Skylake/Broadwell Roadmap Update @Vr-zone

[Fjodor2001]

He's talking about foundries' 10nm.

The article also says this about Intel's 10 nm:

An Intel Executive recently predicted 10nm would be available in 2017 in a candid interview on GulfNews.com out of Dubai of all places:

“We have been consistently pursuing Moore’s Law and this has been the core of our innovation for the last 40 years. The 10nm chips are expected to be launched early 2017,” said Taha Khalifa, general manager for Intel in the Middle East and North Africa region.

Mr. Khalifa is a 24 year Intel veteran so he should certainly know.

 

[witeken]

The article also says this about Intel's 10 nm:

What don't you understand from this quote from January's earnings call, just 2 weeks ago:

Brian Krzanich - Chief Executive Officer
We are timing on 10-nanometers. We are not going to come out with when we’ll be introducing a 10-nanometer to the marketplace in general, probably until the end of this year. So again as we go through the year probably by the investor meeting in November, we’ll give you an outlook on how and what timing is for 10-nanometers.

Emphasis mine.

Edit, don't forget:

We felt like we went on a little early with 14nm as far as timing and performance and features and we saw actually competitors adjust to that. So we're gonna be a little bit more prudent, a little smarter about signaling to the industry exactly when, what and where. And you'll have to trust a little bit the 50 year history we have with Moore's Law and that we should be able to keep it going for 51 or 52 years. So we're gonna be a little careful there about that signaling exactly when, what and where.” --Brian Krzanich, CEO Intel, IM’14

“We do think that we've been giving too much insight too far in advance and so we'll talk about 10nm some time in the next 12 or 18 months [from Nov '14] and when it's appropriate.” Stacy Smith, CFO Intel, IM’14

“But you are seeing in the fourth quarter, you’re seeing the front edge of the startup cost associated with the 10-nanometer and that’s kind of right in line with the historical timing of what you’d expect.” Stacy Smith, CFO Intel, IM’14

If that's not enough, consider this: even Intel, Brian Krzanich doesn't yet know when they'll launch 10nm. Yields may not improve anymore, say from tomorrow on, and then 10nm will never be launched! If yields skyrocket to 99% tomorrow, then you can be sure as [duh] that 10nm will launch as fast as possibluh.

 

[Fjodor2001]

So why did the Intel representative Taha Khalifa say that Intel 10 nm will be released in 2017? Did he lie, make it all up, or intentionally mention a much later launch date than Intel expects?

Why would he do that? Wouldn't it be better to not say anything at all, if he did not know?

My guess it that Taha Khalifa is very aware of the internal Intel release estimates for 10 nm, and he provided the estimate that is commonly accepted within Intel. Whether his statements were sanctioned to become public by Intel HQ or not is another story. Maybe he said more than he was allowed to, but I find it hard to believe his estimate is inaccurate.

 

[witeken]

As I said on the previous page:

Obvious conservatism is obvious.

 

[mikk]

Brian Krzanich also said Cherry Trail is coming late 2014. I wouldn't give him too much credibility.

 

[imported_ats]

Seriously? That's how you want to take my post, like I'm some idiot who couldn't possibly have factored all those points in (plus a few others that aren't common knowledge) before making my synopsis?

:| OK.

Yes, because if you understood the graph, you would understand that Haswell didn't reduce costs over IVB, the 22nm process improvements reduced costs even further during Haswell's life time. And the same graph showed Broadwell reaching or beating parity with Haswell at roughly the start of the 3rd Quarter of ramp.

Its like people have never heard of volume ramp before.

 

[imported_ats]

lmfao, you just told an electrical engineer who works in the CPU industry to learn to read graphs about CPUs.

So what you are saying is that an electrical engineer who's designed some of the biggest most complex chips in the industry and designed the interconnection networks for numerous chips told another EE who also works in the industry his interpretation of a graph was off? Yeah, that actually happens all the time in the industry.

 

[Fjodor2001]

I think the reason Intel is shying away for commenting on 10 nm at Investor's Meetings and alike is that they are scared shitless of a repeat of the 14 nm delay embarrassments. I.e. better to not say anything at all in public forums where they can be held accountable until 10 nm is in the stores more or less.

That does not mean that some old estimate for 10 nm launch made years ago is still valid, it just means they will not comment on it any longer at Investor's Meetings and similar.

 

[Redentor]

I think the reason Intel is shying away for commenting on 10 nm at Investor's Meetings and alike is that they are scared shitless of a repeat of the 14 nm delay embarrassments. I.e. better to not say anything at all in public forums where they can be held accountable until 10 nm is in the stores more or less.

That does not mean that some old estimate for 10 nm launch made years ago is still valid, it just means they will not comment on it any longer at Investor's Meetings and similar.

I agree.

 

[frozentundra123456]

I have to say, I dont really know, but late 2016 or 2017 sounds reasonable for 10 nm to me. It is a bit of semantics perhaps too. We are already a month into 2015, and we dont really have a strong line-up of 14 nm products. However, there were a spattering of 14nm products out before the holidays, so technically intel reached their guidance of 2014 availability. Same could easily happen with 10 nm.

What I am really more concerned with though, is the quality of the products. Being late is one thing, but being late with an underwhelming product is quite the worse situation. So far, I think that would have to describe Broadwell. And I dont think the excuse that it is the formfactors hold up, because the product is late already, so there should have been extra time to refine the rest of the package. The only compelling product that I have seen with Broadwell is the Dell 13 inch laptop, which seems to have adequate performance and exceptional battery life. But until we see a complete test (hint to this site!!) we wont know for sure how good it is and how much improvement is due to Broadwell vs Haswell.

I am also very concerned that Skylake is going to disappoint. Despite intel saying the problems with 14nm are solved, Broadwell is not really that impressive, and Skylake will be built on the same process. It almost seems like intel is getting into what I call the "NCS" that has plagued another company, i.e. "next chip syndrome". It is always something just over the horizon that is going to transform the product line-up. With intel it is graphics and BOM costs that are trying to be improved, but after all the delays we are still in the not quite there yet situation. I fear the same fate for those who were hoping for finally some substantial cpu improvement in Skylake. (I seem to recall the "wait for haswell" when ivb came out, and that turned out into a decent but not spectacular improvement).

And believe me, I take no pleasure in saying this. I dont think anyone could classify me as an intel basher(although there are plenty of those). I am a die hard x86 fan, and would love to see intel take a big bite out of ARM/Android in the mobile field (without having to resort to contra revenue to do it) an make some big performance gains on the cpu side of x86. Unfortunately, all these delays have come at the worst possible time in relation to making inroads into the android/arm ecosystem. It just gives arm more time to refine their process nodes and become even more firmly entrenched with stronger products.

 

[witeken]

Since Dennard scaling died, it's mostly been about the accumulating improvements, which Intel likes to show:

 

[Ajay]

Those are paid off long ago. And 22nm was used longer than planned, so those (using 22nm 6 months longer and 14nm 6 months shorter) cancel each other out.

Well - we can't know that at all yet until 14 nm is shipping for a couple of years. Even then, we'd need a report on Intel's process development financials for both processes, which we are not likely going to get.

 

[Fjodor2001]

Assuming the roadmap in the OP is correct, what could be the reason for Intel providing a different inaccurate roadmap just a few weeks ago at CES 2015?

My guess: Intel wants to give the OEM:s some reason to release Broadwell based products. If it was known to the public that Skylake will be released 4-4.5 months from now, a lot of consumers would skip Broadwell and wait for Skylake. Then OEMs would not have any incentive to release Broadwell based products.

 

[ninaholic37]

We felt like we went on a little early with 14nm as far as timing and performance and features and we saw actually competitors adjust to that. So we're gonna be a little bit more prudent, a little smarter about signaling to the industry exactly when, what and where. And you'll have to trust a little bit the 50 year history we have with Moore's Law and that we should be able to keep it going for 51 or 52 years. So we're gonna be a little careful there about that signaling exactly when, what and where.” --Brian Krzanich, CEO Intel, IM’14

Is he saying that Moore's Law will probably die in 2016 or 2017?

Since Dennard scaling died, it's mostly been about the accumulating improvements, which Intel likes to show:

My old laptop has a dead battery. So I guess 0 x 2 = 0 hours. :awe:

 

[witeken]

Is he saying that Moore's Law will probably die in 2016 or 2017?

Fear not! I've also got quotes to cover that question:

First, from William Holt, the person at Intel in charge (EVP) of Moore's Law:

“Well, let me start with saying that I'm not about to start predicting the end, since anybody who's tried that has been wrong. So I'm not going to try that. The other thing that I'd refer back to is, you know, Craig many years ago said when asked this kind of a question is that, yes there's a wall out there.. somewhere, potentially; and he was going to run into it as fast as he could. So we have no intention of slowing down. If we slow down, it will just be because we can't keep up. So we'll see. The goal is to keep pushing that wall out, and that's what we're doing right now, and as far as hitting, we're not going to slow done because we see it on the horizon.” --William Holt, Intel, IM'13

This one's from Bohr:

“In development and research, we see scaling continues at least another 10 years, which is the same answer we gave 10 and 30 years ago. It’s hard to see beyond 10 years.
Although it gets harder every time, we are still developing technology that’s lower cost-per-transistor than the previous node. I remember when one micron was terrifying to us.
Today our 22 nm process is Intel’s highest yielding, lowest defect technology. In a year or so, our 14 nm process will match that, but it will take a lot of work.” --Mark Bohr, Intel fellow, IEDM 2014

EETimes:“It was an impressive job, and one on which Intel is still working to improve yields (see below). Attendees seemed to agree, packing into the room for the talk and giving the loudest and most spontaneous round of applause I heard at IEDM [2014].”

Two other semi-random comments:

The mission is to really utilize Moore's Law. We have it. We believe we lead at it. We drive it. We define Moore's Law as a company.” --Brian Krzanich, CEO Intel, IM’14

“We will not take the foot off the [Moore's law] pedal here.” --Brian Krzanich, CEO Intel, IM’14

 

[Ajay]

Assuming the roadmap in the OP is correct, what could be the reason for Intel providing a different inaccurate roadmap just a few weeks ago at CES 2015?

My guess: Intel wants to give the OEM:s some reason to release Broadwell based products. If it was known to the public that Skylake will be released 4-4.5 months from now, a lot of consumers would skip Broadwell and wait for Skylake. Then OEMs would not have any incentive to release Broadwell based products.

Yeah, it seems that Intel want's some ROI on BW, even though Skylake's development has been excellent (from what Intel said last year). From what
witeken has posted, neither BW or SKL will be fully ramped, process wise, till Q2 - so we are waiting till around Q3 for the best products from each CPU line. It seems likely that OEM pressure will be to push out SKL products unless Intel discounts BW or staggers SKUs to slot BW into key areas.

Intel could hold back Skylake specs, from the public, till Q3 just to make more room for BW. The those who buy desktops for performance (gamers, etc.) can jump in on SKL for the rest of the year.

 

[Fjodor2001]

@Witeken: You do know BK was in total denial of 14 nm delays until forced to admit it when reality caught up with him, and then admitting one delay after another as products could not be released on time as promised earlier? I'd not put much faith in his comments, and I think you are naive to blindly accept what he has stated some time ago for a fact, and that roadmaps cannot change over time. I think you're more clever than that to be honest.

 

[witeken]

I've never said that roadmaps can't change, quite the opposite in fact. But so far all evidence suggests a Q4 ramp. 10nm was always researched and developed with a '15 ramp in mind, so it is shouldn't be a surprise. A delay of course also wouldn't be a surprise.

 

[Fjodor2001]

But so far all evidence suggests a Q4 ramp. 10nm was always researched and developed with a '15 ramp in mind, so it is shouldn't be a surprise.

Hmmm... I'm not sure what you're basing that on. As I understand it, Intel presented some 10 nm roadmap long ago, and has since refused to comment on it. That does not make it certain that the original release date estimate is still valid, does it?

 

[Abwx]

@Witeken: You do know BK was in total denial of 14 nm delays until forced to admit it when reality caught up with him, and then admitting one delay after another as products could not be released on time as promised earlier? I'd not put much faith in his comments, and I think you are naive to blindly accept what he has stated some time ago for a fact, and that roadmaps cannot change over time. I think you're more clever than that to be honest.

The fact that it s one year that they are ramping their 14nm and that only a 82mm2 chip is shipped in marginal quantities doesnt seem to entail some people s beliefs, now we have expectations that this decidely low production node will be amortized and replaced by an even more costly and complexe node in a matter of 18 months after 14nm 4C are introduced.

 

[III-V]

Yes, because if you understood the graph, you would understand that Haswell didn't reduce costs over IVB, the 22nm process improvements reduced costs even further during Haswell's life time. And the same graph showed Broadwell reaching or beating parity with Haswell at roughly the start of the 3rd Quarter of ramp.

Its like people have never heard of volume ramp before.

Considering Idontcare has a PhD, and was a process engineer for Texas Intstruments, he certainly knows the economics of semiconductor fabrication fab better than you.

In all likelihood, either you've misunderstood what he was saying, or he did not communicate his thoughts effectively. To be calling him ignorant, however, is really rather laughable.

 

[DrMrLordX]

My old laptop has a dead battery. So I guess 0 x 2 = 0 hours. :awe:

El oh el.

Sounds like a winner. All it needs is a snazzy OEM name like "Bungbungame".

 

[Abwx]

Considering Idontcare has a PhD, and was a process engineer for Texas Intstruments, he certainly knows the economics of semiconductor fabrication fab better than you.

In all likelihood, either you've misunderstood what he was saying, or he did not communicate his thoughts effectively. To be calling him ignorant, however, is really rather laughable.

I wont negate someone s merits knowledge wise but in this, trivial, very case Imported ATS was right, i also pointed that this was erroneously interpreted in post64.

http://forums.anandtech.com/showpost.php?p=37127778&postcount=64

 

[mikk]

Here is an older version of this Roadmap. Posting this because there are more descriptions in it. The left edge represents the earliest RTS schedule plan. We can see for Skylake-S RTS plan was July-October. Skylake-S plan didn't change in the newer Roadmap from Vr-zone. Skylake-U in that Desktop Roadmap plan accelerated apparently. By the way: Skylake-U in a Desktop Roadmap? Aimed for NUC and similar devices or why is this? Skylake-U for Ultrabooks shouldn't differ though.

 

[witeken]

At least that confirms that CNL-S won't launch until at least Q3'16. But it's quite an old roadmap, when BDW-E wasn't yet delayed.