intel delays 14nm?-process problems

[piesquared]

Not working? This could be the great equalizer.

http://www.xtremesystems.org/forums/...4nm-in-Ireland

Uh oh.

[nforce4max]

Eventually they're going to hit more and more problems with every shrink and development cycles will get longer. With each cycle it will get more and more costly for Intel to make each new generational step. It will drive costs of their products higher while AMD is failing due to the constrains and complications that come from using SOI.

[mrmt]

Originally Posted by piesquared

Not working? This could be the great equalizer.

The process introduction was just delayed in Ireland, this has nothing to do with the health of the process.

[ShintaiDK]

Intel got atleast 3 14nm fabs. And Fab42 is an entirely new plant with only 14nm.

Ireland was also the last announced.

A slowdown in demand has prompted Intel to delay the introduction of 1272 manufacturing to its Irish subsidiary by six months, pushing it out to late next year.

Thats one view. Another could be they cant get enough tools ready. 14nm is one of the fastest conversions.

I assume piesquared also think 32nm was in deep trouble.



Sorry there piesquared. AMD just aint gonna return.

[piesquared]

Originally Posted by mrmt

The process introduction was just delayed in Ireland, this has nothing to do with the health of the process.

No offense, but these explanations seem more plausible,

Market understands Intel's product roadmap troubles. Qualcomm larger market value might be correct.

Intel killing itself with pursuit of Moore's Law for no business advantage. I can confirm (ex-mobile designer) 14nm is late, not working. 1272/14nm design shuttle tape outs all pushed out due to problems with process.
14nm has many issues and seriously now SOI being possible fix.

At this stage design teams have unstable 0.1 silicon models. No meaningful Baseband processor design can be started based on current state of 14nm.

I hear Intel's engineering team came clean and concluded current approach to 14nm would not work after spending $2B.

Plan was to ship 14nm parts in 2013 and now that plan has been scrapped. 0% chance of shipping a single 14nm chip in 2013.

On top of that Intel marketing this year was claiming 22nm was not competitive in market but "just wait until 14nm".

Intel now plans to to ship LTE chips fabricated at TSMC for next 2 years. However, TSMC is business smart. It is not going to give Intel a good wafer price over its long term customers like Qualcomm that buys many more chips in the chipset (wifi, LTE, RF and SOCs chips) unlike Intel LTE with Atom fabricated by an Intel internal fab.

http://www.eetimes.com/electronics-n...d-chip-company

[nforce4max]

Originally Posted by mrmt

The process introduction was just delayed in Ireland, this has nothing to do with the health of the process.

I still want single atom transistors even though there has been functional examples since the 1990s (Ti and several government labs) that were developed as proof on concept.

[mrmt]

Originally Posted by piesquared

No offense, but these explanations seem more plausible

Really? First 14nm chips weren't due 2013 but 2014. Second why aren't we seeing the same movements in the US fabs, just in Ireland? Third, where did the author got that 2 billion number? 2 billion is a small fraction of the new 14nm fabs, the investment is far bigger than 2 billion.

I remember reading the same thing about Intel 22nm, but what we saw in the end was the fastest process ramp up in Intel history.

[ShintaiDK]

Note he is quoting a random comment user. One that writes alot like what piesquared dreams about

But again, who ever took this guy serious.

[Idontcare]

Originally Posted by piesquared

No offense, but these explanations seem more plausible,

More plausible? That is called confirmation bias.

You are scouring the internet looking for little more than anonymous comments posted at the end of unrelated articles in hopes of finding something that supports your preferred world view.

Confirmation bias (also called confirmatory bias or myside bias) is a tendency of people to favor information that confirms their beliefs or hypotheses.[Note 1][1] People display this bias when they gather or remember information selectively, or when they interpret it in a biased way. The effect is stronger for emotionally charged issues and for deeply entrenched beliefs.

[piesquared]

Originally Posted by Idontcare

More plausible? That is called confirmation bias.

You are scouring the internet looking for little more than anonymous comments posted at the end of unrelated articles in hopes of finding something that supports your preferred world view.

You know what they say about assumptions..

In accuality, I read the thread at the xtremesystems link which is where I read the quotes.

[piesquared]

Originally Posted by ShintaiDK

Note he is quoting a random comment user. One that writes alot like what piesquared dreams about

But again, who ever took this guy serious.

What's taking me serious have to do with somebody else's comment??*twirlingfingerontemple*

[CHADBOGA]

Originally Posted by piesquared

No offense, but these explanations seem more plausible,

http://www.eetimes.com/electronics-n...d-chip-company

From the comments you have quoted attached to that link, is also the following:

AMD have decided that they are going to leave the Semi-conductor market because they know that they can't make competitive products anymore and that Kaveri is a woeful failure and are embarrassed by the conduct of their fanboys.

AMD bankrupt Q1 2014

Just how seriously are we meant to take anonymous comments on some website?

[Ferzerp]

IDC, I vote that you forcibly make this picture this guy's avatar image.

[jpiniero]

Yeah this is totally about demand. Could be a sign they are pessimistic about Airmont's ability to drive smartphone sales.

[bononos]

Originally Posted by CHADBOGA

....

......

The great equalizer indeed (OP's comment).

[BD231]

What reason does intel have to switch to 14nm when they have zero competition and their current parts will remain competitive for years to come. Zero.

[boxleitnerb]

Originally Posted by BD231

What reason does intel have to switch to 14nm when they have zero competition and their current parts will remain competitive for years to come. Zero.

No. Intel has the obligation to their investors and shareholders to make money. You don't make money if you don't sell products. And you don't sell products if they don't advance. They could stick with 22nm and just make the chips larger. Oh wait...margins going down the toilet (relatively speaking).

[cytg111]

its friggin insane this talk about competition, sales and demand blablabla .. there must be, atleast, a billion factors involved in maximizing profits, wich intel, of course, is all about.
If intel has no competition, they will, of course, fill a performance segment with their product before they introduce a new one (or sales declines to a level where the cost benefit ... is not beneficial). It is not friggin hard to understand, Intel is not evil, do'er of all good or anything in between, its a friggin company that specializes in profit. Wtf dudes.

[BD231]

Originally Posted by boxleitnerb

No. Intel has the obligation to their investors and shareholders to make money. You don't make money if you don't sell products. And you don't sell products if they don't advance. They could stick with 22nm and just make the chips larger. Oh wait...margins going down the toilet (relatively speaking).

Oh ok so what you're telling me is Intel doesn't release a " new product " on the same process anymore /ignore

Originally Posted by cytg111

its friggin insane this talk about competition, sales and demand blablabla .. there must be, atleast, a billion factors involved in maximizing profits, wich intel, of course, is all about.
If intel has no competition, they will, of course, fill a performance segment with their product before they introduce a new one (or sales declines to a level where the cost benefit ... is not beneficial). It is not friggin hard to understand,Intel is not evil, do'er of all good or anything in between, its a friggin company that specializes in profit. Wtf dudes.

Are you done?

That was completely unnecessary.
-ViRGE

[boxleitnerb]

[redacted]
I didn't say that at all. Obviously they will still stand by their tick-tock model which means two product lines for each process node. But what you said sounded like Intel could stick with 22nm for much longer, way beyond Broadwell. I explained to you why that won't be the case.

[Sable]

Originally Posted by CHADBOGA

From the comments you have quoted attached to that link, is also the following:



Just how seriously are we meant to take anonymous comments on some website?

[Lonbjerg]

Originally Posted by BD231

What reason does intel have to switch to 14nm when they have zero competition and their current parts will remain competitive for years to come. Zero.

Google: stockholders

The value of your post: Zero

[StrangerGuy]

Welcome to peak silicon!

[KompuKare]

Originally Posted by StrangerGuy

Welcome to peak silicon!

Well not quiet yet but Intel will eventually have to took-took-tick or worse. A shrink every two years cannot be sustained forever.

The other thing is that their die sizes have been going down (4 core):
Nehalem 296 mm² (Wiki lists the same area for Clarksfield & Lynnfield)
Sandy Bridge 216 mm²
Ivy Bridge 133 mm2
so assuming that their wafers stay the same (unsure when 450mm is due), that means the number of chips they can supply per plant has risen (assuming the processes required at 14nm for instance don't require more steps or steps which are more time consuming).

So they must have calculated that at least at the beginning of 14nm ramp-up they do not need Leixlip. 22nm>14nm is a full node,isn't it? Therefore if they had the same amount of fabs on 14nm and were able to do process the same amount of wafers they would be able to supply twice the amount of transistors and unless they blow all that transistor budget on new features, even Intel may not be able to generate that amount of additional volume.

[piesquared]

Besides stockholders to answer to, there is also the ARM stampede to battle. Using that tired old meme of intel not have any competition to explain intelss execution problemss has gotten ridiculous.