[Sweclockers] AMD Zen coming in Q3 2016, will be on 14 nm

[R0H1T]

Within 95 W TDP I think an APU with 8C/16T and iGPU it totally feasible on 14 nm.

I think it's more like 4c/8t or true eight cores kinda like the tiny console APU's, your config is also less likely cause they'll probably have to ditch the IGP in order to achieve with a reasonable amount of die space. Also with the socket being FM3 I'd say that's highly unlikely.

 

[Fjodor2001]

I think it's more like 4c/8t or true eight cores kinda like the tiny console APU's, your config is also less likely cause they'll probably have to ditch the IGP in order to achieve with a reasonable amount of die space. Also with the socket being FM3 I'd say that's highly unlikely.

I guess it depends on what the requirements on CPU frequency and number of iGPU cores are. But yeah, maybe you're right. Possibly there could instead be an 8C/16T pure CPU version without iGPU (replacing FX8350 and friends), and then 4C/8T + iGPU versions.

 

[turtile]

I think Zen is going to continue the module approach but the cores will share less resources than Bulldozer. Probably either 2 or 4 core modules sharing certain components.

 

[MiddleOfTheRoad]

I call that utter bullshit. In this world there are too many "I know better than you" people and so that is the reason why one single person makes a lot difference.

Agreed, Let me quote a little TechReport:
"Keller has an impressive resume. In the 80s and 90s, he worked at DEC, where he co-architected the Alpha 21164 and Alpha 21264 processors. He only worked at AMD between 1998 and 1999, but he definitely left his mark on the company. In addition to his aforementioned roles on the K8 and the x86-64 spec, Keller co-authored the Hyper-Transport specification and was part of the system engineering team behind the K7."

That doesn't even mention all the iPhones and iPads that are now powered by his chip designs. He is a rock star and was probably the reason AMD was able to leapfrog Intel in performance when the K7 was released.

 

[III-V]

Agreed, Let me quote a little TechReport:
"Keller has an impressive resume. In the 80s and 90s, he worked at DEC, where he co-architected the Alpha 21164 and Alpha 21264 processors. He only worked at AMD between 1998 and 1999, but he definitely left his mark on the company. In addition to his aforementioned roles on the K8 and the x86-64 spec, Keller co-authored the Hyper-Transport specification and was part of the system engineering team behind the K7."

That doesn't even mention all the iPhones and iPads that are now powered by his chip designs. He is a rock star and was probably the reason AMD was able to leapfrog Intel in performance when the K7 was released.

Oh, here we go again with the computer chip Jesus stuff, where people literally believe he's come back from the the dead Apple to save us from Satan Intel.

 

[Arachnotronic]

Oh, here we go again with the computer chip Jesus stuff, where people literally believe he's come back from the the dead Apple to save us from Satan Intel.

+1

Does anybody here know the name of the Haswell lead architect? How about Sandy Bridge? Nehalem?

 

[witeken]

+1

Does anybody here know the name of the Haswell lead architect? How about Sandy Bridge? Nehalem?

It's never just one person. The development of state of the architecture like Haswell is years of work of many people. To get an idea, CTRL+F "architect" and at the bottom you'll find a lot of Intel architects what exactly they did or researched.

http://pages.cs.wisc.edu/~rajwar/papers/ieee_micro_haswell.pdf

One of the lead architects of Nehalem: https://intel.activeevents.com/sz14...ww?PERSON_ID=292B769CBEFA5D3382910F68257056F4.

 

[scannall]

Oh, here we go again with the computer chip Jesus stuff, where people literally believe he's come back from the the dead Apple to save us from Satan Intel.

I don't think anyone here is expecting AMD to all of a sudden cream Intel. Just that Jim Keller is a very talented guy, and that it will be interesting to see what they come up with.

And since this is a CPU focused thread and forum, it seems an appropriate thing to speculate on.

Though, being a thread about the AMD Zen I'm not sure where Intel comes into it.

 

[Fjodor2001]

+1

Does anybody here know the name of the Haswell lead architect? How about Sandy Bridge? Nehalem?

Do you think Apple would have been the same without Steve Jobs too? He was also just one guy.

 

[witeken]

Do you think Apple would have been the same without Steve Jobs too? He was also just one guy.

This is something very different. A CEO decides the direction of a company. It's up to the employees to execute towards that vision, roughly speaking. Just having 1 person capable of leading the design of an envisioned competitive architecture is just one part of the jigsaw puzzle: you've also got to have the people and the budget to research and develop the product.

 

[Fjodor2001]

This is something very different. A CEO decides the direction of a company. It's up to the employees to execute towards that vision, roughly speaking.

Likely Jim Keller decides on the direction of the uArch development work. I think that can be just as important.

 

[Arachnotronic]

Likely Jim Keller decides on the direction of the uArch development work. I think that can be just as important.

Even if Keller is a chip development god, he's still working under significant financial constraints relative to Intel. He's also starting from a much lower technology "base". Intel has successful architectures and it can build atop those ideas already. AMD, since its technology isn't as good, need more "new" ideas, and developing them costs money.

 

[ShintaiDK]

+1

Does anybody here know the name of the Haswell lead architect? How about Sandy Bridge? Nehalem?

Nope. Because its the combined effort of 1000s of engineers that makes the product. And any single person isnt any more important than the rest.

Same as the names behind GCN, Kepler or Maxwell. Who can name those? Snapdragons? Exynos?

But Jim Keller, he can singlehanded make it all! Specially on such a low budget.

Do people even wonder how insulting that is to the rest of the engieers in AMD?

 

[cytg111]

Nope. Because its the combined effort of 1000s of engineers that makes the product. And any single person isnt any more important than the rest.

Same as the names behind GCN, Kepler or Maxwell. Who can name those? Snapdragons? Exynos?

But Jim Keller, he can singlehanded make it all! Specially on such a low budget.

Do people even wonder how insulting that is to the rest of the engieers in AMD?

Why is this guy not banned yet? Too many alts, too much work? Whatever, have your fanfest circle, Im out.

 

[Fjodor2001]

Even if Keller is a chip development god, he's still working under significant financial constraints relative to Intel. He's also starting from a much lower technology "base". Intel has successful architectures and it can build atop those ideas already. AMD, since its technology isn't as good, need more "new" ideas, and developing them costs money.

You do realize Apple was close to bankruptcy before Steve Jobs joined the company again? Today it's the tech company everyone looks up to.

Things change fast in the IT business. Head count, current market cap, and past successes do not guarantee future success. Nokia went from the absolute leader in mobile phone tech with 40% market share to nearly nothing in just a few years. Sony, Altavista and others are other examples of unexpected decline. The IT history is full of market game changers that no one could predict in advance, or they'd be filthy rich from placing their bets in the stock market.

Only the paranoid survives...

 

[MiddleOfTheRoad]

+1

Does anybody here know the name of the Haswell lead architect? How about Sandy Bridge? Nehalem?

That's the advantage of working for small companies like AMD or DEC. He was co-architect of those Alpha CPU's -- I seriously doubt anyone at Intel gets that much opportunity/influence over a final design.

 

[frozentundra123456]

Well, as usual, we have deteriorated to extremes with no middle ground. Keller could make a difference. Granted it takes a lot of engineering talent and resources to design a chip, but there are critical design choices at various points of development that a single person could influence. At least, perhaps one person could steer the company away from disastrous decisions, even if the ultimate product is limited by the resources of the company.

OTOH, he has a good track record, but that does not necessarily mean everything he designs will be brilliant, and company resources and politics may also limit what he can do. So both the assumption that he can do nothing and the opposite that he will somehow come up with a miracle seem unwarranted.

In any case, it is just total guesswork now. Accounting for possible delays and giving time to build up the stock, it could be close to two years before we see final product. We dont even know yet the market it will be focused on, low power or high performance.

Edit: another caveat is that cpus are a mature product now. Innovation and performance improvements are much harder to come by than they were in the early days of cpu development. So the chances of a great breakthrough simply because one person is aboard would seem much less than in the earlier days.

 

[witeken]

That's the advantage of working for small companies like AMD or DEC. He was co-architect of those Alpha CPU's -- I seriously doubt anyone at Intel gets that much opportunity/influence over a final design.

From Intel CPU architect:

For features, the project members propose ideas, marketing proposes ideas, and our big customers propose ideas. We develop them for a while, and then whittle them down to the subset that fits our die area and schedule targets. Sorry for the oversimplification, but that is it in a nutshell.

http://www.reddit.com/r/IAmA/comments/15iaet/iama_cpu_architect_and_designer_at_intel_ama/?limit=500
And:

Q:What level of autonomy do you have when someone comes up with a promising idea?
A: Yes, but the bar is very high to get anything into silicon. But anyone can propose a feature.

 

[MiddleOfTheRoad]

Do you think Apple would have been the same without Steve Jobs too? He was also just one guy.

+1
One guy invented the light bulb. One guy invented the modern assembly line. K7 was also developed by Keller on a shoestring budget and rocketed past Intel performance. I have no doubt that Keller will create a great piece of technology -- I'm a hell of a lot more concerned that Global Foundries can actually manufacture the damn thing in a timely fashion.

 

[frozentundra123456]

So Keller is the modern day equivalent of Edison or Henry Ford?? OK........ if you say so.

 

[MiddleOfTheRoad]

Well, as usual, we have deteriorated to extremes with no middle ground. Keller could make a difference. Granted it takes a lot of engineering talent and resources to design a chip, but there are critical design choices at various points of development that a single person could influence. At least, perhaps one person could steer the company away from disastrous decisions, even if the ultimate product is limited by the resources of the company.

OTOH, he has a good track record, but that does not necessarily mean everything he designs will be brilliant, and company resources and politics may also limit what he can do. So both the assumption that he can do nothing and the opposite that he will somehow come up with a miracle seem unwarranted.

Agreed. I suspect that AMD is probably handling the situation: "Whatever Keller wants, Keller gets." He's got a 20 year track record of success. If management was smart (which clearly has not been the case for AMD for many years of its history) -- they'd probably be best served to let the engineers run the show and not try to micro manage the project.

 

[MiddleOfTheRoad]

So Keller is the modern day equivalent of Edison or Henry Ford?? OK........ if you say so.

I don't think he's on the same level as Steve Jobs -- but I'd probably say he's a lot like a modern Steve Wozniak. It's kinda hard for any current CPU architect to be an Edison or Ford since the industry is so mature at this point.

 

[Fjodor2001]

Just to remind everyone how quickly things can turn around in this business:

 

[CHADBOGA]

Why is this guy not banned yet? Too many alts, too much work? Whatever, have your fanfest circle, Im out.

You generally don't get banned for making well thought out posts, that have very reasonable content in it, unless of course you are posting at Semi-Accurate's forums.

Perhaps you would be happier there?

 

[el etro]

You generally don't get banned for making well thought out posts, that have very reasonable content in it, unless of course you are posting at Semi-Accurate's forums.

Perhaps you would be happier there?

Trolls can still be trolls without inflicting the rules, they just have to throw the bait at other posters.

SA Forums is a very good forum, with posters with high technical level. Whoever got banned there was banned due to insistently trolling some threads.