AMD got burned when they did it the other way around in the Kaveri (2014) era, when they bet on unified memory, transparently passing pointers between gpu/cpu

AMD got burned when they did it the other way around in the Kaveri (2014) era, when they bet on unified memory, transparently passing pointers between gpu/cpu with unified virtual memory etc, and hoped that software would start to utilize GPUs more with OpenCL and other vendor neutral APIs providing portability.