Shop OBEX P1 Docs P2 Docs Learn Events
TSMC begins construction on $9B 300-mm fab — Parallax Forums

TSMC begins construction on $9B 300-mm fab

LeonLeon Posts: 7,620
edited 2010-07-20 08:42 in General Discussion
I found this description of TSMC's new fab for the next generation of 28nm chips very interesting:

www.edn.com/article/509809-TSMC_begins_construction_on_9B_300_mm_fab.php

The amount of money it costs to develop those facilities is staggering.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Leon Heller
Amateur radio callsign: G1HSM

Post Edited (Leon) : 7/20/2010 12:41:44 AM GMT

Comments

  • HollyMinkowskiHollyMinkowski Posts: 1,398
    edited 2010-07-19 23:27
    "will have 104,000-sq meters in clean room space. "

    OMG, that's going to be a big place.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    http://justasm.blogspot.com/
  • LeonLeon Posts: 7,620
    edited 2010-07-19 23:45
    And 8,000 workers!

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Leon Heller
    Amateur radio callsign: G1HSM
  • HollyMinkowskiHollyMinkowski Posts: 1,398
    edited 2010-07-19 23:49
    To bad Parallax doesn't own a fab like that smile.gif

    Just think of the fun stuff that would get made.

    I expect that in the years to come ordinary people
    will be able to turn out intricate devices like processors
    on their desktop using a nano assembler.
    (I can't wait 4it)


    Leon, I see that you are in Britain?
    You stay up quite late and surf the
    web, like I do wink.gif

    (I can't really write good code till after midnight)

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    http://justasm.blogspot.com/

    Post Edited (HollyMinkowski) : 7/19/2010 11:54:46 PM GMT
  • LeonLeon Posts: 7,620
    edited 2010-07-20 00:15
    I suffer from insomnia.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Leon Heller
    Amateur radio callsign: G1HSM
  • ercoerco Posts: 20,260
    edited 2010-07-20 00:25
    Nano assemblers? NANO ASSEMBLERS? How about Heathkit assemblers?

    I'm still awaiting the resurgence of vacuum tubes, morse code and 40 meter communication after the upcoming Atomic Flame Deluge. My old Ham radio gear is ready & rarin'!

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    ·"If you build it, they will come."
  • LeonLeon Posts: 7,620
    edited 2010-07-20 00:34
    Morse is still popular here in the UK, even though it isn't required for a full license any more. Valves/tubes should survive a nuclear strike better than semiconductors, of course.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Leon Heller
    Amateur radio callsign: G1HSM

    Post Edited (Leon) : 7/20/2010 12:40:06 AM GMT
  • UnknownUnknown
    edited 2010-07-20 04:53
    Will SD cards survive an Electro Magnetic Pulse?· I know there are Faraday Cages but I wouldn't know how big you would have to build one to survive a large pulse because the general public hasn't prepared for one before.

    Anyway, I'll probably be burned to a crisp if it ever happens.
  • Beau SchwabeBeau Schwabe Posts: 6,568
    edited 2010-07-20 05:26
    If it helps, we are using TSMC's process rules for the 180nm Propeller II.

    I know that there are smaller and smaller process out there, but what starts to happen as you go smaller and smaller is that certain silicon characteristics that were once considered 'recessive' become more and more 'dominant' and take on more quantum characteristics as you get deeper into the silicon. Without good empirical testing, it's difficult to simulate and characterize this behavior, especially when the process is relatively new. In this sense it is beneficial to piggy-back off of the larger companies by letting them throw $$$ at it to work out the process characteristics.

    The smallest process I have worked on was 65nm about 6 years ago, but I assure you it was a custom very expensive process I did while at National Semiconductor. The main stream process at the time was 110nm, and this was something that had just been started... 180nm being the prior main stream process.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Beau Schwabe

    IC Layout Engineer
    Parallax, Inc.
  • LeonLeon Posts: 7,620
    edited 2010-07-20 08:42
    Thanks, Beau. It seems to be the makers of FPGAs, like Xilinx and Altera, that use the new processes first, presumably because the designs are very regular and it's relatively easy to predict the characteristics. People keep saying that because of the problems with smaller geometries we will reach a limit, eventually, but they seem to find ways round them. Physical properties apart, it must be difficult deciding which process to use from a cost point of view, with the larger ones being cheaper per wafer, but yielding fewer parts.
    Chuckz said...
    Will SD cards survive an Electro Magnetic Pulse? I know there are Faraday Cages but I wouldn't know how big you would have to build one to survive a large pulse because the general public hasn't prepared for one before.


    Anyway, I'll probably be burned to a crisp if it ever happens.

    When I worked on military comms at Racal, our equipment had to survive a nuclear EMP, albeit at reduced performance. Our prototypes were actually tested for immunity, using a special government facility (they don't explode one of their devices, of course).

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Leon Heller
    Amateur radio callsign: G1HSM

    Post Edited (Leon) : 7/20/2010 8:52:04 AM GMT
Sign In or Register to comment.