Kornel Kisielewicz <kornel.kisielewicz@gmail.com> wrote:
One example is object hierarchy via oct or kd trees.
Would you like to implement these spatial containers? I think this would be a very useful contribution. Ideally your implementation would be generic, i.e. usable with fixed-size point types or the variable-size uBLAS types; at the end of the project we would then be able to evaluate the benefit of removing the size field. 2D & 3D please. FWIW here's my take on the issues being discussed in this thread: - Not every platform has a GPU on to which work can be offloaded. - Although I doubt that wasting a word on a size field has much effect on speed, I would want to avoid it when dealing with large collections of points because of the memory overhead. - I have using uBLAS for some simple 2D matrix transformation and found it a little bit more difficult than I expected; a special-purpose API, even if implemented as a thin wrapper around uBLAS or something else, would have the benefit of simplifying things for the end user. Regards, Phil.