Samsung Claims Mass Production on 14-Nanometers

[Arachnotronic]

We have started mass production wafers in 14-nanometer. So, in 14-nanometer, our progress is well on track, including process and the yields. Here we're talking when I mention yields being on track, this is for a very advanced product. It'll be our most advanced applications processor. Yielding at full specifications targets with all of the characteristics for speed and power as well.

http://engage.vevent.com/index.jsp?eid=1557&seid=76

Interesting and well-chosen phrasing with "yielding at full specifications targets"...doesn't say much about how many of the die actually work. This is, in effect, saying that the ones that do work...work to spec

Probably worth some discussion...

 

[oobydoobydoo]

isn't apple using Samsung for A9/X? Thats probably what Samsung is producing. Apple always likes to pay extra for the most advanced process.

 

[witeken]

More than half a year earlier than TSMC. I guess TSMC's Apple win didn't last too long and we will see FinFET products in 2015. Not bad, I didn't expect they would still start HVM in Q4 like they promised ~a year ago. Doesn't mean that my signature quote doesn't hold anymore, though.

 

[Arachnotronic]

isn't apple using Samsung for A9/X? Thats probably what Samsung is producing. Apple always likes to pay extra for the most advanced process.

It's Exynos.

 

[III-V]

isn't apple using Samsung for A9/X? Thats probably what Samsung is producing. Apple always likes to pay extra for the most advanced process.

That has only been true for this year's line of products. Traditionally they've lagged behind others.

 

[Sweepr]

I guess this new chip should power Galaxy Note 5's Exynos variant. It's probably too early for a Galaxy S6 appearance (~April 2015) but it should be ready by ~October 2015. Let see how far they can push those A57/A53 cores with FinFETs.

 

[Fjodor2001]

I guess this new chip should power Galaxy Note 5's Exynos variant. It's probably too early for a Galaxy S6 appearance (~April 2015) but it should be ready by ~October 2015. Let see how far they can push those A57/A53 cores with FinFETs.

In that case the competition will effectively only be about half a year behind Intel on 14 nm. Sure, Intel has some 14 nm chips on the market already, but only to claim 2014 availability. It will not be an actual high volume product until 2015Q1 and beyond.

Galaxy S6 will be a high volume product right from the start though. S6 volumes are HUGE.

If this continues, I wonder how long it'll be until the competition has caught up with and eventually surpassed Intel on process tech.

 

[kimmel]

Galaxy S6 will be a high volume product right from the start though. S6 volumes are HUGE.

As long as we are doing baseless speculation yet again. 99% of the volume will be the Qualcomm variant with the 1% made up by the yielding "14nm" Exynos. Which will have virtually the same density as previous nodes. Looks like Intel's process lead is intact.

 

[krumme]

As long as we are doing baseless speculation yet again. 99% of the volume will be the Qualcomm variant with the 1% made up by the yielding "14nm" Exynos. Which will have virtually the same density as previous nodes. Looks like Intel's process lead is intact.

With a density that is more like Intel 14nm. Wake up dude have you been living in a cave? Something new actually happened. We are even going to see 10nm euv MIDNODE logic h2 2016 from either ss or tsmc - as it seem probably ss.
And its good yes?
If you think its bad news better change your goalpost now.

 

[witeken]

In that case the competition will effectively only be about half a year behind Intel on 14 nm. Sure, Intel has some 14 nm chips on the market already, but only to claim 2014 availability. It will not be an actual high volume product until 2015Q1 and beyond.

Galaxy S6 will be a high volume product right from the start though. S6 volumes are HUGE.

If this continues, I wonder how long it'll be until the competition has caught up with and eventually surpassed Intel on process tech.

I'm not going to bother to elaborate, since you refuse to learn the correct facts. Intel's process lead is only expanding.

“It's a true 14nm technology. There's lots of 14nm technologies around, and they're not all created equal.” --William Holt

 

[krumme]

Can someone explain why ss is going so fast? It puzzles me. Whats their incentive and broader strategy. And how did they do it and for what cost? What/who is going to pay for it and why?

 

[Khato]

Interesting and well-chosen phrasing with "yielding at full specifications targets"...doesn't say much about how many of the die actually work. This is, in effect, saying that the ones that do work...work to spec

Can't help but chuckle at their wording there because it's entirely dependent upon what those 'targets' are. If their target for this quarter was to have non-functional silicon that fries itself when power is applied and that's what they got, then that statement would be true.

Note that there's a key omission in the quoted statement - functional.

 

[NTMBK]

Here's hoping the yields get good enough for us to see 14nm GPUs and APUs by the end of next year.

 

[krumme]

Here's hoping the yields get good enough for us to see 14nm GPUs and APUs by the end of next year.

Agree. We could really use some 14nm gpu...

But i guess apple, ss and the top phone brand is using all newest processes for their phones. Its profitable market and they are protecting extremely valuable brands here. My guess thats whats driving 14nm finfet and 10nm euv progression so fast now. Galaxy and iphone.

Keep an eye on sonys mobile sensor tech for the next 2 years. They are on full steam and we are probably less than a year from curved sensors in products that enables cheaper, smaller and better lenses . Again its mobile as the driving force.

 

[Abwx]

Can someone explain why ss is going so fast?

http://www.strategyand.pwc.com/glob...lobal-innovation-1000/top-20-rd-spenders-2013

 

[Arachnotronic]

Can't help but chuckle at their wording there because it's entirely dependent upon what those 'targets' are. If their target for this quarter was to have non-functional silicon that fries itself when power is applied and that's what they got, then that statement would be true.

Note that there's a key omission in the quoted statement - functional.

Khato,

I can help add more color as I finished listening to the presentation.

They expect to "ship" the wafers in the oven now in the first quarter of 2015. This doesn't mean that their yield rates are particularly high, but they do have functional silicon that should be ready to go.

IMO, Galaxy S6 Int'l edition will have this 14nm Exynos.

 

[liahos1]

i think there is a high certainty that apple is working with samsung/glofo in malta and austin at 14nm.

If you listen to their analyst day they speak about a key customer as a catalyst for samsung licensing their 14nm process to glofo.

the tool hookup project managers I've spoken with along with senior process engineers at samsung say that was it was apple's insistence that they tie up. although I have heard that glofo's hookup schedule has been slipping due to just general lack of contractor logistics/infrastructure in place in malta. it sounds like a clusterf*ck up there tbh.

Apparently apple didnt really want to work on 20nm at TSM but was sort of forced to after samsung balked at the idea of building 20nm capacity when mostly everyone ex apple wasnt really that interested in it. apparently Tim Cook REALLY wants to build in the USA. interestingly i've heard the 2h module in line 17 is being built for logic and not dram/nand as initially expected. which ties up to a comment samsung made on their q3 earnings about exiting the year with 30% of their wafer start capacity at 14nm. thats over 100k wafers a month i think. enough to fit lots of qcom, apple, nvidia etc.

The #1 paranoia that samsung has had about foundry is being tied to the hip to apple so I think there strategy has been to diversify their customer base by going after other leading edge customers TSM has traditionally served. This was further corroborated by the comments during the samsung lsi analyst day with respect to having a broad level of customer's in place at 14nm.

all just sort of based on the convos i've had over the past few months. most recently 2 weeks ago.

really interesting though that intel didnt spend any time talking about their foundry strategy during the analyst day. the tool hookup contractor i talked to that had worked in austin was actually working in israel standing up inte's 10nm facility in kiryat. he made some comments with regard to their fab structures not really being designed to do general purpose foundry. i need to further dig into this. He was speculating 10nm in Chandler could be stood up for Apple in the future but that was pure hypothesizing.

 

[Fjodor2001]

I'm not going to bother to elaborate, since you refuse to learn the correct facts. Intel's process lead is only expanding.

“It's a true 14nm technology. There's lots of 14nm technologies around, and they're not all created equal.” --William Holt

Sure, not all 14 nm are created equal. Nobody's arguing about that. One may be better than the competition in one aspect, and worse in another.

But do you really have the metrics data to show that Intel 14 nm is better than Samsung 14 nm in every aspect? If so, then please share it with us. I'm talking about transistor density, power consumption, price per transistor, max performance, yield, etc.

And are the differences really big enough to matter?

 

[NostaSeronx]

And are the differences really big enough to matter?

Intel's version is better than both Samsung's and TSMC's combined.

Unless, the semiconductor company is partnered with Intel's foundry. It's just a R&D dump that can literally kill companies.

https://www.youtube.com/watch?v=uTcR3eC-n5o#t=568

 

[III-V]

Sure, not all 14 nm are created equal. Nobody's arguing about that. One may be better than the competition in one aspect, and worse in another.

But do you really have the metrics data to show that Intel 14 nm is better than Samsung 14 nm in every aspect? If so, then please share it with us. I'm talking about transistor density, power consumption, price per transistor, max performance, yield, etc.

And are the differences really big enough to matter?

The only place where Intel has lagged their competition is with price (I imagine their labor expenses are quite a bit higher), and occasionally density. The data's been posted here numerous times, and you've certainly seen it. Performance in particular has been a big advantage of Intel's, with everyone lagging by at least 3 years.

Yield data is a well-kept secret, but I'd be surprised if anyone has average yields as high as Intel. They have gone great lengths to ensure they have very high yields -- apparently they use computational lithography out the wazoo.

 

[meloz]

This is, in effect, saying that the ones that do work...work to spec

The only thing that matters is when they can churn out enough volume to launch products, and how soon those products can get in the hands of consumers.

Do people think Samsung have a significant manufacturing lead over TSMC? Of these two whom will be able to produce on 14nm (large volume, not PR wafers) first?

 

[meloz]

In that case the competition will effectively only be about half a year behind Intel on 14 nm.

If this continues, I wonder how long it'll be until the competition has caught up with and eventually surpassed Intel on process tech.

What on Earth?!

 

[ShintaiDK]

I see Samsung learned the PR game. Shame we have to wait so long for real products

 

[AtenRa]

Intel will not have the advantage of FFs over Planar cmos anymore from 2015 onwards. That will have a tremendous affect in low power products from now on and that will make Intel to have to work harder.

 

[witeken]

IMO, Galaxy S6 Int'l edition will have this 14nm Exynos.

Very optimistic. Let's just pretend 20nm doesn't exist and no money has been spend developing the technology (). Let's also pretend is doesn't take more than half a year to get big volumes out of a new node into the market.