My 5-year-old woke me up the other night and it took a while to get back to sleep so I used the time to think about sorting 2D data point on the Parallella (it worked! I got back to sleep at least before the sun came up).
Many algorithms that manipulate two dimensional data need to first sort the data from an un-ordered point cloud into something that is more manageable. The application that I have in mind is DeLaunay Triangulation where each point is connected to it's nearest neighbours .
The parallella seems to be well suited to this type of processing.
I've written up my initial thoughts on a Parallella-friendly algorithm here: . Please comment if you have any ideas, especially if I've missed something.
Thanks,
nick