uni-muenster.de
>
> _______________________________________________
> Gate-users mailing list
> Gate-users at lists.opengatecollaboration.org
> http://lists.opengatecollaboration.org/mailman/listinfo/gate-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opengatecollaboration.org/mailman/private/gate-users/attachments/20170111/b96a748e/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: eimi-logo_web.jpg
Type: image/jpeg
Size: 8441 bytes
Desc: not available
URL: <http://lists.opengatecollaboration.org/mailman/private/gate-users/attachments/20170111/b96a748e/attachment.jpg>


More information about the Gate-users mailing list