the z pre-pass is useful as a general technique to reduce fragment call
and improve performance for complex shading or middle-aged graphic card,
at the same time the pass is building occlusion culling.

You might be able to get better performance for special use by skipping it indeed,
but large scene with complex shading/lighting will usually get improved.

Different 3d cards can react differently (on some, like iPhone/iPad pre z-pass to reduce fragment call is bad because the card already does it internally as a strategy).

I got 210 FPS in demo ManualUse,
and after I modify MStandardRenderer.cpp to skip Z pre-pass processing, I got 310 FPS.

Query is expensive in OpenGL I think.
This is useful to reduce draw, but the checking process also takes time.
Or I'm missing something?

Thank you.

(Sorry for English)