14nm GF volume production with low double-digit yields? Apparently yes.

[witeken]

IEDM: TSMC, Intel and IBM 14/16nm Processeshttps://www.semiwiki.com/forum/content/4110-iedm-tsmc-intel-ibm-14-16nm-processes.htmlhttps://www.semiwiki.com/forum/content/4110-iedm-tsmc-intel-ibm-14-16nm-processes.html
Samsung and Globalfoundries Beat TSMC to FinFET – What This Means For AMD, Nvidia and Intel.

It's quite pathetic, really. Here goes my rant. TL;DR in title (according to the interview, GF HVM is actually in H1, contrary to what WCCF Tech says).

TSMC were up first. They talked about the improvements that they had made going from their 16FF to the second generation 16FF+ under the title An Enhanced 16nm CMOS Technology Featuring 2nd Generation FinFET Transistors and Advanced Cu/low-k Interconnect for Low Power and High Performance Applications. They already reported on the basic 16FF process last year so this is an update.

Note the 2nd generation. What the deal with that? Remember how TSMC, IBM, Samsung and Global Foundries, in a desperate act to deceive people and deny Intel's momentous Research & Development prowess, together decided to skip a shrink and call it a new node, forcing Intel to educate the public -- consisting of few investors and tech enthusiasts -- that they've been shipping the, suddenly worshiped, FinFET technology since 2011 and have now started volume -- real volume, that is -- production of a second and much improved gen of FinFET innovation; accidentally called 14nm.

Now apparently, in a sudden act of disguised comeback to further mystify Intel 4 year industry leadership advantage, TSMC made a few little changes to their process and called it a day.

Just too bad many people don't realize what the numbers mean or even bother thinking about the implications of the following:

They admitted that they have had yield problems, which is public knowledge. 22nm is the highest yielding process in Intel history and 14nm is now almost at the same level. It is shipping in volume.

...After already being informed about...

There is anticipation of some minor ramp that we’re aware of in the fourth quarter. There is capacity in place currently. We have equipment that has been relegated to the 14 and 16 nanometer note, primarily 14. And so at this time we have capacity in place to take care of that need as they begin to ramp slowly. So we don’t see a significant ramp in Q4 in FinFETs. We see that occurring later or in 2015 and it’s really hard to project when. The current anticipation is they’re all being overly optimistic as to when they’re going to solve their problems. But the yields on the major companies right now is in the 10% to 20%. And so it's not giving them much indication as to when they’ll grow that problem, that area. So it appears to me right now and the problems aren’t really consistent in one area. They vary in the processing and possibly design.

I'm happy that I'm not the Apple financial guy who's in charge of all the low-yielding wafers Apple buys.

If the delusion wasn't yet complete...

Both Globalfoundries and Samsung beat TSMC to FinFETs with their 14nm LPE and LPP nodes. Announcing that mass production has already begun. This news comes courtesy of a Digitimes interview with GLobalfoundries’ Senior Vice President Chuck Fox.

The cooperative effort between both companies means that 4 worldwide fabs will have full 14nm LPE and LPP manufacturing capability in 2015. With a number of those fabs already shipping 14nm products in volume. This is especially important because TSMC has only recently managed to achieve risk qualification on its 16nm FinFET+ process. With volume production scheduled for 2H 2015, putting it behind Samsung by nearly eight months.

[...]

So what does this all mean for Red, Green and Blue. I’m referring to AMD, Nvidia and Intel naturally.
Intel has already began shipping small, low power 14nm products this year with high performance parts coming in the middle of next year. So the company isn’t significantly ahead of the industry as it was once in terms of process nodes. Which is very exciting for the rest of the industry players. Because it puts everyone on a relatively equal playing field.

Indeed, I could not have said it any better. The marketing insanity put everyone on an equal playing field.

Everyone; not everything.

 

[Arachnotronic]

Lol @ mass production of 14nm at GloFo...suuuure.

 

[waffleironhead]

Where in your links is the "low double digit yields" part? I read thru a few times and missed it.

 

[Abwx]

Still, analysts had concerns regarding Samsung's persistently low FinFET yield rate. "Samsung's yield has been around 30-35% since the beginning of this year," Peng said. "We haven't seen any improvement. Apple and Qualcomm will shift more of their orders to TSMC if it can provide enough capacity."

http://www.eetimes.com/document.asp?doc_id=1324617


And from the OP links :

https://www.semiwiki.com/forum/content/4110-iedm-tsmc-intel-ibm-14-16nm-processes.html

Bottom line: Intel is ahead (by their own reckoning). IBM has the most perfect process for server processors.

IBM talked about their High Performance 14nm SOI FinFET CMOS Technology with 0.0174μm2 embedded DRAM and 15 Levels of Cu Metallization. Of course this is a process that GlobalFoundries will take over when the acquisition of IBM's semiconductor division is complete.

 

[witeken]

Where in your links is the "low double digit yields" part? I read thru a few times and missed it.

http://seekingalpha.com/article/259...-results-earnings-call-transcript?part=single

But the yields on the major companies right now is in the 10% to 20%.

 

[Maxima1]

...............Samsung..............
..................../..\....................
................../......\..................
................/.WIN!.\.................
............../_______\...............
....Glofo...................AMD......

 

[positivedoppler]

n a desperate act to deceive people and deny Intel's momentous Research & Development prowess, together decided to skip a shrink and call it a new node, forcing Intel to educate the public

Wow, Intel calling themselves momentous prowess and belittling that the general public like me needs to be educated? If Intel is really this full of itself it's going to make me want to throw up. No doubt Intel is an industrial leader, but the amount of Narcissism in that paragraph...really that's how Intel think?

 

[witeken]

Wow, Intel calling themselves momentous prowess and belittling that the general public like me needs to be educated? If Intel is really this full of itself it's going to make me want to throw up. No doubt Intel is an industrial leader, but the amount of Narcissism in that paragraph...really that's how Intel think?

Now sure why use Intel instead of the author the that paragraph,

but you don't seem to understand what general public means. The general tech public knows little more than the fact that a lower nm name is better, and since the names are the same, Intel obviously can't be much ahead.

I guess it's called overcompensating.

 

[ShintaiDK]

Wow, Intel calling themselves momentous prowess and belittling that the general public like me needs to be educated? If Intel is really this full of itself it's going to make me want to throw up. No doubt Intel is an industrial leader, but the amount of Narcissism in that paragraph...really that's how Intel think?

Its not an statement from Intel is it.

 

[soccerballtux]

this is all the more reason to buy FX-8310 over core i3. I can't wait for the performance upgarde when my CPU gets the 16nm software update.

 

[ShintaiDK]

this is all the more reason to buy FX-8310 over core i3. I can't wait for the performance upgarde when my CPU gets the 16nm software update.

AM3+ isnt getting any new CPUs. Wrong socket. Also you have to wait till 2016 or later.

 

[soccerballtux]

Now sure why use Intel instead of the author the that paragraph,

but you don't seem to understand what general public means. The general tech public knows little more than the fact that a lower nm name is better, and since the names are the same, Intel obviously can't be much ahead.

I guess it's called overcompensating.

the smaller the process lead gets, the more they have to compensate...

 

[ShintaiDK]

the smaller the process lead gets, the more they have to compensate...

But the lead is expanding. Not to mention Intel is still the only one with lower transistor cost below 28nm.

 

[witeken]

the smaller the process lead gets, the more they have to compensate...

For the sake of clarity, I meant that maybe I'm just overcompensating the diminishing process lead propaganda with some hyperbolic statements .

 

[NTMBK]

You should chill about the deceitful PR statements, it's bad for your health to take it so serious. Marketing is marketing, they will say whatever they can get away with to big up the company. Proof is in the pudding: if they claim to be on par with Intel, they need to provide products to show that. Truth will out in the end, it always does.

Merry Christmas witeken :thumbsup:

 

[monstercameron]

You should chill about the deceitful PR statements, it's bad for your health to take it so serious. Marketing is marketing, they will say whatever they can get away with to big up the company. Proof is in the pudding: if they claim to be on par with Intel, they need to provide products to show that. Truth will out in the end, it always does.

Merry Christmas witeken :thumbsup:

So who killed Kennedy?...some truths never get out.

 

[NTMBK]

So who killed Kennedy?...some truths never get out.

<----P&N is that way

 

[witeken]

You should chill about the deceitful PR statements, it's bad for your health to take it so serious. Marketing is marketing, they will say whatever they can get away with to big up the company. Proof is in the pudding: if they claim to be on par with Intel, they need to provide products to show that. Truth will out in the end, it always does.

But it isn't just PR, tech sites are also proclaiming it, as you can see in the last quote.

Merry Christmas :thumbsup:

 

[ShintaiDK]

But it isn't just PR, tech sites are also proclaiming it, as you can see in the last quote.

"tech sites"?

 

[teejee]

But the lead is expanding. Not to mention Intel is still the only one with lower transistor cost below 28nm.

I really doubt this "lower transistor cost" for Intel includes all the investments. Or am I wrong?
And 14nm probably has much higher investment cost than earlier nodes.
The foundries on the other hand of course include the investment cost in their wafer price.

 

[Abwx]

Proof is in the pudding

It would be cautious to let someone else taste the thing, particularly if it s one of thoses infamous Christmas puddings...

 

[ShintaiDK]

I really doubt this "lower transistor cost" for Intel includes all the investments. Or am I wrong?
And 14nm probably has much higher investment cost than earlier nodes.
The foundries on the other hand of course include the investment cost in their wafer price.

They do have lower cost. Same with 22nm.

Its mainly about tools and design cost. For CPUs we talk billion $ design cost to archive the lower cost.

EUV is the saving grace, but EUV is still not here. Whenever EUV comes, then the 28nm wont be the most cost effective node in terms of transistor cost for the foundries anymore.

 

[witeken]

I really doubt this "lower transistor cost" for Intel includes all the investments. Or am I wrong?
And 14nm probably has much higher investment cost than earlier nodes.
The foundries on the other hand of course include the investment cost in their wafer price.

Investment is just part of the standard costs of a tech company, if you're a healthy company. Cost per transistor really is all about Moore's Law. You multiply the amount of transistor per mm² by the price per mm². If you can't reduce that number, even if you can have 1.9x more transistor on your die, the die will cost 1.9x as much, so capabilities may go up, but costs won't (and ultimately the customer will pay for that), and that's concerning because no one wants 3 billion transistors, each one at the price of one 10µm transistor.

 

[Chevron]

AM3+ isnt getting any new CPUs. Wrong socket. Also you have to wait till 2016 or later.

People keep saying this, but has there been any official word from AMD to say no Excavator(or anything else) for AM3+? Or are people just going on rumours?

 

[NTMBK]

People keep saying this, but has there been any official word from AMD to say no Excavator(or anything else) for AM3+? Or are people just going on rumours?

They also haven't officially denied that their next gen CPU will be made from vanilla pudding. Doesn't mean it's going to happen.