Shower thought: What good is an excellent ranked PC avatar if it has a grab-pass/screen-grab on it?

For each material slot you have, you have its respective shader, however, since shaders are not included into performance rankings, and only for fallbacks. They are not accounted for, otherwise.

Shaders take GPU time, so if a shader from a material slot took an extra 1 ms to render at 90fps, it would decrease the framerate to 82fps. A shader fallback system in the case a shader is too expensive to run in a material would assist in having a more effective performance rank medium.

Shader GPU time could be calculated in the SDK by simply having a hidden camera render each shader/material on a sphere in a different/hidden scene. This method would also prove performant as materials could be evaluated for their frame time. GPU time limits (e.g. 0.02 ms or 0.2 ms) could be implemented on a per-shader or per-avatar basis to help better calculate how optimized an avatar is or how optimized its material slots/shaders are.

A more functional system would to test for certain shader features such as tessellation, grab-pass (screen-space reflections/refractions) and raymarching, and put those in their respective performance ranks for shaders specifically.