Sign in with
Sign up | Sign in

GlobalFoundries Unleashing 7-nm Process Tech in 2017

By - Source: Xbit Labs | B 20 comments

GlobalFoundries is expected to ramp up to 7-nm process technology by 2017.

During the Common Platform Technology Forum last week, GlobalFoundries and its partners disclosed some of their long-term plans. This included ramping up production using 14-nm-XM process tech by 2014, and possibly going into mass production with 7-nm fabrication technology by 2017.

Xbit labs reports that GlobalFoundries will initiate production using 20-nm process technology in 2013. The hybrid 14-nm-XM process, which combines 14-nm FinFET transistors with 20-nm back-end-of-line (BEOL) interconnected flow, will be ramped up in 2014. Then in 2015 the company plans to intro 7-nm technology and the 10-nm-XM hybrid process which will utilize 10-nm FinFET with 14-nm BEOL.

According to a chart provided by GlobalFoundries, 28-nm LPH parts for the "wireless/mobile computing/ digital consumer" sector have already entered production, and will be followed by 20-nm LPM parts for "wired applications/networking" sometime this year. Later on the company will begin manufacturing 14-nm XM parts for both categories, followed by 10-nm XM parts in early 2015.

The report said that members of the Common Platform alliance plan to speed up development and deployment of new process technologies by launching the tech before it’s approved by IBM. This is expected to speed up time-to-market and maturity of the latest manufacturing tech, but at a higher cost for the foundries.

GlobalFoundries, which seems unfazed by the current PC decline, will reportedly be able to process around 190,000 300-mm wafers per month by late 2013 including 20-nm, 28-nm, 32-nm, 40-nm and other process technologies. Meanwhile, rival Taiwan Semiconductor Manufacturing Company (TSMC) already processes over a million 300-mm wafers per month, and is expect to boost its output by the end of 2013.

 

Contact Us for News Tips, Corrections and Feedback

Discuss
Display all 20 comments.
This thread is closed for comments
Top Comments
  • 19 Hide
    gamerk316 , February 13, 2013 1:48 PM
    Scheduled for 2017 means release at 2022...
  • 10 Hide
    m32 , February 13, 2013 3:52 PM
    AMD thought they were going to have 22nm tech 2 years ago. #DONTBELIEVE
Other Comments
  • 2 Hide
    icemunk , February 13, 2013 1:35 PM
    Sweet! It will be interesting to see the technologies we will be able to have with the 7nm process
  • 19 Hide
    gamerk316 , February 13, 2013 1:48 PM
    Scheduled for 2017 means release at 2022...
  • 7 Hide
    slabbo , February 13, 2013 1:53 PM
    Global Foundries always paper launches tech, but the real launch happens years after TSMC does it first.
  • -4 Hide
    sivaseemakurthi , February 13, 2013 2:17 PM
    Like they 'unleashed' 32nm and 45nm before that!
  • 2 Hide
    Onus , February 13, 2013 2:33 PM
    Regardless of the year this comes out, it certainly points to substantial power saving and reduced heat. I just hope they don't run into problems with longevity.
  • -2 Hide
    realibrad , February 13, 2013 3:07 PM
    keep in mind that as we get smaller, OC ability falls. 7 nm does not give a lot of head room for a large range of power.

    Hopefully we see a large enough performance jump to cover the lack of OC headroom.
  • 1 Hide
    maxinexus , February 13, 2013 3:43 PM
    GF pfff even if they get to 7nm their yields will be so small that nobody would bother to go through the hussle.
  • 10 Hide
    m32 , February 13, 2013 3:52 PM
    AMD thought they were going to have 22nm tech 2 years ago. #DONTBELIEVE
  • -5 Hide
    de5_Roy , February 13, 2013 4:44 PM
    is this even for steamroller future apus? looks like it'll be for low power mobile socs, not for higher performance apus and cpus. the process for the apus doesn't seem to be ready yet...
  • -4 Hide
    drbaltazar , February 13, 2013 4:49 PM
    Intel still have yet to find a cooling solution (a la 32 nm but for 22nm )and these guy are ready for 20 nm ?lol!I bet if Intel find a cooling solution they LL be ready before lol.reaching smaller is relatively easy!cooling smaller is the main problem now
  • 2 Hide
    dalethepcman , February 13, 2013 4:59 PM
    I thought 10nm was as small a fab process that could go and be cost effective. At sizes smaller than 10nm the distance between channels/transistors required to insulate from electron overflow would be cost prohibitive (more than half the wafer would be empty space.)

    Unless they found a new way to make electricity travel, I don't see anyone making a CPU out of these revolutionary new transistors until much later than 2016.
  • 2 Hide
    anham , February 13, 2013 6:09 PM
    dalethepcmanUnless they found a new way to make electricity travel, I don't see anyone making a CPU out of these revolutionary new transistors until much later than 2016.

    In a since they did. FinFET transistors leak less power, look into it.
  • 2 Hide
    freggo , February 13, 2013 7:22 PM
    So by about 2020 we will have to get used to articles about the new processors using 800pm technology...working it's way down from there ?

    Funny how the US insists on inches, acres and miles, but uses the decimal systems in computer chips [nm], bullet sizes [9mm], drug quantities [kilo] and sodas [2liter bottles or 1/3rd liter cans].

  • 1 Hide
    Fokissed , February 13, 2013 7:47 PM
    freggoSo by about 2020 we will have to get used to articles about the new processors using 800pm technology...working it's way down from there ?Funny how the US insists on inches, acres and miles, but uses the decimal systems in computer chips [nm], bullet sizes [9mm], drug quantities [kilo] and sodas [2liter bottles or 1/3rd liter cans].

    Cans are 12oz in the U.S.
  • 0 Hide
    wdmfiber , February 14, 2013 12:10 AM
    drbaltazarIntel still have yet to find a cooling solution (a la 32 nm but for 22nm )and these guy are ready for 20 nm ?lol!I bet if Intel find a cooling solution they LL be ready before lol.reaching smaller is relatively easy!cooling smaller is the main problem now

    Ivy Bridge runs hotter than Sandy because the Intel used a cheap thermal paste under the IHS, rather than soldering. Any blame directected at the smaller fab process was/is false. Plus the problem is user(enthusiast) fixable:
    http://www.youtube.com/watch?v=XXs0I5kuoX4
    And mobile CPU's are fine, smaller packaging - no IHS.
  • 0 Hide
    mamailo , February 14, 2013 12:40 AM
    de5_Royis this even for steamroller future apus? looks like it'll be for low power mobile socs, not for higher performance apus and cpus. the process for the apus doesn't seem to be ready yet...

    In fact they do.Is not a new transitor type (FinFET) wich would require many adaptations from the curent design, it just need the new electrical libraries and then send the CAD servers to a geometry reduction.

    freggoSo by about 2020 we will have to get used to articles about the new processors using 800pm technology...


    Not realy. At 5nm the Elctron-Hole pairs flow has been detected in silicon but it seen to be a material end of the road.At 1nm the Tunnel Effect and the Transistor Effect are equaly important.The potencial barrier (base of the transistor) not longer hold up.

    BTW: due the heisenberg principle, the angstrom is commonly used at that scales.The pm is too small to be measured and be of utility.
  • 0 Hide
    wdmfiber , February 14, 2013 1:56 AM
    Agreed. 800nm likely isn't even possible(quantum computer future anyway) but if it were it's way beyond 2020 technology. 5nm by 2020 and serveral more reductions(halving) later... 1nm ~2028
    source:
    http://en.wikipedia.org/wiki/5_nanometer
    interesting chart on wikipedia, starting at 10um back in 1971
    @realibrad, your OC theory would just be magic with a thick old fab process, who needs billions of tranistors? What are they thinking? ...lol
  • 0 Hide
    ojas , February 14, 2013 3:48 AM
    Well...Intel goes 14nm in 2014, then 10nm in 2016 and 7nm in 2018....hahahaha no GF isn't hitting 7nm in 2017...
  • 1 Hide
    agnickolov , February 14, 2013 4:12 AM
    mamailoBTW: due the heisenberg principle, the armstrong is commonly used at that scales.The pm is too small to be measured and be of utility.

    Perhaps you meant Angstrom?
    https://en.wikipedia.org/wiki/Angstrom
  • 0 Hide
    puddleglum , February 14, 2013 12:58 PM
    I'm tired of bogus paper releases. I'll believe it when I see it.