Re: Permission For opensource optimus support?
Logically speaking, most IGP units can be considered "dedicated" but they are not.
Industry standards, the technical specifications typically, dictate the definitions outside of the common english dictionary. An Integrated graphics card is not there for sound controller, storage controller, wifi controller either, but it's not dedicated.
The work I've done designing GUI for process automation systems over the years has since changed to third party solutions for the hardware but still generally consider dedicated to have it's own pathway for the information to reach it, without having to go through another unit. They also have their own memory and never borrow system memory. In this industry we have used proprietary, vax/vms, unix, and pc hardware over the years but the graphics processing standards have remained pretty much consistent as within the growth of the technology.
None of the businesses I've helped set up systems for would buy a mother board with a graphics card that is dependent on another gpu to obtain its information, for ANY graphics intensive solution. What little I've been able to dig up on "optimus" technology all point to all graphics data/traffic having to go THROUGH the intel igpu, making the geforce chip nothing more than an extension, or part of the igpu. Without the complete removal of the dependency of the igpu for any form of processing data, the card is not TECHNICALLY a dedicated gpu.
There may be arguments logically, but standard wise, it's a severe deviation from dedicated gpu technology.
Again, it would seem to me to be relatively easy to integrate a function or small set of functions that would permit the igpu to be used for simple through put, since it's already required as such via hardware design. This should be neither labor intensive nor costly. But if Nvidia feels it would be, then they could turn to the open source community for assistance or even full development.