You can simply improve performance by at least 4 or 5 times by using arrays / matrices instead of relatively expensive object calls.
This went from 7 fps to 30 fps comfortably on my machine with no functional change.
It is a little less intuitive to read, however I believe the performance benefit is significantly improved.
I will also experiment with GPU calculations now that the atom definitions are all number matrices. It makes it relatively trivial to add this functionality now.
You can simply improve performance by at least 4 or 5 times by using arrays / matrices instead of relatively expensive object calls.
This went from 7 fps to 30 fps comfortably on my machine with no functional change.
It is a little less intuitive to read, however I believe the performance benefit is significantly improved.
I will also experiment with GPU calculations now that the atom definitions are all number matrices. It makes it relatively trivial to add this functionality now.