Nvidia's latest driver has opengl changes.

[degasus]

Maybe faster bounding box because of ARB_shader_atomic_counter_ops. And there is a chance to get destAlpha + logicOps (kibrys shadows) with ARB_fragment_shader_interlock. But both are not implemented right now.

Feel free to ask Tinob about using ARB_parallel_shader_compile for his hacky async fifo.