1. This forum section is a read-only archive which contains old newsgroup posts. If you wish to post a query, please do so in one of our main forum sections (here). This way you will get a faster, better response from the members on Motherboard Point.

Intel's new 3D transistors will leave competiton for dead

Discussion in 'Intel' started by Nomen Nescio, May 13, 2011.

  1. Nomen Nescio

    Nomen Nescio Guest

    This is one of the biggest breakthroughs in integrated circuits
    since I don't know when.
    In 2020 we will have the equivalent of a pizzabox cluster rack
    on a single package the size of today's i7.
    Mind-boggling indeed.
     
    Nomen Nescio, May 13, 2011
    #1
    1. Advertisements

  2. Nomen Nescio

    Yousuf Khan Guest

    On 11-05-13 11:03 AM, Nomen Nescio wrote:
    > This is one of the biggest breakthroughs in integrated circuits
    > since I don't know when.
    > In 2020 we will have the equivalent of a pizzabox cluster rack
    > on a single package the size of today's i7.
    > Mind-boggling indeed.
    >


    Might wanna wait until it actually comes out before passing judgement.
    Sometimes these announcements' biggest benefit is just the announcement
    itself.

    The benefit that Intel is touting here is that it'll have a
    fully-depleted channels. Well other technologies have provided these
    similar benefits such FDSOI (Fully Depleted Silicon On Insulator)
    technology.

    http://www.eetimes.com/electronics-blogs/dew-point/4215728/Intel-putting-fins-on-at-22-nm

    Now, it might be a bigger advantage than FDSOI, we won't know that until
    comparisons can be made.

    Yousuf Khan
     
    Yousuf Khan, May 14, 2011
    #2
    1. Advertisements

  3. Nomen Nescio

    Robert Myers Guest

    On May 14, 2:07 am, Yousuf Khan <> wrote:
    > On 11-05-13 11:03 AM, Nomen Nescio wrote:
    >
    > > This is one of the biggest breakthroughs in integrated circuits
    > > since I don't know when.
    > > In 2020 we will have the equivalent of a pizzabox cluster rack
    > > on a single package the size of today's i7.
    > > Mind-boggling indeed.

    >
    > Might wanna wait until it actually comes out before passing judgement.
    > Sometimes these announcements' biggest benefit is just the announcement
    > itself.
    >
    > The benefit that Intel is touting here is that it'll have a
    > fully-depleted channels. Well other technologies have provided these
    > similar benefits  such FDSOI (Fully Depleted Silicon On Insulator)
    > technology.
    >
    > http://www.eetimes.com/electronics-blogs/dew-point/4215728/Intel-putt...
    >
    > Now, it might be a bigger advantage than FDSOI, we won't know that until
    > comparisons can be made.
    >

    As things stand now, there is a hole in the AMD/IBM/ARM anything to
    predict Intel's demise that is big enough to drive a truck through.

    If you want to build *one* of something, and don't mind what you pay
    for it, hire IBM.

    If your performance requirements are modest and/or low, take your pick
    between AMD and ARM.

    If you expect performance requirements to keep increasing and you
    realize that there is only one customer in the world who doesn't care
    about cost (and even that customer is getting edgy), don't bet on
    anyone but Intel, tri-gate transistors and/or any semiconductor
    technology you can imagine notwithstanding. There is no hint (and
    there never has been a hint) anywhere that anyone can compete with
    Intel simultaneously on manufacturing cost and performance.

    It's true. No one knows how any particular technology will play out.
    I'm sure that IBM and Global Foundries will eventually build the
    technological equivalent of tri-gate transistors. The questions are
    when and at what cost. If someone eventually beats Intel, it will be
    because someone has figured out how to do bleeding edge manufacturing
    at a lower cost than Intel. Don't hold your breath.

    Robert.
     
    Robert Myers, May 15, 2011
    #3
  4. Nomen Nescio

    Yousuf Khan Guest

    On 15/05/2011 1:07 PM, Robert Myers wrote:
    > It's true. No one knows how any particular technology will play out.
    > I'm sure that IBM and Global Foundries will eventually build the
    > technological equivalent of tri-gate transistors. The questions are
    > when and at what cost. If someone eventually beats Intel, it will be
    > because someone has figured out how to do bleeding edge manufacturing
    > at a lower cost than Intel. Don't hold your breath.


    It's actually already right there for the IBM alliance: SOI. More
    specifically FDSOI (fully-depleted SOI). They've been using this
    technology since the 90nm node. Intel has been avoiding SOI since that
    time, and trying to find every technology that give it the equivalent
    featureset as SOI without having to use SOI. They came up with HKMG as
    the big marketing announcement in the last node, and now FinFETs.

    FinFETs increase the gate-source capacitance in exchange for higher
    drive current. It's not going to have nearly the same level of advantage
    as HKMG gave them, and HKMG really didn't give them all that much of an
    advantage in power-savings or performance during the last node either,
    despite all of their hype for that technology too. As I said, these
    technology announcements are for the sake of the announcement, nothing else.

    Yousuf Khan
     
    Yousuf Khan, May 16, 2011
    #4
  5. Nomen Nescio

    Robert Myers Guest

    On May 15, 9:57 pm, Yousuf Khan <> wrote:
    > On 15/05/2011 1:07 PM, Robert Myers wrote:
    >
    > > It's true.  No one knows how any particular technology will play out.
    > > I'm sure that IBM and Global Foundries will eventually build the
    > > technological equivalent of tri-gate transistors.  The questions are
    > > when and at what cost.  If someone eventually beats Intel, it will be
    > > because someone has figured out how to do bleeding edge manufacturing
    > > at a lower cost than Intel.  Don't hold your breath.

    >
    > It's actually already right there for the IBM alliance: SOI. More
    > specifically FDSOI (fully-depleted SOI). They've been using this
    > technology since the 90nm node. Intel has been avoiding SOI since that
    > time, and trying to find every technology that give it the equivalent
    > featureset as SOI without having to use SOI. They came up with HKMG as
    > the big marketing announcement in the last node, and now FinFETs.
    >
    > FinFETs increase the gate-source capacitance in exchange for higher
    > drive current. It's not going to have nearly the same level of advantage
    > as HKMG gave them, and HKMG really didn't give them all that much of an
    > advantage in power-savings or performance during the last node either,
    > despite all of their hype for that technology too. As I said, these
    > technology announcements are for the sake of the announcement, nothing else.
    >

    A friend recently inquired of me as to the nature of my computers. I
    have a few. He was most interested in the one AMD box I own. It's
    obsolete and I never bought it particularly for performance to begin
    with. Other than the fact that the processor is dual core and that
    Microsoft online tests concluded that said box could run Vista were I
    so much of a fool as to want to upgrade, I know next to nothing about
    the machine.

    Here is the secret I have been struggling to conceal all these years:
    it's better for me to buy from one vendor because it's hard enough for
    me to keep up with the technology at Intel. That is to say, I'm
    simply too stupid to keep track of what the also-rans in this business
    are up to. OK? That, and the fact that Intel has a clean balance
    sheet, not something that is an unknowable combination of the combined
    accounting trickery of IBM, AMD, and Global Foundries. Had it had to
    run microelectronics as a money-making business, IBM would have been
    out of the game long ago. I conclude, from repeated interactions from
    you, that you simply don't process facts that are inconvenient for
    your world view. Why don't we let it go at that? I have better
    things to do with my neurons and you simply aren't interested in
    anything that doesn't fit your preconceived notions. I hope that
    whoever is following Intel at Goldman Sachs is either smarter or more
    honest than you are, but I'm not going to hold out hope.

    If and when it becomes clear that ARM is a significant player in
    problems of interest to me, maybe I'll take some time to learn
    something beyond what real computer architects have already told me.

    Robert..
     
    Robert Myers, May 20, 2011
    #5
  6. Yousuf Khan wrote:
    > On 11-05-13 11:03 AM, Nomen Nescio wrote:
    >> This is one of the biggest breakthroughs in integrated circuits
    >> since I don't know when.
    >> In 2020 we will have the equivalent of a pizzabox cluster rack
    >> on a single package the size of today's i7.
    >> Mind-boggling indeed.
    >>

    >
    > Might wanna wait until it actually comes out before passing judgement.
    > Sometimes these announcements' biggest benefit is just the announcement
    > itself.
    >
    > The benefit that Intel is touting here is that it'll have a
    > fully-depleted channels. Well other technologies have provided these
    > similar benefits such FDSOI (Fully Depleted Silicon On Insulator)
    > technology.
    >
    > http://www.eetimes.com/electronics-blogs/dew-point/4215728/Intel-putting-fins-on-at-22-nm
    >
    >
    > Now, it might be a bigger advantage than FDSOI, we won't know that until
    > comparisons can be made.
    >

    The one advantage I took from the various releases was low cost manufacturing
    using (mostly) established processes and equipment. If that prediction turns out
    to be correct, then it would seem to have a market advantage.

    I freely admit I haven't followed SOI of any kind in ages, and press releases
    are written by optimists and filtered through mostly tech challenged publication
    channels, but Intel does have a pretty good record of getting things from the
    lab to the market, even if the market sometimes rejects the idea of change.

    The most interesting thing to me was the implication that power efficiency was
    going to be better. I definitely see that as becoming more important than raw
    speed gains in both cloud and home markets. And Intel is putting more effort
    into trying to compete with ARM, as people buy more toys, low power is critical.
     
    Bill Davidsen, May 24, 2011
    #6
  7. Robert Myers wrote:

    > If and when it becomes clear that ARM is a significant player in
    > problems of interest to me, maybe I'll take some time to learn
    > something beyond what real computer architects have already told me.
    >

    So you don't own a smart phone, microwave, HDTV, cable box, or anything like
    that? Intel has said they are going to put more effort into competing with ARM
    in those markets, they think those are problems of interest.
     
    Bill Davidsen, May 24, 2011
    #7
  8. Nomen Nescio

    Robert Myers Guest

    On May 24, 10:57 am, Bill Davidsen <> wrote:
    > Robert Myers wrote:
    > > If and when it becomes clear that ARM is a significant player in
    > > problems of interest to me, maybe I'll take some time to learn
    > > something beyond what real computer architects have already told me.

    >
    > So you don't own a smart phone, microwave, HDTV, cable box, or anything like
    > that? Intel has said they are going to put more effort into competing with ARM
    > in those markets, they think those are problems of interest.


    So far as I have been able to understand the business, Intel has not
    ever been successful in any market other than x86 and directly
    supporting technologies, at least not in recent memory.

    The only argument that makes sense for Intel to compete with ARM is
    that arm is a potential threat to x86--the core of its business.

    I don't even own a dumb cell phone, and the only thing I require from
    my microwave oven is that it work. I don't watch television. The
    only place where ARM might be of interest to me is in very low power
    massively parallel processing (often erroneously referred to as
    supercomputing).

    Robert.
     
    Robert Myers, May 25, 2011
    #8
  9. Nomen Nescio

    Yousuf Khan Guest

    On 5/24/2011 10:51 AM, Bill Davidsen wrote:
    > The one advantage I took from the various releases was low cost
    > manufacturing using (mostly) established processes and equipment. If
    > that prediction turns out to be correct, then it would seem to have a
    > market advantage.


    Well, it's an advantage for Intel vs. Intel's current manufacturing
    process. AMD has been using SOI for most of this decade, so they have a
    different manufacturing process, so FinFET technology is not so
    necessary for them. It seems FinFET and SOI address the same problems,
    although it's not inconceivable that you can combine both of them, and
    get a bigger effect from them.

    > The most interesting thing to me was the implication that power
    > efficiency was going to be better. I definitely see that as becoming
    > more important than raw speed gains in both cloud and home markets. And
    > Intel is putting more effort into trying to compete with ARM, as people
    > buy more toys, low power is critical.


    I don't think Intel will be able to compete against ARM in ARM's market
    any more than ARM will be able to compete against Intel in Intel's
    market. There is some talk that Apple may put one of its ARM chips in
    one of its next generation laptops. I figure that'll probably be a
    success, simply because of the Apple legions willing to buy anything
    Apple puts out. However, I doubt it's going to anything more than an
    Apple-only phenomenon, much like the tablet market. A lot of new tablets
    out, but few successes beyond Apple's Ipad.

    Yousuf Khan
     
    Yousuf Khan, May 28, 2011
    #9
    1. Advertisements

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Similar Threads
  1. MBoyd00000
    Replies:
    2
    Views:
    940
    MBoyd00000
    Apr 28, 2004
  2. crispy

    dead dead dead

    crispy, Sep 9, 2003, in forum: Laptops
    Replies:
    2
    Views:
    806
    crispy
    Sep 10, 2003
  3. Replies:
    1
    Views:
    414
  4. Larry LaMere

    Dead Battery - Leave In Or Not

    Larry LaMere, May 14, 2006, in forum: Laptops
    Replies:
    5
    Views:
    473
    Larry LaMere
    May 16, 2006
  5. boni

    What's the best lap top computer to buy?

    boni, Jan 9, 2008, in forum: Motherboard Purchase Recommendations
    Replies:
    0
    Views:
    517
  6. Replies:
    0
    Views:
    1,417
  7. jacatone

    How do P.O.S.T. cards work?

    jacatone, Apr 20, 2010, in forum: Motherboard General Discussion
    Replies:
    1
    Views:
    1,173
    onadal
    Sep 2, 2010
  8. fbsduser

    Which is better (for testing amcient x86 OS's)

    fbsduser, Apr 20, 2011, in forum: Motherboard Purchase Recommendations
    Replies:
    0
    Views:
    1,067
    fbsduser
    Apr 20, 2011
Loading...