[KG] AMD: We will ship 20nm products in 2015

[Fastx]

I thought they might skip 20nm and go right to 16nm FinFET, but doesn't look like it now.

July 22nd, 2014 at 12:00 am


Although there are rumours that Advanced Micro Devices and Nvidia Corp. are not really happy with 20nm process technology of Taiwan Semiconductor Manufacturing Co., they will inevitably have to use it. A vice president of AMD said last week that the company’s 20nm chips would be out next year.

“20nm is an important node for us. We will be shipping products in 20nm next year and as we move forward […],” said Lisa Su, senior vice president and chief operating officer of AMD. “If you look at our business, it is quite a bit more balanced between the semi-custom, embedded, […] professional graphics […] as well as the more traditional sort of client and graphics pieces of our business. [20nm] technology plays in all of those businesses.”

This is not the first time when AMD outlines its plans to use 20nm process technology for its chips. However, this is the first time when AMD reveals that it will use the manufacturing process for graphics processing units, semi-custom accelerated processing units and embedded products. For some reason Ms. Su did not say whether AMD plans to make central processing units using 20nm process technology.

amd apu beema mullins puma jaguar x86 fusion AMD: We will ship 20nm products in 2015

20nm process technology will help AMD to boost functionality and performance of its products. Besides, it will allow AMD to further cut down power consumption of its GPUs and semi-custom Fusion system-on-chips.

http://www.kitguru.net/components/graphic-cards/anton-shilov/amd-we-will-ship-20nm-products-in-2015/

 

[Bateluer]

Wasn't 16nm FinFet the marketing term for 20nm?

 

[Fastx]

Wasn't 16nm FinFet the marketing term for 20nm?

Check this on 20nm - 16nm FinFET
http://www.extremetech.com/computin...m-finfet-tapeout-of-big-little-cortex-a57-soc

 

[jpiniero]

Wasn't 16nm FinFet the marketing term for 20nm?

TSMC is calling their 20 nm with finfet 16 nm.

 

[NostaSeronx]

This will help just spread the word. Intel did this at the 28-nm... I mean the 22-nm node. So, their 14-nm node by definition is the same density as TSMC and Samsung.

=====
GlobalFoundries has cancelled;
14-nm UTBB FDSOI

And, there is no life signs of;
14-nm SOI FinFETs

 

[Fastx]

Quote below from the above link I posted.

The dirty little secret in all of this, of course, is that 16nm FinFET is actually 20nm R&D + FinFET technology bolted on. Not only is this a stark illustration in how arbitrary node names have become across the entire industry, including Intel, it’s also proof of just how much the semiconductor industry as a whole is struggling to deliver meaningful performance gains.

http://www.extremetech.com/computin...m-finfet-tapeout-of-big-little-cortex-a57-soc

 

[Ajay]

Well, this was clear from the recent con-call.

Also:

KitGuru Says: It is noteworthy that neither AMD nor Nvidia try to be the first to adopt 20nm process technology. Apparently, TSMC’s prices for 20nm production are rather high, hence, fabless chip designers want to postpone adoption to ensure maximum yields and minimize the costs of their products.

And the yields were terrible, which made it even more expensive. I don't know where TSMC is at w/20nm, but it seems like only Apple can afford it right now.

 

[3DVagabond]

Well, this was clear from the recent con-call.

Also:

And the yields were terrible, which made it even more expensive. I don't know where TSMC is at w/20nm, but it seems like only Apple can afford it right now.

Is there a 20nm HP process available yet from TSMC? I don't think there is which would explain why we haven't seen GPU's, as well as the cost/yields too.

 

[NostaSeronx]

TSMC is at volume and the yields are good. The process is cheaper than it was in risk and early production. Most fablesss companies already have 20-nm designs since 2012.

20-nm FinFETs require a completely new design while 20-nm planar requires only restructuring. So, it would be best for brand new architectures to be on FinFETs. While having shrinks of existing architectures on planar.

Just because Intel can do FinFETs doesn't mean everyone can do FinFETs anytime soon. This is why UTBB FDSOI and DDC Bulk is such a big deal.