bit-tech.net

TSMC 16nm to hit volume production this year

TSMC 16nm to hit volume production this year

TSMC

Taiwan Semiconductor Manufacturing Company (TSMC) has said that its 16nm FinFET manufacturing process will be at volume levels of production towards the end of this year.

This puts the company ahead of its peers, with much of the industry still ramping up on 20nm production. TSMC itself is still at relatively low volumes for 20nm FinFET, with volume production only starting this month and total production for 2014 expected to be just 10% of the company's output, but be leveraging knowledge gained through creating that process TSMC thinks it can transition to 16nm more rapidly.

"Our 16FinFET yield improvement has been ahead of our plan. This is because we have been leveraging the yield, learning from 20SoC. Currently, the 16FinFET SRAM yield is already close to that of the 20SoC process." said TSMC CEO Mark Liu.

The announcement was made at the company's investor conference where it was discussing its latest financial results. Partially thanks to this speedy transition to 16nm TSMC expects full year sales and profits to improve by 10%.

As well as financial positives for TSMC, the move to 16nm FinFET will have performance advantages too with the company estimating improvements of around 15%. "It will be the highest performance technology amongst all available 16 and 14nm technologies in 2014. This progress is well ahead of Samsung," Liu boasted.

Liu also revealed that 20 products based on the 16nm process were scheduled for production in 2014.

TSMC is the largest contract semiconductor manufacturer in the world with it producing the likes of Nvidia's graphics card chips as well as working with AMD, Apple, Qualcomm, VIA and many others.

6 Comments

Discuss in the forums Reply
Alecto 17th January 2014, 22:02 Quote
"It will be the highest performance technology amongst all available 16 and 14nm technologies in 2014. This progress is well ahead of Samsung," Liu boasted.

How does that relate to Intel's 14 nm process (to be used for Broadwell)?
rollo 17th January 2014, 22:29 Quote
No clue, Rumours around though that Apple paid mega bucks to buy out alot of the early production.

Maxwell will likely be the first gpu released under the process.
fluxtatic 18th January 2014, 05:35 Quote
Eh...I'm not buying it. They're just now ramping up production of 20nm, and they expect us to believe they're jumping right to 16nm for volume production this year?
Bindibadgi 18th January 2014, 09:05 Quote
I've heard and read their process' vary between "14-20nm". I'm not sure the specifics as TSMC itself is very tight with real info.
Eriol 20th January 2014, 21:22 Quote
Quote:
Originally Posted by Alecto
"It will be the highest performance technology amongst all available 16 and 14nm technologies in 2014. This progress is well ahead of Samsung," Liu boasted.

How does that relate to Intel's 14 nm process (to be used for Broadwell)?

AFAIK 16 nm FinFET isn't actually 16 nm, but just 20 nm with FinFET. It's actually on a similar level with Intel's 22 nm Tri-Gate. Intel is way ahead in the process tech game.

If you're a pessimist it seems like 16 nm FinFet is the first process you'll get any decent benefit over the current 28 nm as far as GPU's are concerned. And if the first production is SoC only and we'll have to wait until late 2015 before getting it in GPUs we're in for a long wait before we see anything better than the 780 Ti.
Bindibadgi 21st January 2014, 02:31 Quote
Quote:
Originally Posted by Eriol
AFAIK 16 nm FinFET isn't actually 16 nm, but just 20 nm with FinFET. It's actually on a similar level with Intel's 22 nm Tri-Gate. Intel is way ahead in the process tech game.

If you're a pessimist it seems like 16 nm FinFet is the first process you'll get any decent benefit over the current 28 nm as far as GPU's are concerned. And if the first production is SoC only and we'll have to wait until late 2015 before getting it in GPUs we're in for a long wait before we see anything better than the 780 Ti.

That's the one!

+rep you.

And yes - there's a process difference between making tiny ARM chips and ~10Bn transistor GPUs.

EDIT: Rep button disappeared??
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums