[AMD] K12 will be on 28nm

[Arachnotronic]

First you can prove that Intel's 14nm process requires a higher voltage for the same design. Your buddy hasn't proven a thing. Operating voltage is a design parameter and a higher operating voltage does not indicate inferiority.

Your logical fallacy is that perf/watt gains can only occur by reducing voltage. what if I designed a circuit that was more efficient but yet required a higher voltage to operate than a less efficient design?

Back in the real world, Intel is presenting its 14nm process at IEDM and claims NMOS saturated drive current improvement of 15% over 22nm and 41% PMOS over 22nm. Intel also claims that the transistors operate at just 0.7V, a reduction from the 0.75V that the 22nm ones operated at in the company's IEDM 2012 paper.

http://electroiq.com/blog/2014/10/i...ategies-on-competing-substrates-at-iedm-2014/

 

[Phynaz]

Nice find. I'm going to give that site a read.

 

[john5220]

@ tential May I then ask, do you think we will be moving away from X86 to ARM anytime soon for the desktop market?

Like video games etc. Would there come a time when X86 will be like a distant past and everything will be ARM?

 

[tential]

@ tential May I then ask, do you think we will be moving away from X86 to ARM anytime soon for the desktop market?

Like video games etc. Would there come a time when X86 will be like a distant past and everything will be ARM?

Best to pose that to the thread and not just at me.

My opinion is No.

 

[Arachnotronic]

Like video games etc. Would there come a time when X86 will be like a distant past and everything will be ARM?

What would cause that?

 

[Phynaz]

@ tential May I then ask, do you think we will be moving away from X86 to ARM anytime soon for the desktop market?

Like video games etc. Would there come a time when X86 will be like a distant past and everything will be ARM?

Soon? No.
Eventually? Maybe.

The problem with replacing x86 is the software legacy. There's literally trillions of dollars of x86 software that would have to be replaced. That's a pretty big mountain to climb.

Btw, the desktop pc is a dead man walking.

 

[beginner99]

The question is: Will the lower power consumption and higher performance be enough to justify 14 nm instead of 28 nm for server CPUs?

The last few years we've seen that later nodes does no longer bring much higher performance. The power consumption will likely be a bit lower, but is it enough to justify the much higher cost? I'm not so sure. I bet it's decisions like that which makes AMD provide such an evasive answer.

Of course, but at what cost. Are you willing to pay 2x as much for 30% power savings? 4x as much... ?

Besides what was already mentioned about cooling costs and total bom, never forget that a server usually runs software costing 100x times as much as the servers hardware. So in the total lifetime cost of a server the CPU price is pretty much irrelevant.

 

[jdubs03]

Saw this:

http://www.kitguru.net/components/graphic-cards/anton-shilov/amd-first-products-made-using-14nm16nm-finfet-are-on-track/

Maybe in 2015 we'll see 28nm CPUs and SoCs with 20nm GPUs, and then 2016 being 14nmFF GPUs and maybe then XXnm CPUs and SoCs. But who knows there isn't much that is certain regarding AMD.

 

[Abwx]

First you can prove that Intel's 14nm process requires a higher voltage for the same design. Your buddy hasn't proven a thing. Operating voltage is a design parameter and a higher operating voltage does not indicate inferiority.

Well, thanks for the enginering point of view but it work the other way around.

Your logical fallacy is that perf/watt gains can only occur by reducing voltage. what if I designed a circuit that was more efficient but yet required a higher voltage to operate than a less efficient design?

Lol, more voltage and higher efficency is just complete non sense, when you increase voltage by 40% TDP increase by 100%, and the perf wil be the same if frequency is unchanged.


I see that to try t, against all evidence, contradicting me you have to make abstraction of the law of physics, keep on saying that i have no facts if it can help you believe whatever fits your preferences...

 

[Phynaz]

Ok, so circuit design has nothing to due with power consumption, it's all about the electrical properties of the manufacturing process.

Explain AMD Faildozer being a a perf/watt disaster then.

Do you believe what you post or are you just a troll?

 

[rtsurfer]

can we expect a K12 AMD to beat a 2500K i5 sandy bridge?

if AMD does this its a win. 2500K will remain a beast for the next 5 years.

K12 is an ARM part.
Zen is the new x86 core, which we hope at least catches up to Sandy Bridge.
But can't be guaranteed looking at AMD's track record.

Sandybridge, Ivybridge, Haswell, Devil's Canyon, Broadwell, Skylake. That's 6 generations of CPUs that people will have.

Devil's Canyon doesn't count as a new generation.
Adding more capacitors to an existing chip & maybe some better binning for better over clocking doesn't make a new generation of processors.

That is like saying that the new AMD 8370E chips are a newer generation of chips then PileDriver 8350.
Tweaking & extracting more performance out of a mature manufacturing process on the same CPU architecture is not a new generation CPU.

 

[Abwx]

Ok, so circuit design has nothing to due with power consumption, it's all about the electrical properties of the manufacturing process.

Explain AMD Faildozer being a a perf/watt disaster then.

Do you believe what you post or are you just a troll?

You can do the best design of the world but it wont manage to compete if the process is too borked, for instance Baytrail run at 0.73V at 2.2-2.4 while the Core M is at 0.95-0.988V at same frequencies, if Baytrail was manufacteured with the same process it would dissipate (0.95/0.73)^2 = 69% more and there s nothing you could do through the design to compensate for such a difference, even 10% difference in voltage is about impossible to gap.

Currently the only advantage of Intel is said lower voltage, that is, process, a FX need 17% more voltage at 3.5 than a 4770K, this induce 37% more TDP, and despite this a FX8370E manage to be at same power efficency than a 4670K on high throughput softs, at equal process the FX would had better efficency than a 4770K, let alone if it s a Kaveri based FX.

As for believing what i write, of course since i rely on factual datas rather than on urban legends, i told you i give no importance to the usual hearsay that we can read here and there and wich seems to be the basis of your forever biaised assumptions.

 

[mrmt]

Maybe in 2015 we'll see 28nm CPUs and SoCs with 20nm GPUs, and then 2016 being 14nmFF GPUs and maybe then XXnm CPUs and SoCs. But who knows there isn't much that is certain regarding AMD.

Anton Shilov is the most incompetent journalist I can think of. Papermaster didn't mention any date when talking about FINFet, just like AMD won't be in the bleeding edge anymore, exactly what Kumar did. He also added some very interesting bits regarding AMD strategy, that they won't fold any division they have today, and that they expect to have competitive products on the bottom end of the market even when considering the node handicap, among others. But despite having these topics available Anton Shilov manages to write a story about a non-existent point in Papermaster's speech, pulling assumptions out of thin air.

 

[NTMBK]

Anton Shilov is the most incompetent journalist I can think of. Papermaster didn't mention any date when talking about FINFet, just like AMD won't be in the bleeding edge anymore, exactly what Kumar did. He also added some very interesting bits regarding AMD strategy, that they won't fold any division they have today, and that they expect to have competitive products on the bottom end of the market even when considering the node handicap, among others. But despite having these topics available Anton Shilov manages to write a story about a non-existent point in Papermaster's speech, pulling assumptions out of thin air.

I wouldn't say he's incompetent. He's very good at what he does. He knows exactly what he is doing every time he posts a sensationalist, misleading, clickbait headline with dubious sourcing and deliberate misinterpretation of quotes. I'm sure he gets lots of ad revenue from his garbage "news" stories.

 

[Phynaz]

You can do the best design of the world but it wont manage to compete if the process is too borked, for instance Baytrail run at 0.73V at 2.2-2.4 while the Core M is at 0.95-0.988V at same frequencies, if Baytrail was manufacteured with the same process it would dissipate (0.95/0.73)^2 = 69% more and there s nothing you could do through the design to compensate for such a difference, even 10% difference in voltage is about impossible to gap.

Currently the only advantage of Intel is said lower voltage, that is, process, a FX need 17% more voltage at 3.5 than a 4770K, this induce 37% more TDP, and despite this a FX8370E manage to be at same power efficency than a 4670K on high throughput softs, at equal process the FX would had better efficency than a 4770K, let alone if it s a Kaveri based FX.

As for believing what i write, of course since i rely on factual datas rather than on urban legends, i told you i give no importance to the usual hearsay that we can read here and there and wich seems to be the basis of your forever biaised assumptions.

Talk about hearsay, do you have anything to back up the bolded quote? So far you have completely failed to provide any evidence of your statements. All you do is deflect onto others and change attempt to change the subject rather than provide citations.

 

[Abwx]

Talk about hearsay, do you have anything to back up the bolded quote? So far you have completely failed to provide any evidence of your statements. All you do is deflect onto others and change attempt to change the subject rather than provide citations.

Stubborness is not an argument, i guess that thoses numbers hurts your brand preference, it s you that are relying on myths and deflections as a mean to patheticaly negates numbers..

As for the bolded part good luck for compensating, at equal competences with the competition, what amount to 20% lower perf/watt due to process

 

[tential]

K12 is an ARM part.
Zen is the new x86 core, which we hope at least catches up to Sandy Bridge.
But can't be guaranteed looking at AMD's track record.



Devil's Canyon doesn't count as a new generation.
Adding more capacitors to an existing chip & maybe some better binning for better over clocking doesn't make a new generation of processors.

That is like saying that the new AMD 8370E chips are a newer generation of chips then PileDriver 8350.
Tweaking & extracting more performance out of a mature manufacturing process on the same CPU architecture is not a new generation CPU.

Generation was meant in a "consumer" sense. Better to have put 6 "years" of chips when talking on here. The point still stands, If AMD actually attempts to still compete in the highend market with a chip that just reaches sandybridge levels of performance, by 2016, that's not great for marketing.

While AMD is trying to catch sandybridge, Intel is on Skylake...

I agree with the other posters in this thread that it's much more likely to be focused on the lower end and that AMD is done trying to pitch high end processors to compete with Intel. Best for AMD to focus on the biggest growth segments than to still compete against all segments vs a vastly wealthier intel.

 

[BSim500]

Stubborness is not an argument, i guess that thoses numbers hurts your brand preference, it s you that are relying on myths and deflections as a mean to patheticaly negates numbers..

All you've done is quote the maximum possible short-burst Turbo OC voltage, not it's normal operating voltage. Either you didn't bother to read the article you're quoting, or you simply aren't honest enough to quote the context of the single images you're posting (as you've been caught doing before):-

"Notice however that the maximum Turbo clocks are the same - 2.60 GHz. Intel told us months ago that Broadwell-Y was designed to run at extremely low power and clock speeds when idle but still reach high performance and high clock speeds on demand, when needed, in short bursts. Most importantly, note the significant TDP difference between these two processors. The Core M 5Y70 only requires a thermal solution designed for 4.5 watts. The Core i5-4200U required a 15 watt system design - more than 3x the potential heat to dissipate.

The Yoga 3 Pro with the Core M 5Y70 is able to pull out a result that is 8% longer than the Core i5-4200U. That's great and all, but is made even more impressive when you learn that the Yoga 3 Pro has a 22% smaller battery in it! Lenovo shipped the latest Yoga machine with a 44 Whr battery compared to the 54 Whr battery found in last years Yoga 2 Pro."

http://www.pcper.com/reviews/Proces...ance-Testing-Broadwell-Y/Power-Consumption-an

Nice noticeable increase in battery life for a chip that (according to you) "uses more power" based solely on its max Turbo OC voltage state that it spends less than 1% of its life in. :thumbsdown:

 

[Enigmoid]

Stubborness is not an argument, i guess that thoses numbers hurts your brand preference, it s you that are relying on myths and deflections as a mean to patheticaly negates numbers..

You can do the best design of the world but it wont manage to compete if the process is too borked, for instance Baytrail run at 0.73V at 2.2-2.4 while Haswell is at 0.95-0.988V at same frequencies, if Baytrail was manufacteured with the same process it would dissipate (0.95/0.73)^2 = 69% more and there s nothing you could do through the design to compensate for such a difference, even 10% difference in voltage is about impossible to gap.

Why are you comparing BT to HW? BT generally operates with less voltage than HW at the same frequency?

 

[Exophase]

You can do the best design of the world but it wont manage to compete if the process is too borked, for instance Baytrail run at 0.73V at 2.2-2.4 while the Core M is at 0.95-0.988V at same frequencies, if Baytrail was manufacteured with the same process it would dissipate (0.95/0.73)^2 = 69% more and there s nothing you could do through the design to compensate for such a difference, even 10% difference in voltage is about impossible to gap.

You can't compare voltage/frequency curves for two totally different CPUs. The design drives this, not just the process. Compare Krait and Cortex-A15 with both on TSMC 28HPM, for example.

 

[NTMBK]

Ashraf emailed AMD. Apparently Devinder mis-spoke, and K12 is on a FinFET process and not 28nm: http://semiaccurate.com/forums/showthread.php?t=8379

(Sorry to steal your thunder Ashraf )

 

[Exophase]

Wonderful, so once again AMD lets an employee completely misrepresent their product. This time their senior vice president.

Here's a tip: if you don't know don't give an answer. Say that you'll reveal it later or something. Even actually saying "I don't know" is better than this.

 

[kimmel]

You can do the best design of the world but it wont manage to compete if the process is too borked, for instance Baytrail run at 0.73V at 2.2-2.4 while the Core M is at 0.95-0.988V at same frequencies, if Baytrail was manufacteured with the same process it would dissipate (0.95/0.73)^2 = 69% more and there s nothing you could do through the design to compensate for such a difference, even 10% difference in voltage is about impossible to gap.

You really do have no idea how products are designed do you? It also sounds like you have little electrical knowledge beyond ohm's law. The number of independent variables you group together in this analogy alone is staggering. Nothing about process technology and silicon design is so easily put in a box as you want to do.

 

[Fjodor2001]

Not Intel. Intel 14nm Atom server chips will eat AMD 28nm chips for breakfast in any metric you can think off.

Including perf/cost?

Nobody buys a server based on perf/costs, but based on TCO for a given scenario. Given that the costs of powering and cooling a processor in datacenter are bigger than the costs of acquisition of a server processors, even when considering the huge price premium over desktop counterparts, perf/watt is the absolute metric for servers. This is why Opteron was wiped out on servers but not on desktops.

So Intel is not better in every metric after all. I.e. AMD will likely be better on perf/cost.

Also, the savings on perf/watt of course must be weighed against the cost of the CPU itself. If the CPU_A is much more expensive than CPU_B, then the TCO of CPU_A will not be lower than for CPU_B, even if CPU_A has better perf/watt. That is when calculated over it's lifetime.

 

[ikachu]

I was kinda shocked when I saw that quote, I know they definitely weren't originally targeting 28nm. I'm glad that it's been cleared up.

Devinder is alright when speaking at all hands meetings and other internal events but he fails pretty hard at speaking with investors.

It's probably worthy of a new thread but did anyone see that AMD is moving from NYSE to NASDAQ?

http://biz.yahoo.com/e/141209/amd8-k.html