You also had the whole thing AMD seemed to be attempting with their first gen APU design, shifting the floating point workload over to GPGPUs rather than use a

You also had the whole thing AMD seemed to be attempting with their first gen APU design, shifting the floating point workload over to GPGPUs rather than use a dedicated floating point unit.