When the number of point clouds is more than 400K,why does my code slow down? #235
Description
Hello,
thank you very much for the invention of the Nanoflann library, which solved many of our problems.
I'm having some problems. When the number of point clouds is less than 400K, when using knnSearch to find the 10 closest points of all points, Nanoflann+OpenMP is faster than PCL+openMP, but when the number of point clouds is greater than 400K, Nanoflann+OpenMP is slower or almost the same as PCL. Interestingly, if you search by radiusSearch, nanoflann is always faster than pcl::KdTreeFLANN. I would like to ask why this happens.
And please,how to solve it?