CGAL users discussion list

[Michael Bieri <>]

Hi again,

Thanks Sebastian for your answer. 

I finally managed to make it work. I'll describe what I did and I also have some questions about it:

- The cgal_create_cmake_script and cgal_create_CMakeLists do not create any TBB-related code, at all, never. Is that correct?

- So I had to merge the TBB-related stuff from your above linked CMakeLists into the CMakeLists of my project.

- Compiling TBB is easy, BUT it does not offer any installation routine. So I had to specify all TBB_* variables in my project with ccmake or -D... manually. Or does CGAL/TBB offer anything to automatically detect those variables?

- Once it compiled, the meshing will run with a reasonable number of threads right away. Or should I configure something manually?

Best regards,

Michael

On Thu, Aug 17, 2017 at 9:02 AM, Sebastien Loriot (GeometryFactory) <> wrote:

On 08/16/2017 12:17 PM, Michael Bieri wrote:

Hi all

The CGAL documentation just says that I have to enable the parallel tag,
like:

typedef CGAL::Parallel_tag Concurrency_tag;
typedef
CGAL::Mesh_triangulation_3<Mesh_domain,CGAL::Default,Concurrency_tag>::type
Tr;

Surprisingly, it compiled right after that modification, even though I
neither set any link flags for the Intel TBB nor did I find any traces
that I already have Intel TBB on my system (Mac Book Pro) at all.

If you copy-paste a cmake script from CGAL it might be automatically
done. I know this a proposal to issue a warning when the user ask
for parallel support while it is not available but it was not
implemented IIRC. One way to check is to verify that
CGAL_LINKED_WITH_TBB is defined.
Otherwise have a look at this cmake script for example:
https://github.com/CGAL/cgal/blob/master/Mesh_3/examples/Mesh_3/CMakeLists.txt

What exactly do I need to do to use parallel meshing? How can I control
the number of threads? Is there an example?

You should refer to the Intel TBB documentation.
Something like this should work:

{
  tbb::task_scheduler_init tsi(6); // 6 threads
  run_parallel_algo1(); // Uses 6 threads
}

Sebastien.

Thanks a lot and best regards,
Michael

--
You are currently subscribed to cgal-discuss.
To unsubscribe or access the archives, go to
https://sympa.inria.fr/sympa/info/cgal-discuss