CGAL users discussion list

[schrodingersnewcat <>]

Hello all,

I have a program that is executing a large number of transformations on
Nef_polyhedra.  As the program executes the memory allocated to it grows
extremely fast.  

The goal of the code is to create a grid which tracks material properties at
each point, so I have to know what materials constitute the voxel
surrounding each grid point (CGAL makes this very easy by the way).  Since
this is intended to be part of a program that I will eventually run on a
large parallel computer, with total execution time around 24 hours, I can't
afford to exhaust all of my available memory in the first twenty or so
minutes.

Code that follows exhibits the problem.  I had to use ps to get memory info
since I can't get valgrind to work with Nef_polyhedra.


#include <CGAL/basic.h>
#include <CGAL/Gmpq.h>
#include <CGAL/Exact_predicates_exact_constructions_kernel.h>
#include <CGAL/Extended_cartesian.h>
#include <CGAL/Polyhedron_3.h>
#include <CGAL/IO/Polyhedron_iostream.h>
#include <CGAL/Nef_polyhedron_3.h>
#include 
#include 

typedef CGAL::Exact_predicates_exact_constructions_kernel Kernel;
typedef Kernel::FT CoordType;
typedef CGAL::Polyhedron_3 Polyhedron;
typedef CGAL::Nef_polyhedron_3 Nef_polyhedron;
typedef Kernel::Point_3 Point;
typedef Kernel::Vector_3 Vector;
typedef Kernel::Aff_transformation_3 Aff_transformation;

int main(int argc, char **argv)
{
        std::string name(argv[0]);

        Point p1(0,0,0);
        Point p2(1,0,0);
        Point p3(0,1,0);
        Point p4(0,0,1);

        Polyhedron P;
        P.make_tetrahedron(p1,p2,p3,p4);
        std::cout << P;

        std::cout << "Make NP" << std::endl;
        Nef_polyhedron NP(P);

        std::cout << "Make Transformation" << std::endl;
        Aff_transformation affTran(CGAL::IDENTITY);

        int count = 10000;
        if(argc >= 2)
        {
                count = atoi(argv[1]);
        }

        std::cout << "ps -eo pid -o comm -o rss -o vsz | grep " << name <<
std::endl;
        std::cin.ignore();
        for(int i = 0; i < count; ++i)
        {
                NP.transform(affTran);
        }
        std::cout << "ps -eo pid -o comm -o rss -o vsz | grep " << name <<
std::endl;
        std::cin.ignore();

        NP.clear();
        std::cout << "ps -eo pid -o comm -o rss -o vsz | grep " << name <<
std::endl;
        std::cin.ignore();

        return 0;
}

--
View this message in context: 
http://cgal-discuss.949826.n4.nabble.com/Memory-Management-with-Nef-polyhedra-tp3404480p3404480.html
Sent from the cgal-discuss mailing list archive at Nabble.com.