juanrga :
h2323 :
juanrga :
During the conference Rory Read mentioned servers, notebooks, tablets, phones... and even medical equipment and casino machines, but didn't mention HEDT. I am not surprised.
Tablets, phones? he said AMD will not swim against the current and go for those low margin highly competitive areas (tablets, phones). The absence of him saying anything about zen really except it's name and node means nothing. He did talk about high margins in that space though (cutting edge processes), which depending on how you take it might be a nod to large, powerful, and expensive zen chips.
Which is your point?
h2323 :
juanrga :
AMD want to the a leader in ARM architecture, not in x86. Of course, Rory knows that Zen will not be faster than Skylake.
He never said that, he never implied that, there is no evidence for this, on the skylake or zen side. I understand that this is your stance but that is all it is, your opinion. It is clear that even if Zen was very fast AMD would not be the x86 leader, they have lost way to much market. I recognize your opinion but everything skylake and zen available does not support your theory, it's truly anyone's guess, sklylake and zen.
Never said what? Never implied what? there is no evidence for what?
You are clearly implying that Zen will not be a high-end desktop part (that is yet to be seen), by taking what Rory Read said out of context to support your point (you have a theory of what Zen will be, stated here and on semi many times based on little to zero information).
Yes he did bring up Tablets and Phones, implying that they will not enter that market any time soon.
Rory Read
"Of course, I think mobility you can get there through semi-custom. But for us to continue to win in that space that march on the low power curve is key. Mark Papermaster did a talk about how we're going to lower our power 25 times over the next x years and that is all going to position us better for that space. But in the mean time, don't go after bad business. If it makes sense, we want that business, but we're not in a position where we could wrap $10 bills around it or if someone is doing $20, God bless them, but I can't -- I'm not going there. It doesn’t make sense"
But nothing in his statements about cutting edge parts Zen/K12 implies that they will not do high end desktop. All he said regarding that is that the margins have to great to cover the yield costs.
Rory Read
"Now, will we move down to Fins? Yes, absolutely, and our next generation products go there and as we introduce them in 16. But we want to do that as you’re catching the cost curve and the yield curve at the right place so that it makes sense for us."
"If, from my perspective you’re running a balanced business on that space. So, I like how Mark Papermaster, Lisa, Keller, those guys are all looking at it, and there is a whole set of people that are really doing the work to say when does the cost curve come over, where do I want to match that into the roadmap,"
"We’ll do test chips and do early parts, but we’re not going to move volume until it’s ready to produce the right kind of return, because that whole Moore’s law is kind of flattening out. You saw it from 28 to 16, 14. That was a very different curve in terms of the benefit, the shrink, and the cost. And it was out of whack at least for the first 24 months and that’s unusual. And so, we are going to be very thoughtful about how we attack that. But make no mistake, we will continue to move down the curve, and we’ll continue to leverage it in the products where it makes sense."
"So, industrial, medical, casino gaming, thin clients, home, gaming, those are all interesting spaces, and big volume places I mentioned, mobility, that could be an area or server where we could partner with the player and really create a deep relationship."
"I could swim backwards on mobility and try and go into cell phones when the margins are going down and the competition is tough or we can try and look out and say where are the opportunities? And Lisa and her team are identifying. That's how we transformed AMD; better efficiency, better model, look to the future."
"ARM architectural leadership; x86 architecture; big leap forward, and we do it in a package where we've had very good conversation with the OEMs in the server space and with the customers, and we tested it because if we didn't see that acceptance, we would have ended the business a year ago."
"Now, will we move down to Fins? Yes, absolutely, and our next generation products go there and as we introduce them in 16. But we want to do that as you’re catching the cost curve and the yield curve at the right place so that it makes sense for us."
There is virtually 0 information on Zen, just peoples opinions on what Keller implied at the core conference. This discussion with Rory Read does not limit the application of Zen in any way shape or form. As he said, Lisa, Papermaster and Keller are looking at what spaces Zen will be applied and that could very well be HEDT. We have no clue.
There is really no reason Zen could not be a huge leap forward. It is all new, it will not be restricted by bulldozer or K10/10h in anyway. Limiting it to Haswell performance also makes little sense, because there is no real information, it could be faster, it could be slower. There is no reason why a new architecture from a different team then previous arc at 14/16nm could not greatly overtake AMD's previous architecture, we'll see.
(I personally think AMD is done with dedicated CPU's, but there is no credible information at all regarding Zen.)