Samsung’s 14nm in mass production?

[user_2013101]

At the company's Jan 29 earnings call, Page 5 of the transcript, Samsung's VP Ji Ho Pak said:

"For the first quarter, we will start 14-nanometer mass production in earnest and begin supply for customer's new product."

EVENT DATE/TIME: JANUARY 29, 2015
THOMSON REUTERS STREETEVENTS EDITED TRANSCRIPT: Q4 2014 Samsung Electronics Co Ltd Earnings Call
http://aod.teletogether.com/sec/20150129/2014Q4_script.pdf

On Feb 16, Samsung announced the mass production of its 14nm process, supposedly officially.

Samsung’s leading-edge14nm FinFET process will be adopted by its Exynos 7 Octa, then expanded to other products throughout the year.

Samsung Announces Mass Production of Industry’s First 14nm FinFET Mobile Application Processor
Samsung Announces Mass Production of Industry

Prior reports merely hot air

It’s now official that the mass production of Samsung’s 14nm started in mid February 2015, for its own Exynos 7 SoC.

In the prior 3-4 months before mid Feb, numerous media stories pumped the idea that Samsung had commenced 14nm volume production including building Apple’s A9 SoC. Now these previous reports look to be false information, probably from a deliberate PR campaign.

What is Samsung’s “mass production”?

Does it mean the 14nm production has reached 50K wafers per month in Feb?

Or, Samsung just started the ramp of 14nm volume?

Or, somewhere in between?

Or, another fabrication?

We will get a better idea after the Galaxy S6 starts shipment. We will find out how many, if any, of the S6 are equipped with 14nm Exynos. Note that Samsung does have a 20nm Exynos 7420 in production. S6 may use the 20nm part.

In addition, Samsung’s 14nm chips will be scrutinized to see if any evidence that Samsung had violated TSMC’s patents and trade secrets, as detailed in the following article and its comments:

02-10-2015
TSMC vs Samsung!

 

[Daniel Nenni]

As you are probably aware Samsung has a very optimistic approach to press releases. I do know that Samsung 14nm is in production with "reasonable" yield which to me means above 50% at this stage in the ramp. The question I have is how will Samsung supply both their internal Exynos SoC demand AND Apple's A9 demand? Remember, Samsung and Apple hold the majority market share of smartphones:

https://www.semiwiki.com/forum/content/4311-mobile-2014-future.html

My guess is that they will supplement internal demand with QCOM and MediaTek SoCs. Still, it is hard to believe that Samsung 14nm will dominate the Smart phone market in 2016. Quite an impressive feat for sure.

 

[astilo]

In the prior 3-4 months before mid Feb, numerous media stories pumped the idea that Samsung had commenced 14nm volume production including building Apple’s A9 SoC. Now these previous reports look to be false information, probably from a deliberate PR campaign.

Why do you think they are false?

We will find out how many, if any, of the S6 are equipped with 14nm Exynos. Note that Samsung does have a 20nm Exynos 7420 in production. S6 may use the 20nm part.

First of all, Qualcomm already admitted to have lost one of the biggest 810 customers (guess who?). So, the amount of chips we are talking about, is definitely huge.
Exynos 7420 is only at 14nm ("Samsung’s leading-edge14nm FinFET process will be adopted by its Exynos 7 Octa"), the 20nm is the 7410 formerly known as 5433.

In addition, Samsung’s 14nm chips will be scrutinized to see if any evidence that Samsung had violated TSMC’s patents and trade secrets, as detailed in the following article and its comments

Patents are patents, not sure what you can really claim instead about trade secrets (very hard to prove, since in principle, they are just secrets, so never disclosed).

 

[Daniel Nenni]

Patents are patents, not sure what you can really claim instead about trade secrets (very hard to prove, since in principle, they are just secrets, so never disclosed).

It will all depend on the email as with the TSMC vs SMIC case. Email is the smoking gun. And calling TSMC "the T company" on email is not going to fool a jury.

 

[IanD]

Daniel, I thought you said on the "TSMC vs. Samsung" thread that the dispute was really about one guy's golden handcuff agreement, not patent infringement as such?

If so, the hints in this thread about patent infringement by Samsung look rather like spreading FUD by people who dislike Samsung and like TSMC...

 

[Andrei]

If so, the hints in this thread about patent infringement by Samsung look rather like spreading FUD by people who dislike Samsung and like TSMC...

You characterized user_2013101 activity on this site pretty well.

Also "mass production" announcements like these by Samsung means they have reached or are near reaching their full volume production. The ramp has been going on since October.

 

[Daniel Nenni]

Daniel, I thought you said on the "TSMC vs. Samsung" thread that the dispute was really about one guy's golden handcuff agreement, not patent infringement as such?

If so, the hints in this thread about patent infringement by Samsung look rather like spreading FUD by people who dislike Samsung and like TSMC...

It is thus far but I'm guessing TSMC has enough to discover email/texting records and that may lead to more legal action, or not. The email trail is what really sunk SMIC and that is very common. One should always assume that anything put in email or on social media can be used against you because it probably will.

 

[user_2013101]

[h=5]Samsung’s prior claims debunked by itself [/h]

Samsung Announces Mass Production of Industry

In the above Feb 16 release, Samsung stated its mass production just started in February. And, no mention of Apple, not even a hint of an “external big customer.” This release debunks its prior claims – some samples of its claims below:

Oct 1, 2014:
Samsung’s president, Kim Ki-nam, “told reporters at headquarters in Seoul that once the company begins to supply Apple with chips using its latest technology, profits ‘will improve positively,’

Samsung is expected to start producing application processors (APs) for clients such as Apple, Qualcomm, and AMD, using its 14-nanometre process around the end of the year.

October 1, 2014
Samsung expects sales boost from Apple chip deal
Samsung expects sales boost from Apple chip deal | ZDNet

Dec 12, 2014
Samsung Electronics began production of Apple A9 in 14nm FinFET
???? IT??? ??! ????

[h=5]Conflicting data on Exynos 7420 [/h]

Before January, Exynos 7420 was reported as 20nm chip. The following story is an example.

"The other variant of the device will come with the Exynos 7420 Octa which is manufactured on 20nm HKG and has 64 bit architecture as well."

Nov 2014
Details About The Samsung Galaxy S6 Leak - Device To Feature 64 Bit Snapdragon 810

Only in the past few months, Exynos 7420 has been reported as a 14nm chip. However, there are doubts, because the benchmark results of this supposedly 14nm 7420 appear to fall behind QCOM’s SD 810 from TSMC’s 20nm. See stories below:

February 16 2015
Qualcomm Snapdragon 810 vs. Samsung Exynos 7420: AnTuTu Benchmark Scores Show Snapdragon Chip Relatively Better Than Exynos

13 Feb 2015
Preliminary Snapdragon 810 vs Exynos 7420 benchmarks show Qualcomm made the faster SoC - with caveats
http://www.phonearena.com/news/Prel...mm-made-the-faster-SoC---with-caveats_id65973

Should a 14nm SoC, with the same 8-core ARM A53/A57, underperform a 20nm one? Or, is Samsung’s 14nm such a lousy imitation of TSMC’s 16nm? Or, Exynos 7420 is really a 20nm part?

[h=5]An attitude issue? [/h]

Here I provided a lot of references. But, some posters offer no data references; instead, question other posters’ motivations. These posters seem to have an attitude issue: they believe they own the truth, delivery little substances, and have the privileges to hurl personal attacks.

These proud posters might have underestimated the readers’ intelligence.

 

[Andrei]

If you actually read your references you will see that Samsung did not make a single false claim in any of those articles. The purported debunking is just fictional fantasy.

The very same link calling the 7420 a 20nm piece uses the same false reasoning by others threads here by confusing the official "Exynos 7" product page describing the 5433 model as a 20nm piece. Yet we have Samsung claiming the "Exynos 7" is now a 14nm piece? Might we be referring to two different chips and this is just shoddy reporting?

That very same source you reference is calling the Snapdragon 810 a 2.8GHz piece. I wonder what exactly what the reliability of this source is mhm? Conflicting data indeed.

Fact is that we have had the 7420 being demonstrated since TechCon beginning October: https://www.youtube.com/watch?v=XrpVceyWO-U

Also if one wouldn't be as blind-sighted by a single data-point, one would see the truly 20nm 5433 Exynos is already outperforming the Snapdragon 810 on the CPU side: AnandTech | Understanding Qualcomm's Snapdragon 810: Performance Preview

Further inspection of those Antutu results also point out the outlying score putting the S810 above the 7420 seems to be the CPU tests. That's interesting. I wonder if that has anything to do with the benchmark application having newer version and 64-bit capability on one platform and a history of non-comparable scores between revisions. Hrmm.

But certainly, that one data point is enough to contradict a history matching the professional publication's and public user results, indeed depicting the 7420 much outperforming the S810:
universal7420 - Geekbench Search - Geekbench Browser
http://browser.primatelabs.com/geekbench3/search?q=avl7420

http://browser.primatelabs.com/geekbench3/search?q=msm8994

The reader's intelligence is not under question but some's motivations and misleading through cherry-picking of data of given posters might very well be.

One could by the same methodology also source that Intel is building Apple's 14nm chip, here are my sources:
http://www.fool.com/investing/general/2014/07/11/will-intel-build-apples-a9-chip.aspx
http://www.digitimes.com/news/a20140711PD205.html

Of course the reader's intelligence will quickly note such absurd claims as false. Hopefully.

 

[user_2013101]

...
Also if one wouldn't be as blind-sighted by a single data-point, one would see the truly 20nm 5433 Exynos is already outperforming the Snapdragon 810 on the CPU side ...

Also "mass production" announcements like these by Samsung means they have reached or are near reaching their full volume production. The ramp has been going on since October.

Congratulation! A poster offers some references, finally.

There are many benchmarks with numerous sub-categories. It’s not difficult to cherry pick some metrics to favor a particular chip. As I quoted, at least some writers cited benchmarks to show Exynos 7420 does not have leads over SD 810.

Nevertheless, I didn’t rule out a 14nm Exynos, nor its possible advantages. I said “conflicting data”, “having doubts,” and, as stated at the beginning of this thread: I like to wait for the real chips inside the Galaxy S6 for comparison and teardown.

This poster, however, has deep convictions, not only in the superiority of Exynos, but also in the “full volume production” of Samsung’s 14nm node that is capable of supplying the needs of both Galaxy S6 and iPhone 7 – pretty much the entire premium smartphone market. That is about 180 millions parts over two quarters. Intel’s 14nm, for comparison, has not yet been able to deliver such huge volume since 3Q14.

I respect this poster’s convictions, or faith. Readers will reach their own conclusions on this subject.

P.S.
Readers may also form an opinion if posters with strong convictions are entitled to the privilege of personal attacks on others.

 

[Daniel Nenni]

Interesting, I just got this email:

http://www.samsung.com/us/foundry/enews/SSI_eNewsletter_0215.pdf



So yes, it seems that Samsung is in production at 14nm.........

 

[Andrei]

It’s not difficult to cherry pick some metrics to favor a particular chip. As I quoted, at least some writers cited benchmarks to show Exynos 7420 does not have leads over SD 810.

Writers reporting on the same reference and dubious single data-point while ignoring the greater picture as reported by many more test-cases. Articles portraying the opposing scenario are only a Google search away. As you say, it is not difficult to cherry-pick.

Nevertheless, I didn’t rule out a 14nm Exynos, nor its possible advantages. I said “conflicting data”, “having doubts,” and, as stated at the beginning of this thread: I like to wait for the real chips inside the Galaxy S6 for comparison and teardown.

Although you say you do not rule out these points, you chose to only portray one side of the discussion which fits the TSMC agenda. Quotes such as these are textbook FUD spreading given the context and proof posted above:

FUD is generally a strategic attempt to influence perception by disseminating negative and dubious or false information:

Should a 14nm SoC, with the same 8-core ARM A53/A57, underperform a 20nm one? Or, is Samsung’s 14nm such a lousy imitation of TSMC’s 16nm? Or, Exynos 7420 is really a 20nm part?

Given your admittance of being a shareholder and following TSMC for 10 years - it is pretty clear where the bias lies. One only has to research post history. It is also very clear that the community is not blind to these obvious biased stances and there has been repeated friction.

In any case- everything will be official in 6 days and we'll see who will be correct, and who will continue to have a clean track-record.

 

[user_2013101]

Uh, oh. Someone is getting emotional, if not hysterical.

Ok, ok, Samsung’s 14nm is superior beyond belief. Samsung is taking over the world. Ok?

It’s just an investment, not a religion. …

 

[user_2013101]

Today, on the following news, ASML up 5%; and, TSM made another all time high.

Feb 24, 2015
ASML: Press - Press Releases - ASML Announces New High Mark for EUV Productivity; TSMC Images More Than 1000 Wafers in a Single Day

P.S.
TSMC just trying desperately and hopelessly to catch up with Samsung. Ok?

 

[Andrei]

Amazing sudden change of topics. To expected when one has run out of arguments or facts. The Illeism is dwindling though, please work on keeping that up in your writings. We don't want to break persona, do we?

 

[IanD]

Today, on the following news, ASML up 5%; and, TSM made another all time high.

Feb 24, 2015
ASML: Press - Press Releases - ASML Announces New High Mark for EUV Productivity; TSMC Images More Than 1000 Wafers in a Single Day

P.S.
TSMC just trying desperately and hopelessly to catch up with Samsung. Ok?

Looking from outside -- with no TSMC or Samsung axe to grind -- every new post like this just further destroys any credibility you might once have had.

 

[user_2013101]

Looking from outside -- with no TSMC or Samsung axe to grind -- every new post like this just further destroys any credibility you might once have had.

Someone does not understand the sarcasm in my post, nor providing any substance.

Nevertheless, this proud poster seems to believe he own the privilege to lecture and attack other posters.

 

[IanD]

Someone does not understand the sarcasm in my post, nor providing any substance.

Nevertheless, this proud poster seems to believe he own the privilege to lecture and attack other posters.

Someone is English and is perfectly capable of distinguishing sarcasm from FUD. On the second point, maybe you should take the log out of your own eye... ;-)

 

[user_2013101]

It’s pretty obvious that Samsung’s latest release, if it is to be believed, debunked all the prior reports that Samsung had won and started A9 production.

It is also impossible for Samsung’s 14nm process to supply both Galaxy S6 and the next iPhone at the same time. If Samsung is indeed to build 14nm Exynos for S6, then it does not have the capacity for A9s.

Unfortunately, in order to obscure such simple observation, some posters resort to personal attacks. These posters’ emotional responses are another confirmation of the observation. When facts contradict their faiths, they turn defensive and emotional.

It’s just an investment, not a religion. …Take easy!

P.S.
Today, TSM made another all time high. It seems to be daily occurrence now. TSMC is supposed to have lost to Samsung all or most of the orders from Apple, QCOM, AMD, NVDA, etc. Either the market had done a monumental miscalculation on TSM; or, the media reports are a result of Samsung’s relentless PR misinformation.

 

[IanD]

Orwell in the 21st century : "TSMC good, Samsung bad"