Techreport - 14nm Atom schedule being moved up

[krumme]

Where I come from, schedules being moved up in the tech industry are almost invariably the exact opposite of a sign of weakness.

From my perspective schedules moved up its mostly a sign of bad planning and management more than adapting to the market. A proof of that is the change in priorities comes long time after the market change.

You might have a brilliant product that gives you the superior advantage to the competitors. If thats the case you very seldom talk about it until very close to release.

Intels 14nm Atom successor might save a lot of battery drainage from the screen and command a selling price to pay for all Intels fixed cost. I just dont believe in such miracles because they unfortunately nearly never happen and they get more and more seldom in this industry.

But if it happens i will be the first one to buy it. Hell if it even gets close i will be there

 

[Homeles]

A company's willingness to disclose information about upcoming products has no meaningful correlation with the upcoming product's success.

 

[krumme]

Why not? The fact that they will have 14nm Airmont in Q3 and possibly a new architecture before the end of 2014 bother you?

No it does not bother me. If its great in great phones that just great.

Does it bother you the qualities of the current Atom need a voltmeter and ohms law to be explained? Because it bother me and so does the sudden appearence of sources that confirm hot air.

 

[Vesku]

So AMD detailing Steamroller at last year's hot chips is a sign of weakness, correct? Because by your logic, it would be.

If you want that parallel to be accurate then it would require AMD to start talking about accelerating Excavator development and release date shortly after the Steamroller Hot Chips preview, something that AMD did not do. If AMD had done so it would closely match Bay Trail Preview -> Public talk of accelerating 14nm Atom line.

I tend to view Intel's information releases from a "talking to OEMs" perspective. Promising an aggressive 14nm Atom release in 2014 is telling OEMs "Our history of long waits between Atom improvements is over, we will have an answer to the 20nm ARM chips scheduled for 2014".

 

[Homeles]

I hold the same view as Idontcare: releasing products before or after they are ready for market introduction is folly, and this view is something that Intel in particular practices. They don't wait for inventory to clear before introducing a next generation product.

14nm Atom will be ready when it's ready. They aren't artificially moving the release date forward -- that would be disastrous. Seriously -- where would the benefit be in that?

Instead, it is almost guaranteed that things are simply running ahead of schedule.

As far as my analogy goes, it is still very relevant. krumme's argument pertains to the amount of noise that Intel's making about their next generation Atom: he believes that increasing amounts of hype are inversely proportional to the success of a product. AMD has not exactly been quiet about Kaveri, so by his logic, Kaveri is doomed to fail.

I feel that "hype" is utterly useless as an indicator of a product's performance. If you want to objectively evaluate the performance of the upcoming Atom processors (and Kaveri as well), you can educate yourself on microprocessor design and read about its architecture. Intel's also good at releasing information about the performance of their processes -- you can piece together a good ballpark figure for Silvermont. Airmont is much more unknown, since we hardly know anything about Intel's upcoming 14nm process.

 

[frozentundra123456]

If you want that parallel to be accurate then it would require AMD to start talking about accelerating Excavator development and release date shortly after the Steamroller Hot Chips preview, something that AMD did not do. If AMD had done so it would closely match Bay Trail Preview -> Public talk of accelerating 14nm Atom line.

I tend to view Intel's information releases from a "talking to OEMs" perspective. Promising an aggressive 14nm Atom release in 2014 is telling OEMs "Our history of long waits between Atom improvements is over, we will have an answer to the 20nm ARM chips scheduled for 2014".

I would consider moving up a release date much more positive than continual delays, such as with Bulldozer, Kaveri, and even the cancelling of Broadwell for the desktop. I see moving up atom as an indication that they are pulling out all the stops to try to make up for the lost opportunity. If you call being late to the market a weakness, I would agree. But that is water under the bridge and cant be changed. Whether moving up atom is too late now despite however competent the chip is, is the real question. After all ARM is thoroughly entrenched now, and may be "good enough" for phones and cheap tablets that intel will have a hard time breaking into that market no matter how good atom is. High end tablets and laptops is another story though.

 

[erunion]

Even when the new CEO talks about trading away some "perfectionism" for faster rollouts?

He talked about getting products out sooner, rather than waiting until they were perfect. Sounded to me he was talking more about side products.
Like xeon phi, the knights corner chips they recently released are a third generation product. 2 generations of larrabee were never sold.

 

[krumme]

In the end more compettition is great for us. Look at the new nexus 7 where krait 200 became krait 300. Qualcomm is dumping s600 all over and same goes for s800.
Intel will do the same and more to, to get some sales comming all this way. That will just give us freaking fast and cheap tablets and phones. Fast and cheap. We will be the winners unlike the near monopoly x86 pc cpu market. I already paid for my next phone. Glad to get a little in return.

 

[Homeles]

Whether moving up atom is too late now despite however competent the chip is, is the real question. After all ARM is thoroughly entrenched now, and may be "good enough" for phones and cheap tablets that intel will have a hard time breaking into that market no matter how good atom is.

This is an excellent point, and should be what is being discussed here.

 

[jpiniero]

For starters, what percentage of consumers over the past 2 decades would you guestimate have even heard of the term "x86" let alone factored it into their purchasing decision process at any given time when buying a computer?

Windows compatibility wasn't important to people? Intel continues to push Windows 8 despite people hating it, because they know the artificial barrier that the x86 license brings is a huge advantage for them.

If Android continues it's path to dominance, it's going to be really tough for Intel to get any kind of pricing premium.

 

[AtenRa]

Tremendous compared to what, exactly?

Compared to 28nm(HPL) and 20nm(SoC)

http://www.cadence.com/Community/bl...-20nm-16nm-finfet-and-3d-ic-technologies.aspx

Sun noted that 20SoC uses "second generation," gate-last HMKG technology and uses 64nm interconnect. Compared to the 28HPM process, it can offer a 20% speed improvement and 30% power reduction, in addition to a 1.9X density increase.

Comparison of 16nm FinFET 7.5T to 20nm 9T: 37% less power at the same speed, or 26% better performance at the same leakage.

And you've misread what he stated: look up "BEOL" and "FEOL."

As far as i know, TSMC 20nm SoC process is a full node shrink over 28nm. Both FEOL and BEOL will be shrink, no Hybrids here. 16nm FinFet BEOL will be the same from 20nm, it is the reason why they only have a 10% shrinkage.

It'll take 20nm ARM SoCs to match Intel's 22nm Silvermont.

Say hello to Tegra 4 at 28nm Quad Core, 1.9GHz and at least 2x the GPU performance of Silvermond. Also, Qualcomm 28nm Snapdragon 800 will be a Quad Core up to 2.3GHz.

Not much later, 14nm Airmont will replace it... leaving the ARM club in the dust for a lengthy period of time.

Airmont will come one year after Silvermont, that is Q3 2014. First ARM 20nm SoCs will be released in 2014. What lengthy period are you talking about ???

The ARM club will finally catch up in 2015 using second-rate 14/16nm processes, however Intel will quickly answer with Airmont's successor (a tock, mind you), where Intel will be easily be the leader for an extensive period of time.

Why do you believe that a Tock using the same 14nm process will provide Intel with that much of a difference when 14/16nm FinFets from GloFo and TSMC will provide so much more performance than 20nm planar???

As far as i can see, it is Intel that runs to catch ARM and not the other way around. Intel left ATOM so far behind that only now have started to move it above the rest of its products in order not to stay behind in the mobile market.

 

[Idontcare]

Windows compatibility wasn't important to people? Intel continues to push Windows 8 despite people hating it, because they know the artificial barrier that the x86 license brings is a huge advantage for them.

If Android continues it's path to dominance, it's going to be really tough for Intel to get any kind of pricing premium.

That's a different topic - one of x86 versus Windows OS.

What you are addressing is how Microsoft managed to claw its way into a monopoly, which had nothing to do with x86 per se as Windows was available for multiple microarchitectures for many years.

Lots of people of course know about Windows, but they don't demand x86, which is the claim to which I was specifically addressing.

 

[jpiniero]

Lots of people of course know about Windows, but they don't demand x86, which is the claim to which I was specifically addressing.

Windows has been available for other architectures, but none of them were ever really geared toward consumers. Demanding Windows is demanding x86.

 

[Homeles]

Long post

You've made some good points, and I'll submit my rebuttal tomorrow.

 

[krumme]

This is an excellent point, and should be what is being discussed here.

Absolutely. But when StrangerGuy made the same point* just from the Intel perspective on x86 side, he got personally attacked by 5 guys straight saying he was living in a fantasy world, hallucinating whatever. And there was lots of indirect personal responses also eg from idc.

When some enters a "cellebration for future-future products thread" and you dont like the post think twice before hitting the personal approach like eg responding people is using drugs like you did to one of my post you didnt like.

Frankly i dont think the same attacks would be tollerated if the label was AMD instead of Intel.

* http://forums.anandtech.com/showthread.php?p=35408603

 

[Idontcare]

And there was lots of indirect personal responses also eg from idc.

I did no such thing, you however seem keen to do so as you are specifically calling me out here. Why?

 

[krumme]

I did no such thing, you however seem keen to do so as you are specifically calling me out here. Why?

Well yes you did here
"Odly out of touch"
http://forums.anandtech.com/showthread.php?p=35408839
And notice i dont call it a personal attack, but beeing indirect personal.
Its only a minor issue and i dont care for it in itself and i also do it myself, and i know you strive and normal reach very high standard here.

Its interesting because when StrangerGuy presents a point of entrenchement explained from the point of Intel he takes a lot of flak, but when the excact same point is beeing presented from the side of arm its suddenly important.

Perhaps we should all train the habbit of asking what people mean, try to understand, instead of doing what is easy try to explain why they are wrong and explain our selves.

 

[tential]

It's funny you mention this, because some Bay Trail SKUs will be using the Celeron monkier.

I'm going to be so confused once Bay Trail comes out.

 

[krumme]

I'm going to be so confused once Bay Trail comes out.

The days of cheap single thread power seems over but people will get stunning battery life for the notebooks. I dont really think many casual users can separate a fast quad core bt from some pentium/celeron ib. Only if the gfx is radically worse/better.
But great battery life or smaller formfactor and even cheaper craptops is always a hit on the market and emerging markets.
With the atom name dead perhaps we will se a core i9?

 

[Phynaz]

Well yes you did here
"Odly out of touch"
http://forums.anandtech.com/showthread.php?p=35408839
And notice i dont call it a personal attack, but beeing indirect personal.

You failed to notice he said the post was out of touch, not the poster.