Skip to Content.
Sympa Menu

cgal-discuss - [cgal-discuss] squared_distance performance

Subject: CGAL users discussion list

List archive

[cgal-discuss] squared_distance performance


Chronological Thread 
  • From: topotun <>
  • To:
  • Subject: [cgal-discuss] squared_distance performance
  • Date: Wed, 3 Oct 2012 12:47:03 -0700 (PDT)

Hello everyone.

I am trying to get a feel for the performance of the CGAL's AABB-based
distance computation routines (/squared_distance/, /closest_point/,
/closest_point_and_primitive/), using the corresponding tutorial in the
online manual. Currently I am simply loading some .off files found on the
web for the polyhedron model and compute distance to the polyhedron from
random points in space.

What I am seeing is that the /squared_distance /is at least two orders of
magnitude slower than either the /closest_point /or
/closest_point_and_primitive /(e.g., ~32ms versus 0.56ms/0.57ms). This
behavior remains consistent for all .off models I have tried. This seems to
contradict the results posted on the CGAL manual page, where all three of
these functions exhibited approximately the same performance. Furthermore,
calling the /accelerate_distance_queries /after the tree is constructed
actually decreases the performance by about half, thus making me question
its functionality.

The timing is done very simply with the OpenMP’s /omp_get_wtime(), /which
might not be the most accurate method, but does allow me to come up with the
aforementioned questions.

Any advice on this matter will be greatly appreciated.

Denis




--
View this message in context:
http://cgal-discuss.949826.n4.nabble.com/squared-distance-performance-tp4655960.html
Sent from the cgal-discuss mailing list archive at Nabble.com.



Archive powered by MHonArc 2.6.18.

Top of Page