CGAL users discussion list

[Ding <>]

Dear all,

There are some statements saying that CGAL can perform 3D DT of 1M random
points under 20 sec (13 or 16 sec). However, I tried the example provided in
the CGAL manual, 3D DT of only 0.125M points took 52.59 sec on a 2.13 GHz Core
2 Duo computer with 2G memory (Mac OSX 10.5). So I'm just wondering if any
options need to be set for better performance. Thank you! 

The testing code is as follows, which is almost the same as provided in the
manual.

/******************* code begin *******************/
#include <CGAL/Exact_predicates_inexact_constructions_kernel.h>
#include <CGAL/Delaunay_triangulation_3.h>
#include <CGAL/Triangulation_hierarchy_3.h>

#include <cassert>
#include <vector>

typedef CGAL::Exact_predicates_inexact_constructions_kernel K;

typedef CGAL::Triangulation_vertex_base_3<K>             Vb;
typedef CGAL::Triangulation_hierarchy_vertex_base_3<Vb>  Vbh;
typedef CGAL::Triangulation_data_structure_3<Vbh>        Tds;
typedef CGAL::Delaunay_triangulation_3<K,Tds>            Dt;
typedef CGAL::Triangulation_hierarchy_3<Dt>              Dh;

typedef Dh::Finite_vertices_iterator Finite_vertices_iterator;
typedef Dh::Vertex_handle            Vertex_handle;
typedef Dh::Point                    Point;

#include <boost/progress.hpp>

int main()
{
  // generating points on a grid.
  std::vector<Point> P;

  for (int z=0 ; z<50 ; z++)
    for (int y=0 ; y<50 ; y++)
      for (int x=0 ; x<50 ; x++)
          P.push_back(Point(x,y,z));

  Dh T;

  {
      boost::progress_timer timer(std::cout);
      T.insert (P.begin(), P.end());
  }


  return 0;
}

/******************* code end *******************/


regards,
df