GloFo preps 14-nm FinFET process for 2013

Three-dimensional transistors are coming!

Okay, so maybe they’re already here. You can already find ’em inside every Ivy Bridge processor on the market today. Independent foundries are catching up to Intel, though. Earlier this morning, GlobalFoundries announed that it’s “accelerated its leading-edge roadmap” with the introduction of a 14-nm FinFET fab process for mobile devices.

The process is dubbed 14nm-XM, short for 14-nm eXtreme Mobility, and it’s “based on a modular technology architecture that uses a 14nm FinFET device combined with elements of GLOBALFOUNDRIES’ 20nm-LPM process.” (FinFET is just another word for three-dimensional transistors. The term describes field-effect transistors with a three-dimensional, fin-like conducting channel.) According to GlobalFoundries, the 14nm-XM process can improve device battery life by 40-60% compared to GloFo’s existing 20-nm process with planar transistors.

GlobalFoundries goes on to note that 14nm-XM test silicon is already “running through” its Fab 8 facility in upstate New York. The firm expects the first tape-outs from customers (that is, customers sending in their final chip designs for manufacture) some time next year.

As we reported last month, GlobalFoundries has inked a partnership with ARM to speed up the migration of mobile processors to FinFET technology. GlobalFoundries says the FinFET design “intrinsically operates at a lower voltage with minimal current leakage, which translates into longer battery life for mobile applications.” Of course, GlobalFoundries doesn’t just cater to ARM SoC vendors—it also manufactures all of AMD’s A- and FX-series processors.

Comments closed
    • mark625
    • 7 years ago

    I call BS on this being ready in 2013. They can barely push out 28nm at this point, and where are their 22nm parts?

    They’ll be lucky to be in production of 14nm products in 2015, much less in 2013.

    No wonder AMD spun off GloFo.

    • DavidC1
    • 7 years ago

    2009-32nm availability, mid-late 2011 product availability
    2014-14nm availability, mid-late 2015 product availability

    The timeline that a FOUNDRY says is different from what a developing company says USING the foundry. Because Foundries are there to spread info to others(to gain business before other foundries).

    You don’t see that with Intel because they don’t have a foundry service, at least not a big one, and not at all until very recently. Intel sends you news about the end result of the process team + design team, not the process team alone because they don’t need to. They sell you finished chips, not process to make your chip, right?

    • WillBach
    • 7 years ago

    Glo-Fo is ready [b<]to take orders[/b<] at the end of next year, not fulfill them. They're probably taking orders of 28 nm and 20 nm parts now.

    • Sam125
    • 7 years ago

    Sounds like good news for GloFo if they can manage to execute well. Intel seems to be losing a bit of their process advantage so now would be a good time for the smaller players to close the gap a bit.

    Speculation on my part – But perhaps that Intel realizes that they’re losing their foundry advantage which is why they’re now open to fabbing devices for third parties? Probably not but it’s an interesting thought nonetheless and makes a lot of sense from a business perspective when your process is no longer leaps and bounds more advanced than the competition.

    • slaimus
    • 7 years ago

    Does this mean FD-SOI is out at IBM/GF?

      • NeelyCam
      • 7 years ago

      I heard a rumor in S|A that it’d be used on 20nm

    • odizzido
    • 7 years ago

    If they actually get properly working 14nm products out in useful numbers in 2013 I will be amazed. More than amazed. Flabbergasted perhaps.

      • Farting Bob
      • 7 years ago

      So would GF because they are only saying they will be doing first test runs on actual customer designs next year. Even if they meet that target they will still be some way off selling 14nm chips to consumers.

    • ronch
    • 7 years ago

    Ok. Now let’s see them pull it off.

    • Star Brood
    • 7 years ago

    Now if only AMD would jump on board with a desktop CPU with this process, we would see some competition at long last.

      • Silus
      • 7 years ago

      Why ? Are you under the impression that a new fab process somehow fixes or diminishes all the shortcomings of AMD’s or any given chip architecture ?

      If you are, well…you’re wrong 🙂

        • Star Brood
        • 7 years ago

        They would be more energy efficient and have more room for the IGP. Both are things AMD really can take advantage of.

          • NeelyCam
          • 7 years ago

          GloFo and Intel 32nm processes had roughly equal performance, but SB had better power efficiency and much higher performance than BD.

            • Star Brood
            • 7 years ago

            I’m not saying that AMD is Intel’s equal or superior on a given node, I’m saying that AMD has a nice chance if they can publish 14nm before Intel does, otherwise they will always be behind.

            • chuckula
            • 7 years ago

            I will bet you a full keg of any micro-brew you would like that AMD not only does not beat Intel to 14 nm, but that Intel has a second-generation 14nm part out on the market before AMD has its first 14nm part available for sale.

            Note that I’m not necessarily talking about GloFo as a whole, just AMD. The side-bet would be a rack of ribs from the best rib-joint you can find that Intel has shipping 14 nm CPUs at least one year before GloFo puts out *any* 14 nm part (including cut-down ARM chips that are nowhere near as complicated as AMD’s products) for commercial sale (experimental test runs don’t count, you have to be able to buy the products in the open market).

            • NeelyCam
            • 7 years ago

            It’s starting to sound like TR is becoming an online betting site..

            We have to be careful so that the puritans don’t shut down US access to the site. I’d be very sad without my daily shortbread

            • BobbinThreadbare
            • 7 years ago

            So did the Phemon 2 (what was the code name, k10?).

            Here’s waiting for Steamroller! *sigh*

      • SetiroN
      • 7 years ago

      I guess you missed the XM (eXtreme Mobility) monicker.

    • flip-mode
    • 7 years ago

    Typo? Should that be 2023? Or maybe GloFo to start prep for 14nm finfet in 2013? GloFo and short time frames have, like, never conspired together.

    • chuckula
    • 7 years ago

    I think there is a small error in this article. I think you really meant to say that GloFo is prepping [b<]PowerPoint Presentation Slides about a [/b<] 14-nm FinFET process for [b<]a meeting in[/b<] 2013. That's about what I would expect here.

      • tay
      • 7 years ago

      Hahahaha. Perfect.

      • bcronce
      • 7 years ago

      I was about to say, if they actually could pull off 14nm in 2013, it would given Intel competition.

      I to am skeptical about a 2013 effective roll out from GloFo.

    • DancinJack
    • 7 years ago

    Oh gosh. Even manufacturing processes are getting stupid names now. Marketing ftw.

    e: Why is GF in all caps in only once instance?

      • Damage
      • 7 years ago

      That instance is a quotation. We don’t follow their… insanity… in our own writing here at THE TECH REPORT.

        • NeelyCam
        • 7 years ago

        Those caps are bound to turn SSK on

        • wierdo
        • 7 years ago

        Lol, when I saw the GF in caps it made me think of them talking like the guys in the Powerthirst parody…

        [url<]http://www.youtube.com/watch?v=qRuNxHqwazs[/url<]

        • willmore
        • 7 years ago

        Then it would be a good time to break a whole can of [sic].

    • NeelyCam
    • 7 years ago

    [quote<]"Early process design kits (PDKs) are available now, with customer tape-outs expected in 2013."[/quote<] Is this like: [url<]http://asic.yonsei.ac.kr/bbs/view.php?id=news&page=4&sn1=&divpage=1&sn=off&ss=on&sc=on&select_arrange=headnum&desc=asc&no=47[/url<] [quote<]"Earlier this year the company said it would be ready to accept orders for 32nm bulk production in late 2009."[/quote<] With that 32nm, the yields were finally decent [b<]three[/b<] years later GloFo "ready" != TSMC "ready". I hope their customers learned from AMD's 32nm fiasco.

      • OneArmedScissor
      • 7 years ago

      Eh…what do you know was made with their 32nm [b<]bulk[/b<] that is your frame of reference?

        • NeelyCam
        • 7 years ago

        Ah good point. I missed that in the race to get FP..

        The linked article implied that GloFo 32nm SOI was actually supposed to be [i<]ahead[/i<] of the 32nm bulk... making GloFo's timeliness even worse And I'm not sure that 32nm bulk ever materialized.. I haven't heard of any products on it.

          • guardianl
          • 7 years ago

          “All of our efforts around next-gen graphics and wireless are focused on 28nm with HKMG and we no longer have a 32nm bulk process. We removed this off our roadmap due to lack of customer demand as most are making the jump from 40/45nm right to 28nm,” said Jon Carvill, the head of public relations at Globalfoundries.

          Source: [url<]http://news.softpedia.com/news/Globalfoundries-32nm-Bulk-Process-Canceled-139063.shtml[/url<] AFAIK Trinity is AMDs last SOI product. Source: [url<]http://www.guru3d.com/news-story/amd_to_adopt_28nm_bulk_cmos_in_2013.html[/url<]

      • tay
      • 7 years ago

      Agreed, won’t be ready till 2016 based on past performance.

    • Jigar
    • 7 years ago

    Did they just skip the process and jump to 14nm ?

      • chuckula
      • 7 years ago

      I don’t know if they jumped to 14 nm or just jumped the shark.

      • NeelyCam
      • 7 years ago

      No:

      [quote<]According to GlobalFoundries, the 14nm-XM process can improve device battery life by 40-60% compared to GloFo's [b<]existing[/b<] 20-nm process with planar transistors.[/quote<] Of course, what "existing" means is anybody's guess. I don't see any AMD roadmaps showing imminent release of 20nm CPUs, for instance.. This is all GloFo marketing talk that we've seen before. My guess is that their 14nm FinFET process is probably around the same point in development as, say, Intel's 10nm process. They just usually tend to make it sound like they are farther than they really are

        • DancinJack
        • 7 years ago

        AMD isn’t the only company having chips made at GF, Neely.

          • NeelyCam
          • 7 years ago

          I used AMD only as an example. But if you know of some other companies releasing GloFo fabbed 20nm soon, please let me know

            • DancinJack
            • 7 years ago

            No, you’re right. I don’t know anyone that’s ready to release a 20nm GloFo made chip. Then again most of the news I’ve seen on 20nm and GloFo is mentioning ARM.

      • ronch
      • 7 years ago

      No. Intel just had the silly idea of doing 22nm.

        • DancinJack
        • 7 years ago

        I think it’s the other way around isn’t it? I think 22nm is the first full node after 32. 28 and 20 are half.

          • ronch
          • 7 years ago

          Er.. It was a joke.

      • mnecaise
      • 7 years ago

      Does look like it. The information on FAB 8, which is the new NY fab, says they’re target was 28nm and below. They discuss 20nm node being released FY12 on the globalfoundries website. And now they’re planning to have 14nm for next year. That’s pretty damn quick — unusually quick for GloFo. I’m going out on a limb with this supposition: Given all the fab space they have dedicated to larger nodes, maybe they decided to jump in the deep end and skip a couple nodes / half nodes in order to try to grab more market share.

Pin It on Pinterest

Share This