Subject: CGAL users discussion list
List archive
- From: Scherrer Daniel <>
- To: "" <>
- Subject: [cgal-discuss] dD Delaunay Triangulation, data dependent performance
- Date: Wed, 7 Oct 2015 14:02:58 +0200
- Accept-language: en-US, de-CH
- Acceptlanguage: en-US, de-CH
- Authentication-results: mail3-smtp-sop.national.inria.fr; spf=None ; spf=Pass ; spf=None
- Ironport-phdr: 9a23:N6z3qR0O7XxTSFHBsmDT+DRfVm0co7zxezQtwd8ZsegRK/ad9pjvdHbS+e9qxAeQG96Lt7Qb0KGK7+jJYi8p39WoiDg6aptCVhsI2409vjcLJ4q7M3D9N+PgdCcgHc5PBxdP9nC/NlVJSo6lPwWB6kO74TNaIBjjLw09fr2zQd6OyZvsnLzus7ToICx2xxOFKYtoKxu3qQiD/uI3uqBFbpgL9x3Sv3FTcP5Xz247bXianhL7+9vitMU7q3cY6Lod8JsKWqrze+E0TKdTES89G2Ez/szi8xfZB0Pb7XQVViAakwFDHhPexBD8RJb49CXg4LlTwi6faPb9Tb0uQzm4p4tiTh7ljG9TGzc/8WzTzPd7jK9Buh+7jwd02MjfYNfGZ7JFYqrBcIZCFiJ6VcFLWnkZDw==
Hello
I experience huge differences in performance when triangulating in 6D. I
tested the following:
const int D = 6;
typedef CGAL::Epick_d<CGAL::Dimension_tag<D>> K;
typedef CGAL::Delaunay_triangulation<K> T;
for (int i = 0; i < 512; ++i){
points.push_back(T::Point((i & 0x3), (i & 0xc) >> 2, (i &
0x30) >> 4,
(i & 0x40) >> 6, (i & 0x80) >> 7, (i & 0x100) >> 8));
}
T t(D);
t.insert(points.begin(), points.end());
This takes about 12s on my PC.
Then I tried the same but scaled the positions by 10 like this:
...
for (int i = 0; i < 512; ++i){
points.push_back(T::Point((i & 0x3) * 10, ((i & 0xc) >> 2) *
10, ((i & 0x30) >> 4) * 10,
((i & 0x40) >> 6) * 10, ((i & 0x80) >> 7) * 10, ((i &
0x100) >> 8) * 10));
}
And this takes more than 1 minute. The scaling itself does not seem to be the
problem. I tried different data sets which performed better or worse for no
obvious reason. I observed factors in computing time up to 30.
Any idea what the triangulation performance could depend on?
Thanks,
Daniel
- [cgal-discuss] dD Delaunay Triangulation, data dependent performance, Scherrer Daniel, 10/07/2015
- Re: [cgal-discuss] dD Delaunay Triangulation, data dependent performance, Marc Glisse, 10/07/2015
- RE: [cgal-discuss] dD Delaunay Triangulation, data dependent performance, Scherrer Daniel, 10/08/2015
- RE: [cgal-discuss] dD Delaunay Triangulation, data dependent performance, Marc Glisse, 10/08/2015
- RE: [cgal-discuss] dD Delaunay Triangulation, data dependent performance, Scherrer Daniel, 10/08/2015
- Re: [cgal-discuss] dD Delaunay Triangulation, data dependent performance, Marc Glisse, 10/07/2015
Archive powered by MHonArc 2.6.18.