GloFo, Samsung partner up on high-perf 28-nm process

Santa Clara — At the GlobalFoundries Technology Conference here in Silicon Valley, GlobalFoundries revealed that it has broadened its partnership with Samsung on 28-nm high-K metal gate manufacturing.

Last year, the two companies, along with IBM and STMicroelectronics, announced plans to sync up their fabs to facilitate production using a low-power, 28-nm HKMG process. Today, GlobalFoundries says it’s teaming up with Samsung on a high-performance, low-leakage 28-nm HKMG manufacturing process that complements the low-power process announced in 2010.

This new, high-power process was “specifically developed for mobile applications,” and it purportedly offers “60 percent of active power reduction at the same frequency or 55 percent of performance boost at the same leakage over 45nm low power (LP) SoC designs.” GlobalFoundries sees the process as suitable for high-end smart phones, tablets, and even notebooks.

The process will be qualified at GlobalFoundries’ fabs in Dresden, Germany and in Saratoga County, New York, as well as at Samsung’s facilities in Giheung, South Korea and in Austin, Texas. The two companies also plan to whip up “a comprehensive System-on-a-Chip (SoC) design platform . . . to enable seamless customer design-in to the multiple global manufacturing sites.”

We’ll have more from the GlobalFoundries Technology Conference for you soon.

Comments closed
    • ronch
    • 8 years ago

    This news probably has more to do with making non-CPU circuits, but I’d like to touch a bit on GF’s CPU-centric processes.

    Intel is usually about 12-18 months ahead of GF/AMD in terms of process technology, so their (Intel) 22nm, having missed 2011, is next year, and 15nm should come in around 2014, if not earlier. GF/AMD has always struggled to keep up with Intel, but now that ATIC can pour in more gold to make smaller silicon, perhaps they can expedite R&D for 15nm etc. AMD historically had 2-year gaps between process shrinks, having introduced 65nm in 2006, 45nm in 2008 (although 45nm Phenom II chips only came out in early 2009 — a bit late), and 32nm only in 2011. But lately they seem to be slipping a bit off schedule (not to mention their new architectural rollouts seem to be missing their schedules as well.. Barcelona, and now Bulldozer, is late). If GF sticks to their 2-year gap they should be rolling out 22nm in 2013, a year behind Intel. I’m sure they have concurrent teams working on future processes, but I just hope they can expedite 15nm so it can come out earlier than 2015 (and perhaps closer to Intel’s 2014 15nm launch).

      • bcronce
      • 8 years ago

      It was my understanding that Intel held off on 22nm because of lack of competition. I’ve read, a few months back, that Intel’s 22nm process is ready to go but Intel postponed the release of their chip at 22nm to “refine” their architecture a bit more.

      I also read that Intel’s 14nm process is almost ready for prime time. So the 22nm to 14nm release effectively should be shorter(hard to not pun this) than the 32nm to 22nm release. Which doesn’t make much sense because the 22nm to 14nm is harder, unless their 22nm was on time but their chip wasn’t ready.

        • OneArmedScissor
        • 8 years ago

        Intel intentionally held off making more money? Uh huh.

        What you probably read is that the manufacturing process itself is ready. They still have to have all the manufacturing equipment built and installed for a gajillion fabs to be able to keep up with the demand they have. Look at how long it took them to get 32nm going for their entire PC and server lines. They still don’t do Atom and Itanium on 32nm. Intel is a huge company with numerous markets to supply. They can’t possibly cover them all at once with a brand new manufacturing process that hasn’t been proved at mass production scale.

          • bcronce
          • 8 years ago

          “Intel intentionally held off making more money?”

          Because competing with one’s self is a great business practice. So, Intel pushes out a new CPU with nearly zero competition from AMD, only to push down its own 32nm chips.

        • NeelyCam
        • 8 years ago

        [quote<]I also read that Intel's 14nm process is almost ready for prime time.[/quote<] You have a link? I haven't heard that anywhere - I though 14nm was expected to be ready in 2014. [quote<]So the 22nm to 14nm release effectively should be shorter(hard to not pun this) than the 32nm to 22nm release. Which doesn't make much sense because the 22nm to 14nm is harder, unless their 22nm was on time but their chip wasn't ready.[/quote<] Actually, I would guess the transition from 32nm to 22nm might have been harder because of the introduction of FinFETs... I'd be surprised if they make another major change to the transistor topology at 14nm. If they "just" scale from 22nm to 14nm, it might not be that hard.

          • bcronce
          • 8 years ago

          “32nm to 22nm might have been harder because of the introduction of FinFETs”

          Interesting to know. Learn something new every day 😛

      • NeelyCam
      • 8 years ago

      If I remember correctly (I can’t find the link), GloFo revised their 22nm plans after Intel’s FinFET announcement, and Rich Wargo said on SemiAccurate forums that AMD is going for FD-SOI on 22nm:

      [url<]http://semiaccurate.com/forums/showpost.php?p=127031&postcount=19[/url<] I think the introduction of FD-SOI was expected to cause a delay, with 22nm products coming out in 2014. 15nm might be arriving "early" to close the gap somewhat, but I think all the companies (GloFo, TSMC, Samsung) were thrown off-track at 22nm by Intel's FinFETs.

        • ronch
        • 8 years ago

        I hope that info is wrong. How could the industry fall behind Intel even further? Perhaps ATIC should pour in more oil money.

    • luisnhamue
    • 8 years ago

    All of this guys just to try fight intel. wow,

      • NeelyCam
      • 8 years ago

      I think their main target is TSMC…

        • willyolio
        • 8 years ago

        only because Intel’s still out of reach even if they partnered with TSMC.

          • Farting Bob
          • 8 years ago

          GloFo isnt AMD. Chip production is a huge industry and while Intel leads the way in process tech and profits GloFo can still make billions while being a year or 2 behind Intel. AMD might suffer as a result, but GloFo wont be hurt, because everyone except Intel (who mostly just stick to their own thing) is behind or on par with them.

    • NeelyCam
    • 8 years ago

    Looks like a clone of this (IEDM 2009):

    [url<]http://www.intel.com/content/dam/doc/technology-brief/32nm-soc-platform-technology-paper.pdf[/url<]

      • codedivine
      • 8 years ago

      P( NeelyCam comment about Intel | Story about process tech) = 1.0

        • indeego
        • 8 years ago

        hah!

        • NeelyCam
        • 8 years ago

        Too true.

    • HisDivineOrder
    • 8 years ago

    Intel’s is better. It’s in 3d. 😉

      • NeelyCam
      • 8 years ago

      Clearly. 🙂 But this is competing with Intel’s 32nm SOC process.

        • willmore
        • 8 years ago

        SOC? Silicon on Carbon?

          • JumpingJack
          • 8 years ago

          System On a Chip

      • Farting Bob
      • 8 years ago

      Spy kids was recently released in “4D”, that should be AMD’s next move.

        • NeelyCam
        • 8 years ago

        Yeah, maybe they could release 22nm in 2008

          • ronch
          • 8 years ago

          Did I miss something here? 2008 was 3 years ago.

            • NeelyCam
            • 8 years ago

            You missed the 4th dimension

        • The Dark One
        • 8 years ago

        If the 4D refers to the smellovision thing they did with that movie, then my Athlon T-bird could do that years ago. It was designed to release a very noticeable smell in case the heatsink had fallen off.

      • bcronce
      • 8 years ago

      How does this guy get voted down? It was clearly non-biased and totally funny.

    • ltcommander.data
    • 8 years ago

    Maybe I missed it, but it would be nice if their press release actually noted a expected timeframe the process will be operational. Instead they just talk about the 2010 low-power 28nm announcement and the H1 2011 32nm process.

      • Risme
      • 8 years ago

      There will be roadmap and possibly other presentation slides floating around the internet later on that will shed some light on that, i believe, probably including TR.

      So did TR get someone there in person this time around? If so, that would be great.

      Personally i’m most interested in the Technology Solutions, Design Solutions: Collaborating to Fully Enable Your Future Designs, Global Capacity: Enabling New Sources of Customer Value • 300mm and Beyond and CEO Panel: “Design Enablement Challenges and Future Solutions” sessions.

      The agenda can be found here: [url<]http://www.globalfoundries.com/gtc2011/agenda.aspx[/url<] Too bad there's no live webcast of the conference, at least i haven't found one.

      • ronch
      • 8 years ago

      Don’t make promises you can’t keep..

Pin It on Pinterest

Share This