No it doesn’t, but it’s not about that. It’s about seeing the difference in geom count (Pstats is not needed for that). Given the fact the geom count is where it should be when objects are in the graph but hidden, and the geom count is reduced correctly when objects are detached from render,
speaks for itself.
If hidden objects were not being sent to the GPU, then the geom count would reflect that. Otherwise, if hidden objects are not being sent to the GPU and the geom count says otherwise, then something is wrong somewhere within the engine.
Ok…the benefit of the doubt…
Even if hidden objects are not going to the GPU, detaching them when they are not needed still reduce geom count, and that is better…which is what my thread was about.
I can see your point of view; Panda reports geom count of hidden objects, but does not send those objects down the graphics pipeline… That could be the way P3D does it.
That could be true… but why did I get a performance increase?
The point is, detaching is better for reducing geom count in the scene.
Understanding this has greatly reduced my geom count. I’ve gone from 700+ geoms to 200+. That’s strong evidence. I’m loving the extra performance I’m getting from the change. 
Keep an eye out for some new screenshots of my project; it’s amazing what little changes can do for a P3D app.
PS,
I guess the title of my thread was a little missleading.