> Why not static allocation, but sharing memory between the greedy chunks of code that can't run parallel to each other?

That's what I wanted to say with my comment actually.