News TSMC gets green light for China fab plans

Discussion in 'Article Discussion' started by Gareth Halfacree, 3 Feb 2016.

  1. Gareth Halfacree

    Gareth Halfacree WIIGII! Staff Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    9,419
    Likes Received:
    307
  2. zimano

    zimano New Member

    Joined:
    1 May 2014
    Posts:
    33
    Likes Received:
    0
    Assuming that TSMC really are ahead of Intel process node development wise, what does this really mean? Intel will lose the advantage of getting more chips per wafer than anyone else but is there more to it than this? For as long as I can remember Intel have had this advantage.
     
  3. Jimbob

    Jimbob New Member

    Joined:
    2 Jul 2009
    Posts:
    189
    Likes Received:
    1
    If they plan to move to 7nm by 2018 then what is the point in opening a new factory producing 16nm in 2018?
     
  4. Gareth Halfacree

    Gareth Halfacree WIIGII! Staff Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    9,419
    Likes Received:
    307
    Because not everyone wants a cutting-edge process. Look at some of the biggest chips out there: GPUs are at 28nm, FPGAs are at 28nm, 32-bit ARM chips (the majority of 'em, in other words) are at 28nm, Atmel's *incredibly* popular range of ATmega microcontrollers are built on a 350nm process...

    There's no way anybody's bulk output will be 7nm by 2018.
     
  5. Bindibadgi

    Bindibadgi Tired. Forever tired.

    Joined:
    12 Mar 2001
    Posts:
    36,319
    Likes Received:
    419
    As a TWese company they cannot legally have a leading edge node manufactured in China. By 2018 they will be on 10-7nm, so 16 will not be leading. It will make them more cost competitive on 16nm.
     
  6. Elton

    Elton Officially a Whisky Nerd

    Joined:
    23 Jan 2009
    Posts:
    8,574
    Likes Received:
    189
    Costs. It costs quite a lot more to develop on a node that is smaller, you need better optics, and above all you need to know how the yields are. Using an older node especially now is a good way to retain good yields and potentially cut down on expensive lithography and optics required with smaller nodes.

    Not to mention that early smaller nodes will undoubtedly have yield issues so stop gap nodes (like 16nm) are pretty great when you need to serve more than just the big CPU/GPU guys.
     

Share This Page