would be nice to see using an entry level GeForce for gameworks plus PhysX workload like the old hack.
Why would it need to be implemented by game Devs the OS and DX12 should automatically take care of it, wasn't that the whole point so game devs could just get on with the job of coding without having to worry about what GPU / GPU's were in a system
I did read a while ago that AMD were supposedly working on a variation of this for combining AMD Zen APU's and compatible GCN cards. This workload sharing would work across all games including DirectX 11, not just DirectX 12 games where the developer decides to implement it. AMD Zen supposedly has two desktop flavours, most notably being a 4-core, 8 thread APU, and an 8-core, 16 thread CPU. The CPU having more cores because it utilises the die space for the graphics for CPU cores. If the rumour is right, you could have a very powerful gaming system with just a Zen APU and a mid level AMD discrete 14 nm GPU.
my understanding is all the Automated things done by dx11 where removed to reduce the the overhead, and now devs actual have to optimize thing correctly for there needs?
If this is true how would you be able to use g-sync and freesync together . Interesting times ahead for sure