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

[GreenChile]

If AMD hired Jesus Christ as lead architect he still wouldn't be able to pull a miracle out of his hat without proper funding. It matters not what Keller has on his resume. Past accomplishments do not equal future successes, especially when underfunded.

 

[ShintaiDK]

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

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

What do you call the claims that a person singlehanded decides how the success of not just one, but two very complicated products will be. Nomatter the talent of others or R&D budget?

If anyone claims a single person can do that. Then they are neither technical or serious.

 

[Atreidin]

Some people are curious to see what will come out of Jim Keller's return, combined with the rest of the shake-ups at AMD. There is also intellectual curiosity about AMD's future designs and how they will fare and their technical details. Will it be faster than Intel's fastest? No, probably not, but I still want to see it because it is interesting. If you think AMD is innately incapable of making anything of interest, maybe you're right, but I don't see the point of constantly shoving the negative forecast down everyone's throat. We get it. AMD is in a jam. They have problems. We know. Anyone who doesn't acknowledge this isn't going to be swayed by more smarmy posts. Nobody needs to be constantly reminded about AMD's financial situation and inane economic theories about the evils of competition.

It would be nice if there could be one conversation about the future designs and the influence of the hired talent, and that's it. I don't see what's so horrible about having a discussion about how maybe someone like Jim Keller could help turn things around, not necessarily single-handed but maybe as an architect of a new plan or just simply a catalyst for positive changes. BS hyperbole about how people who don't think he's just another cog in a machine automatically equate him to a messiah aren't helping.

 

[Zstream]

What do you call the claims that a person singlehanded decides how the success of not just one, but two very complicated products will be. Nomatter the talent of others or R&D budget?

If anyone claims a single person can do that. Then they are neither technical or serious.

While there are hundreds of tasks that are completed by engineers, there are maybe one or two heads of engineering that set direction and roadmap.



I would say one person, in the right position, can put a company in a better position to execute.

 

[Shehriazad]

As much as I would like to believe such a ridiculous rumor..I will treat it as such until AMD themselves tells us a release date for such a chip.

With AMDs situation I'd almost be surprised if they even managed to switch to 16nm FinFet anytime soon.

Where is that 14nm supposed to come from? Since they rely on the "open" foundries....pffffff.

And where would such a monster chip come from? AMD just had such a design laying around? Because it sure as hell didn't have the R&D budget.

8 core with hyperthreading...16 "core" but not a modular design? (Since AMD claimed before that modular design is over with Excavator)

I mean sure an AMD "16" core that can actually single thread as strong as Intels current gen stuff + AMDs aggressive pricing would be a wet dream for any gamer I bet...but one has to wonder how such a rumor even forms.

That said..if such a chip should ever come out...I would not give a rats tail about the 95w.

Anything below 100 for a chip with 8 cores that delivers strong performance is a very good chip imo. 100 and below is very easy to cool down as opposed to 200+++w TDPs for AMDs current "high end" cherry picks.

 

[Idontcare]

+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.

Dirk Meyer was the lead architect for the K7.

 

[scannall]

Some people are curious to see what will come out of Jim Keller's return, combined with the rest of the shake-ups at AMD. There is also intellectual curiosity about AMD's future designs and how they will fare and their technical details. Will it be faster than Intel's fastest? No, probably not, but I still want to see it because it is interesting. If you think AMD is innately incapable of making anything of interest, maybe you're right, but I don't see the point of constantly shoving the negative forecast down everyone's throat. We get it. AMD is in a jam. They have problems. We know. Anyone who doesn't acknowledge this isn't going to be swayed by more smarmy posts. Nobody needs to be constantly reminded about AMD's financial situation and inane economic theories about the evils of competition.

It would be nice if there could be one conversation about the future designs and the influence of the hired talent, and that's it. I don't see what's so horrible about having a discussion about how maybe someone like Jim Keller could help turn things around, not necessarily single-handed but maybe as an architect of a new plan or just simply a catalyst for positive changes. BS hyperbole about how people who don't think he's just another cog in a machine automatically equate him to a messiah aren't helping.

This is exactly right. I am interested in seeing what comes out. I make no claims as to what that might be, nor is anyone else. I'm a great lover of technology, no matter who makes it. But there are those that go rabid with some odd form of internet Tourette's syndrome when something is said that might possibly be positive for AMD.

 

[III-V]

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

Interestingly enough, it was the switch to Intel that marked the beginning of that. Yet people are calling for the end of that relationship, which is nonsense.

Apple created entirely new markets with the iPhone and iPad. Whatever AMD cooks up cannot and will not do that. It's entirely possible for AMD to be successful in the post-Bulldozer family era, but they can't create the "next big thing."

The problem with Keller is that people are just so, so desperate for AMD to do well, that he's been hyped to hell and back. In reality, Keller can only do as much as he has the resources to do. I have no doubt he is very talented, but he is just one man, with an ever-dwindling number of people under his command.

Jim Keller was only with AMD for a single year, before he left the first time around.

Dirk Meyer was the lead architect for the K7.

And he ended up playing a huge role in putting AMD on its deathbed.

 

[CHADBOGA]

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.

No. We have seen multiple instances now where people who refused to drink the anti-Intel, anti-Nvidia Kool Aid got banned for presenting their views in a perfectly respectable manner.

SA Forums are terrible, occasionally people with high technical level will drop in to attempt to correct the misinformation running rampant, but it would be one hell of a stretch to describe their regulars as having a high technical level.

 

[beginner99]

In huge corporations a single person still can have a huge influence and I'm not talking about the CEO. So it's possible we will get surprised.

However this is almost 2 years away. That's a long time. I will be disappointed if the can't manage to use the iGPU for something else than graphics by then by default like say for physics while graphics run on a dGPU.

 

[Shehriazad]

So AMD plans to stay alive throughout 2015 via a Kaveri refresh, console semicustom parts...a whack GPU 300 series and mobile Carrizo cores.

at least 2 of those things are going to bring nothing but losses...will AMD even be around for 2016 to release the darned thing? Then again...this rumor might just be some nerds' wishful thinking and Zen ends up being the greatest failure in the history of CPUs.

 

[Erenhardt]

The problem with Keller is that people are just so, so desperate for AMD to do well,

Looking at your and some others posts, I have to disagree. Looks like people are desperate for amd to fail.

 

[AtenRa]

So AMD plans to stay alive throughout 2015 via a Kaveri refresh, console semicustom parts...a whack GPU 300 series and mobile Carrizo cores.

They did with Kaveri, AM3+, Opterons, Consoles, Embedded and GPUs in 2014, why do you believe they will not in 2015 when they will have new products like Carrizo and new GPUs ??

Carrizo will be a huge improvement for the Mobile segment, it could make a big difference.

 

[Ancalagon44]

While there are hundreds of tasks that are completed by engineers, there are maybe one or two heads of engineering that set direction and roadmap.



I would say one person, in the right position, can put a company in a better position to execute.

This.

Remember, one person approved the Bulldozer idea. Many talented engineers worked on it, but they did so at the direction of an executive who decided that this was the strategy to follow.

We all know how Bulldozer turned out.

I don't expect Jim Keller to lay the circuits by hand, I expect him to give the engineers a target to work towards that makes sense for the market.

 

[III-V]

Looking at your and some others posts, I have to disagree. Looks like people are desperate for amd to fail.

You disagree with what I have to say? How predictable.

I actually make a pretty big effort to keep my biases in check. If you disagree with me, the problem is likely not on my end.

In huge corporations a single person still can have a huge influence and I'm not talking about the CEO. So it's possible we will get surprised.

Yes, it's possible. Unfortunately, people here do not focus on what probable, and blow what's possible out of proportion.

 

[AtenRa]

No matter how good your design is, if you dont have the right tools you are coming second best. But from 2016 onwards, AMD will have the right tools (14nm FF) to leverage all the knowledge they have both in CPU and GPU and make a big comeback.

 

[Spawne32]

So AMD plans to stay alive throughout 2015 via a Kaveri refresh, console semicustom parts...a whack GPU 300 series and mobile Carrizo cores.

at least 2 of those things are going to bring nothing but losses...will AMD even be around for 2016 to release the darned thing? Then again...this rumor might just be some nerds' wishful thinking and Zen ends up being the greatest failure in the history of CPUs.

I'm thoroughly convinced after the recent post of their quarterly earnings and the 7% layoffs that they will be bought out or be making a deal with someone like samsung in the near future. I don't think they are viable at all within the next 2 years to bring that income back without an infusion of new money and a change in leadership. This might as well be the straw that broke the camels back for any potential future designs that may or may not have brought them back.

 

[Shehriazad]

Long story short...I doubt that anyone would mind if AMD brought out a killer CPU in 2016...especially with their killer-pricing.

We will just have to see if this last straw of AMD will be worth holding onto...or if it rips before they can even grab it.

I mean the possibilities are great tbh.

-8 cores, 16 threads
-No Bulldozer modular failure design -> IPC improvements high enough to somewhat catch up to Intel
-Possible use of HBM as dedicated L4(?) cache since by then they will have had some experience with it.
-DDR4 which should be far more affordable and better clocked on average by then.
-14nm, thus lots of space on die for either an igpu or HBM.
-If iGPU is present, possible use of advanced HSA, which they by then will also have had more experience with.

But all of this is just me fantasizing about what I would like. If they managed to bring their IPC/Single threaded up to what the current Intel 5K series is TODAY...then this chip would likely have a right to exist. I mean AMD is again the first to jump onto HBM and HSA...by 2016 possible kinks would have been worked out (I suppose)...so this COULD be nice...but that's the last I'll say/guess about this until we get official news in 2016...or hear about AMD being bought out by another giant. xD

 

[ShintaiDK]

No matter how good your design is, if you dont have the right tools you are coming second best. But from 2016 onwards, AMD will have the right tools (14nm FF) to leverage all the knowledge they have both in CPU and GPU and make a big comeback.

Wouldnt they have to use TSMCs 16FF for high performance parts? Since the 14FF is a low power process designed for ~5W devices.

But again...always wait..better in the future? But it never have been, has it?

 

[AtenRa]

Wouldnt they have to use TSMCs 16FF for high performance parts? Since the 14FF is a low power process designed for ~5W devices.

There will be high-perf 14nm FF.

http://www.globalfoundries.com/technology-solutions/leading-edge-technology/14-lpe-lpp

• 14nm FinFET Technology
  • 14LPE – Early time-to-market version with area and power benefits for mobility applications
  • 14LPP – Enhanced version with higher performance and lower power; a full platform offering with MPW, IP enablement and wide application coverage
• Immediate availability
  • PDK and DM available now for design starts
  • Silicon maturity on track at Fab 8, New York
  • MPWs since 2014
• Wide range of applications
  • Mobile and wireless – lower watts per GHz
  • Computer, network and storage – more performance per watt

Also to note, LLE and LPP have the same rules, so you can directly port one design from one process to the other with minus minimal time and resources spend.

But again...always wait..better in the future? But it never have been, has it?

Isn't the same for Intel's GPU, always next year ??

 

[ShintaiDK]

LPP is still a low power process. GloFo doesnt have a process on its own. So its just to look at what Samsung got.

 

[AtenRa]

LPP is still a low power process. GloFo doesnt have a process on its own. So its just to look at what Samsung got.

Both Samsung and GloFo will have 14nm FF LPP in 2016.

http://globalfoundries.com/docs/def...dries-14nm-collaboration---final.pdf?sfvrsn=2

 

[Gikaseixas]

Some people are curious to see what will come out of Jim Keller's return, combined with the rest of the shake-ups at AMD. There is also intellectual curiosity about AMD's future designs and how they will fare and their technical details. Will it be faster than Intel's fastest? No, probably not, but I still want to see it because it is interesting. If you think AMD is innately incapable of making anything of interest, maybe you're right, but I don't see the point of constantly shoving the negative forecast down everyone's throat. We get it. AMD is in a jam. They have problems. We know. Anyone who doesn't acknowledge this isn't going to be swayed by more smarmy posts. Nobody needs to be constantly reminded about AMD's financial situation and inane economic theories about the evils of competition.

It would be nice if there could be one conversation about the future designs and the influence of the hired talent, and that's it. I don't see what's so horrible about having a discussion about how maybe someone like Jim Keller could help turn things around, not necessarily single-handed but maybe as an architect of a new plan or just simply a catalyst for positive changes. BS hyperbole about how people who don't think he's just another cog in a machine automatically equate him to a messiah aren't helping.

Great post

Lately I avoid getting into arguments about future AMD products because these threads always attract the anti-AMD brigade.
My 2 cents... Keller is a lead engineer and will have influence on what's being developed right now. That's why he was brought aboard, to make decisions that hopefully could help the delivery of better products. Arguing that Intel needs hundreds or thousands is a bit childish since at the end of the day only one design direction will get the nod.

 

[witeken]

No matter how good your design is, if you dont have the right tools you are coming second best. But from 2016 onwards, AMD will have the right tools (14nm FF) to leverage all the knowledge they have both in CPU and GPU and make a big comeback.

Right tools is relative. They will still be 1-2 node generations behind with 20FF like they've been for a long time already.

 

[teejee]

In a large R&D project there are at least three levels of technical decision meetings in the hiearchy. Difficult issues always end up at the highest level.The person responsible for the highest level takes at least ten decisions per week concerning these difficult issues, that menas more than 1000 tough decisions during a three year project. This person have a very big influence on the outcome.