CGAL users discussion list

["Sebastien Loriot (GeometryFactory)" <>]

Did you try your program with a model with a size similar to the 
benchmark? If yes, did you observe similar timing?

The mesh you're using is pretty huge. I tried on a 703512 faces mesh and 
it took me 75s of pre-processing for 10 points on my laptop (10m
might not be surprising then depending on your computer).

Also, you might be getting a memory issue (swapping). Did you check that?

Sebastien.


On 07/08/2016 05:43 AM, GaryGu wrote:
> Hi,
>
> I have the a problem of computing surface distance about that the processing
> time is too long.
>
> The input data I used is big. The surface face is about 3427880 and the
> vertices size is 1714008.
>
> Here is what I did.
> I used the library of CGAL 4.8.1 - Triangulated Surface Mesh Shortest Paths.
> First, I load the input data into the structure of polyhedron.
> Then I use following codes to set up the start point and the target point
>
> Surface_mesh_shortest_path shortest_paths(polyhedron);
> vertex_descriptor startVertex = vertices[source];
> vertex_descriptor endVertex = vertices[target];
>
> // add the first vertex as source
> shortest_paths.add_source_point(startVertex);
> // compute the surface distance
> FT startToEnd =
> shortest_paths.shortest_distance_to_source_points(endVertex).first;
>
> However it took too long to process the computation about more than 10
> minutes.
> In addition, my program could work efficient when data size is 50000 points
> about 2~3 seconds processing time.
>
> I saw the benchmark. The biggest dataset used is 17495 points, and it is
> efficient.
> It can process the 187519 queries per second.
>
> So I want to ask that is it normal that processing such big dataset have to
> take more than 10 minutes or even longer ?
> Or is it any solution can speed up the computation ?
>
> Thanks,
> Gary
>
>
>
>
> --
> View this message in context: 
> http://cgal-discuss.949826.n4.nabble.com/The-problem-of-surface-distance-computation-tp4662048.html
> Sent from the cgal-discuss mailing list archive at Nabble.com.