(Semiwiki) Intel 14nm Delayed Again?

[Fjodor2001]

It was a defect density issue. The last thing that Intel wants is to lose its process advantage.

Then why did Intel make an official statement saying:

"[...] the decision to postpone mass production of Broadwell was because of slow demand for personal computers in general [...]"

Sure, Intel does not want to lose it's process tech advantage. But as long as they keep the lead sufficiently large, they can release new node tech at a rate that brings the most profit and return on invested R&D money. I.e. make sure to sell enough 22 nm products before releasing 14 nm. Remember, just because they delay release of 14 nm does not mean it will delay 10 nm and beyond. They can keep researching that at full speed in parallel, so they have it ready "in store" to release when desired to keep the competition at a sufficient distance.

 

[ShintaiDK]

It took ~20 months from Haswell tapeout to release. Sandy Bridge took ~19 months.

If someone needs someone to compare with. But kinda kills the drama...

 

[Homeles]

I think 7nm will use III-V, maybe already at 10nm, if we can believe Paul Otellini:


Source: http://apcmag.com/intel-looks-beyond-silicon-for-processors-past-2017.htm (2009)

Germanium-based would come first, and would still be be "post-silicon."

 

[Fjodor2001]

It took ~20 months from Haswell tapeout to release. Sandy Bridge took ~19 months.

If someone needs someone to compare with. But kinda kills the drama...

So since Broadwell was taped out in September 2012, it ought to be released about 19 months after that, i.e. April 2014. But Broadwell-K won't be released until 2015Q1. I.e. almost 1 year late. So yeah, there's some drama...

 

[witeken]

Germanium-based would come first, and would still be be "post-silicon."

I know, but it is also said in the article that Intel claims a 10x lower power consumption and 1,5x higher speeds with III-V indium antimonide, so I assumed that they meant III-V materials, not a 1 or 2 node solution (and I'm not sure if you can conside SiGe actually post-silicon). Did they actually know in 2009 that they would need to use SiGe at 14nm before they could switch to III-V materials at 10 or 7nm?

 

[witeken]

So since Broadwell was taped out in September 2012, it ought to be released about 19 months after that, i.e. April 2014. But Broadwell-K won't be released until 2015Q1. I.e. almost 1 year late. So yeah, there's some drama...

Broadwell-K will be released in Q3, won't it?

 

[Homeles]

So since Broadwell was taped out in September 2012, it ought to be released about 19 months after that, i.e. April 2014. But Broadwell-K won't be released until 2015Q1. I.e. almost 1 year late. So yeah, there's some drama...

Where are you getting this 2015 crap? It's still Q4 2014 until Intel or reasonably reliable sources state otherwise.

I thought we already buried the hatchet in this discussion, but clearly this conspiracy theory excites you. Let go already.

Broadwell-K will be released in Q3, won't it?

Broadwell in general is set for availability in Q3, Broadwell-K is set for Q4, if I remember correctly.

I know, but it is also said in the article that Intel claims a 10x lower power consumption and 1,5x higher speeds with III-V indium antimonide, so I assumed that they meant III-V materials, not a 1 or 2 node solution (and I'm not sure if you can conside SiGe actually post-silicon). Did they actually know in 2009 that they would need to use SiGe at 14nm before they could switch to III-V materials at 10 or 7nm?

I do consider SiGe post-silicon. What matters is that the industry moves to a better channel material than pure silicon.

What I have been thinking is that Intel will stay with Silicon at 14nm. I don't think they're ready for it, and I think they'd be making noise about it. I fully expect 14nm to be a tweaked, shrunken version of 22nm, without any surprises. I believe they'll move to SiGe or Ge at 10nm. From the sounds of it however, they're pushing InSb. I don't really know what they're going to choose, but they're ready to move onto something else, and they'll be doing it soon. I think they'll be moving to TFETs soon as well.

 

[Fjodor2001]

Where are you getting this 2015 crap?

It was 2014Q4, but that was before the latest delay. So now it's 2015Q1.

 

[Homeles]

The latest delay, which is coming from an unreliable source, yes?

 

[witeken]

Then why did Intel make an official statement saying:

"[...] the decision to postpone mass production of Broadwell was because of slow demand for personal computers in general [...]"

Look at the source of your link, which makes it more clear: Intel May Introduce First “Broadwell” Chips in Q3 – Report.

The decision to postpone mass production of Broadwell was conditioned not only by slow demand for personal computers in general and microprocessors in particular, but also by yields that were below Intel’s comfortable level.

Followed by BK's official statement in which he concludes that "It was simply a defect density issue." I don't know where the part of "slow demand..." comes from, but Intel only said that it's because of defect density issues, so it was probably a rumor.

Sure, Intel does not want to lose it's process tech advantage. But as long as they keep the lead sufficiently large, they can release new node tech at a rate that brings the most profit and return on invested R&D money. I.e. make sure to sell enough 22 nm products before releasing 14 nm. Remember, just because they delay release of 14 nm does not mean it will delay 10 nm and beyond. They can keep researching that at full speed in parallel, so they have it ready "in store" to release when desired to keep the competition at a sufficient distance.

I hope it doesn't delay 10nm, but according to your logic there should be a delay because Intel misses 1Q of missed Broadwell/Cherry Trail sales. We'll see if 10nm goes into production in 2015 or Q1 2016 as you'd expect without delays.

Personally, I think that if a process is ready for volume production, it should start as soon a possible: all the fixed costs of a new process node are done (e.g. R&D and equipment), so there is no reason to delay the new node because that new node will bring huge cost/transistor improvements.

While googling, I also found this interesting article: Intel: We know how to make 10nm chips
Intel wants to use EUV for 10nm, but it probably won't be ready.

 

[Fjodor2001]

Look at the source of your link, which makes it more clear: Intel May Introduce First “Broadwell” Chips in Q3 – Report.

Followed by BK's official statement in which he concludes that "It was simply a defect density issue." I don't know where the part of "slow demand..." comes from, but Intel only said that it's because of defect density issues, so it was probably a rumor.

I think Intel said both. From where did you get that Intel did not say the part "the decision to postpone mass production of Broadwell was because of slow demand for personal computers in general" too? Because it's been reported by a number of credible sources.

I hope it doesn't delay 10nm, but according to your logic there should be a delay because Intel misses 1Q of missed Broadwell/Cherry Trail sales. We'll see if 10nm goes into production in 2015 or Q1 2016 as you'd expect without delays.

Personally, I think that if a process is ready for volume production, it should start as soon a possible: all the fixed costs of a new process node are done (e.g. R&D and equipment), so there is no reason to delay the new node because that new node will bring huge cost/transistor improvements.

http://www.pcmag.com/article2/0,2817,2453384,00.asp

The interesting part is that DigiTimes did not claim there was another glitch or any other technical reason for the rumored delay. Instead, Intel is supposedly hanging back on releasing Broadwell due to "the slow digestion of Haswell processor inventories in the market," according to the tech site.

If so, analyst and former Intel employee Ed McKernan said such a decision could make sense.

"It is unlikely in Intel's history to hold back from launching new processors on a new process technology, regardless of the yield. The ASPs of the new processors typically are very high, and deliver gross margins that pay off even when the yield is low. Also, it was always necessary to launch a new CPU in a timely fashion to keep the treadmill running in order to stay ahead of AMD," McKernan wrote on Seeking Alpha.

But with the "threat of AMD ... gone" as Intel's rival shifts towards graphics-based client solutions, Intel now has some more leeway as to when to ramp the successor to its 22nm Hawell products, he said.

"Intel could very easily stretch out its current 22nm Haswell for quarters and not see a downturn in PC market share, all the while increasing manufacturing margin," McKernan said."

 

[Homeles]

DigiTimes isn't reliable either.

 

[Fjodor2001]

The latest delay, which is coming from an unreliable source, yes?

It's been cited by a number of credible sources:

http://www.electronicsweekly.com/news/general/intels-14nm-mobile-delayed-till-2015-2014-02/

http://www.techspot.com/news/55676-rumor-intel-broadwell-cpus-delayed-until-late-2014.html

http://www.extremetech.com/computin...-intel-delayed-14nm-deployments-until-q4-2014

Just to name a few.

Apparently only the low power U and Y series mobile Broadwell CPUs will now be released in 2014Q4, and Broadwell-K in 2015Q1.

 

[witeken]

I think Intel said both. From where did you get that Intel did not say the part "the decision to postpone mass production of Broadwell was because of slow demand for personal computers in general" too? Because it's been reported by a number of credible sources.

I don't know, but I think BK's words ("simply a defect density issue") are more credible than your credible sources.

http://www.pcmag.com/article2/0,2817,2453384,00.asp

Doesn't this just confirm that Intel won't delay a new process if it's ready? So if Intel starts volume production in Q1 like they said we'll see Broadwell in Q3, not in Q1 2015.

 

[witeken]

It's been cited by a number of credible sources:

http://www.electronicsweekly.com/news/general/intels-14nm-mobile-delayed-till-2015-2014-02/

http://www.techspot.com/news/55676-rumor-intel-broadwell-cpus-delayed-until-late-2014.html

http://www.extremetech.com/computin...-intel-delayed-14nm-deployments-until-q4-2014

Just to name a few.

Apparently only the low power U and Y series mobile Broadwell CPUs will now be released in 2014Q4, and Broadwell-K in 2015Q1.

An unreliable rumor being quoted by reliable sources doesn't make it reliable.

 

[Homeles]

Yep.

 

[Fjodor2001]

An unreliable rumor being quoted by reliable sources doesn't make it reliable.

Those credible sources wouldn't quote it if they thought it was unreliable.

 

[Fjodor2001]

I don't know, but I think BK's words ("simply a defect density issue") are more credible than your credible sources.

But are you sure that's what he actually said? How do you know he didn't say the other part as well ("the decision to postpone mass production of Broadwell was because of slow demand for personal computers in general"), which has been quoted by many credible sources. Are you saying all those sources just made that up? And if so, what proof do you have of that?

Doesn't this just confirm that Intel won't delay a new process if it's ready? So if Intel starts volume production in Q1 like they said we'll see Broadwell in Q3, not in Q1 2015.

There are several reasons for why Intel would intentionally delay the release of 14 nm. Some of which was quoted in the last section of my previous post.

 

[jpiniero]

The Digitimes article suggests that this second delay into mostly 2015 is also being caused by poor sales of Haswell.

 

[witeken]

But are you sure that's what he actually said? How do you know he didn't say the other part as well ("the decision to postpone mass production of Broadwell was because of slow demand for personal computers in general"), which has been quoted by many credible sources. Are you saying all those sources just made that up? And if so, what proof do you have of that?

Yes, I'm sure. Here is a quote from Intel's conference call for Q3 in October:

Operator
Our next question comes from the line of John Pitzer of Credit Suisse. Your line is open, sir.

John Pitzer - Credit Suisse
Brian just a follow-up on the Broadwell. I am just kind of curious why the push out? Can you elaborate a little bit? Is this is technical issue and if it is, why are you confident you overcame it or is this really a market issue? I.e. there's a little bit excessive Ivy Bridge inventory which is pushing out Haswell, which is the ramp, which is pushing Broadwell. If you could elaborate on that that would be great.

Brian Krzanich - Chief Executive Officer, Director
Sure. It's absolutely not the latter. It was simply a defect density issue. This was on the issue -- as we develop these technologies, what you are doing? You are continually improving the defect densities and those resulted in the yield, the number of die per wafer that you get out of the product and what happened as you insert a set of fixes in groups, you will put four or five, maybe sometimes six or seven fixes into a process and group it together, run it through and you will expect an improvement rate occasionally as you go through that. The fixes don't deliver all of the improvements [stock], we had one of those.

Why do I have confidence? Because, we have got back now and added additional fixes, gotten back onto that curve, so we have confidence that the problem is fixed, because we have actually data and defects and so that gives us the confidence that we are to keep moving forward now and that happens sometimes in these development phases like this, so that's why we are going to over it a quarter.

Member of Broadwell and Haswell are incompatible for the unit for the most part these will slide into existing systems and it delivers the next generation of the low-power Broadwell. Why? Which is you give even more capability for fanless core products, so we and our OEM partners have a strong desire to get Broadwell to the market, so if I could there would be nothing slowing down. This is a small blip in schedule and we will continue on from here.


I didn't even think he said it so clear.

Source: http://seekingalpha.com/article/174...sses-q3-2013-results-earnings-call-transcript (CTRL+F keyword: yield)

 

[Homeles]

Those credible sources wouldn't quote it if they thought it was unreliable.

Well, they're wrong.

http://techland.time.com/2012/05/14/digitimes-apple-rumors/

Will you ever take no for an answer?

 

[Fjodor2001]

Will you ever take no for an answer?

Will you?

 

[Homeles]

I'll take that as a no.

 

[Meghan54]

Those credible sources wouldn't quote it if they thought it was unreliable.

Really? You find copying and pasting rumors from an admittedly dubious source on various websites without fact checking makes those websites credible? Amazing.

 

[jpiniero]

Intel could release only the ULT models sometime in the 4Q and push the others into 2015 and not contradict what they have been saying though. I don't even think they ever committed Broadwell to Q3 anyway, it was always "later in 2014".