AMD’s x86 Nolan APU will be Fabricated on the 28nm Process not 20nm
I had the opportunity to talk to an Industry Insider with knowledge of AMD's supply chain and we discussed quite a lot of interesting topics. Naturally I will not be revealing his identity or company for obvious reasons. We covered alot of AMD's roadmap possibilities, some of which will be different from what other publications (and even we ) have reported on occasion. While the information at the time of revelation is correct, keep in mind that there is nothing stopping AMD from changing their roadmaps in the future as they have done countless times already. I have split the talk into four pieces. This is the first.
AMD's Nolan APU is actually on 28nm - Industry Insider
Lets start with Nolan APU. Nolan is AMD's x86 project and were mentioned as 20nm (on official slides) however, our source says that is not actually true. According to him Nolan APU is on the 28nm process while Amur will actually end being on a lower node (20nm). That is an extremely interesting revelation to say the least. He went into further specifics about Nolan APU as well, that it will only be a 'mild' update to the Beema/Mullins platform and be FP4 BGA packaged (same as Carrizo). He further went on to state that if AMD employed a smaller-than-28nm node it will most probably be used for Amur or the successor to Amur. To summarize, it would appear that Nolan stays at 28nm while Amur shifts to a lower process node in all likelihood.
Nolan and Amur is AMD's special platform - an attempt to break into the low power mobility market and compete against Intel's Cherry Trail. While switching to 20nm would have given AMD a much needed advantage, from the looks of it, AMD will be staying with 28nm with this particular SKU. Nolan should be based on the Puma+ architecture and will be virtually identical to its ARM counterpart apart from the fact that it is x86 and not ARM. Both have a release time stamp of Q3 2015. Amur is said to support both Android and Linux while Nolan will support both Android and Windows.
This also raises questions on which process AMD might end up utilizing, for a low power die, the 28nm lithographic processes of both TSMC and GloFo will suffice, and there would be negligible difference between the two anyways. And thats not all, our source also mentioned that AMD will stick to 28nm process for most of its offerings throughout 2015 and maybe even 2016 - something that comes as a considerable surprise to us. Stay tuned for the remaining writeups of our discussion on AMD (and also Intel), they will be up soon.