If that's your goal, why don't you benchmark them instead of asking us things we don't know?
What will you be using the results of the benchmark for? Once you do know which Kd-Tree is fastest, what will you do with it?
What have you done so far? Do you have the kd-tree benchmark running? Did you get JK's tree working with that? Did you get mine working with that? If you want to benchmark them, I think it makes more sense for you to ask for help with that, rather than asking how fast the various trees are (which nobody knows since we never bothered to benchmark all of the latest trees). I'm actually kind of curious about this too, so I'm willing to offer some help if you'll explain what you need done and what you've tried already.