TSMC & Samsung Developing 3rd Gen 16 and 14-Nano FinFET Processes

[Sweepr]

During IR to present yearly result for 2015, TSMC made an announcement that it is planning to enter mass-production system of chips produced by 16-nano FinFET Compact (FFC) process sometime during 1st quarter of this year. TSMC had finished developing 16-nano FFC process in 4th quarter of last year. Co-CEO C.C Wei of TSMC explained that 16-nano FFC process focuses on reducing production cost more than before and implementing low electricity.

...By adding FFC process, TSMC now has total of 3 16-nano processes.

“We are going to develop a derivative process that will follow Gen.2 14-nano process and lead mobile chip and foundry markets.” said Vice-President Bae Young-chang of Samsung Electronics’ System LSI Business Department’s Strategy Marketing Team.

Unlike 32-nano and 20-nano processes, it seems that 14 and 16-nano processes will become ‘longevity’ processes because it has become difficult to reduce line width as processes are entering 10-nanos. If it is not easy to secure yields, then it is going to lead to increase of cost for production of chips. This is why 28-nano process is still the main process after long years.

“It is predicted that Samsung Electronics and TSMC will develop Gen.3 14 and 16-nano derivative processes and focus on securing customers before they finish developing high-tech 10-nano process at the end of this year.” said a representative of this industry. “Even if 10-nano process is commercialized, there still will be many fabless businesses that will use 14 and 16-nano processes that cost relatively cheaper.”

http://english.etnews.com/20160121200001

 

[ShintaiDK]

16FFC seems to be the A10 variant.

 

[Andrei.]

16FFC seems to be the A10 variant.

Just no. 16FFC is a IoT/ULP node that has absolutely no usage in anything requiring performance.

 

[Arachnotronic]

Just no. 16FFC is a IoT/ULP node that has absolutely no usage in anything requiring performance.

But, according to this Semiconductor Industry Professional, 16FFC is for the Apple A10, absolutely!

https://www.semiwiki.com/forum/content/5108-who-will-get-next-bite-apple-chip-business.html

:biggrin:

But yes, 16FFC is actually for IoT and low-cost mobile devices. Anybody building a high performance mobile SoC at TSMC will be using 16FF+.

 

[witeken]

Yeah, we should really "worry" about TSMC: http://www.fool.com/investing/gener...the-troubling-story-behind-intels-chip-m.aspx.

And if I remember correctly, since Samsung did some learning from TSMC (to say it euphemistically), from them as well.

Intel, as IDM that has dominated the manufacturing space for decades and has pioneered Tick Tock method, will be left behind.

 

[Arachnotronic]

Intel, as IDM that has dominated the manufacturing space for decades and has pioneered Tick Tock method, will be left behind.

Intel didn't really even have a process lead until, like, 45nm (HKMG)

I think the real issue that some people have with Intel and their statements about mfg prowess is that they way, way oversell their "advantage" and even try to lead people into thinking that their lead is growing which IMO it's not.

I still don't know if this is a case of arrogance or just not having a clue what the competition is doing. I think it's the former, though.

 

[el etro]

16FFC From TSMC(seems to come more processes from them but i don't have the names); and 14LPC(ultra-low power)/14LPA(a bit more power oriented than LPP, around LPE in the performance/leakage metrics)/14LPH(true successor to 14LPP, performance optimized process that targets tablets and notebooks) from Samsung.

And Seronx claims IBM 14SOI Finfet will be ready to mass production at GlobalFoundries in 2016.

 

[Ajay]

And Seronx claims IBM 14SOI Finfet will be ready to mass production at GlobalFoundries in 2016.

Well, that might be true (for use in Power 9). I wish it was going to be for Zen, then I might actually believe that it will perform competitively.

 

[dark zero]

Wondering if Zen+ is 14LPH

 

[raghu78]

Wondering if Zen+ is 14LPH

The next few years are good for fabless companies as TSMC and IBM/Samsung/GF close the process gap at 10nm and 7nm against Intel. TSMC looks the best to deliver on completely erasing the process node gap against Intel. TSMC has stated Q4 2016/Q1 2017 volume production for 10nm and 7nm a year later, in Q1 2018.

http://www.fool.com/investing/gener...the-troubling-story-behind-intels-chip-m.aspx

TSMC 7nm products will be competing with Intel 10nm products in 2018, 2019. AMD has a good opportunity to compete with Intel on competitive process nodes against Intel, something which they could not do in the past. AMD needs to grab the opportunity and deliver excellent products with their foundry partners.

 

[el etro]

Well, that might be true (for use in Power 9). I wish it was going to be for Zen, then I might actually believe that it will perform competitively.

But IBM process is sized as Samsung process is, no?

Wondering if Zen+ is 14LPH

Is 14LPP.

 

[Ajay]

But IBM process is sized as Samsung process is, no?

I don't know. I would think the design rules would be very different due to differing electrical characteristics even if physical dimension where the same.