Looking at our working sets we find:

• ray[direction, maxdistance]
• all the objects

In order to simplify each threads job we now have

• objects in tree-leaf
• tree-leaf
• tree

With the multiple ideas that the tree will:

• reduce object discovery time
• tree-leaves will fit better into memory than loading the entire object set at once

Am I correct in my explanation of the logic behind this datastructure?