They're almost there but I don't think it'll be as supported or easy to hack on with. mostly. I believe you'll only have it through APUs - which a few are getting into the same watts profile. Perf via datasheet looked similiar - I'm not sure how they compare in terms of global memory bandwidth, link speed, and latency. Both are integrated close to the GPU, but I suspect AMD integrated a bit better. OTOH probably less flaws with nvidia on the software side, and these can easily be considered performance related.
