tbb and opengl scenegraph

I’m considering the idea of create a multicore (with parallel_for or so)scenegraph for my rendering system.
I ask if is possible since i think that is the rendering section the bottleneck(the gpu) and if this can gain performance or no.
I think is something related to the Gustafson’s/Amdahl’s law: if i have to do computational expensive calculations in my scenegraph i can gain performance , else if the scenegraph is used for a low expensive calculations i not gains nothing, rather i can get lower performance for the Threads overhead .

But i think to use the scenegraph for do all in my engine , then i can encapsulate and launch heavy parallel functions in this, this can grow the work and grow the performance.

There is some examples of engines or tutorial with the use of tbb(of intel) and SceneGraph in general?

ps. I do these experiments only for understand well the parallel programming for fun, I not think to creates a commercial engine, I am aware of this