Otellini: 22-nm Ivy Bridge production began last quarter

While AMD’s first 32-nm processors were starting to hit retail shelves last quarter—and AMD apparently had to contend with yield issues at GlobalFoundries—Intel was busy churning out 22-nm Ivy Bridge chips. At least, that’s what Intel CEO Paul Otellini stated during his company’s third-quarter earnings conference call, according SeekingAlpha’s transcript:

During the third quarter, we began volume production of Ivy Bridge on our 22 nanometer process technology. 22 nanometers will usher-in the era of 3-D transistors, which will pay dividends in power, performance and density for generations to come.

Intel actually demonstrated running Ivy Bridge systems back in May, so working chips were definitely coming out of Intel’s fabs at that time. Volume production is a whole other story, though, and that’s what Otellini talked about this week.

Now, don’t hold your breath—the latest whispers from the rumor mill suggest Ivy Bridge might not become commercially available until next spring. That ought to give AMD a little time to sort out its 32-nm manufacturing problems with GlobalFoundries, and maybe crank out faster, cheaper FX-series chips that put up more of a fight. (Thanks to Fudzilla for the tip.)

Comments closed
    • ronch
    • 8 years ago

    I was really hoping BD will present a credible alternative to IB. It’s a good thing for AMD that I’m in no terrible hurry to upgrade, which means I can hold out until Steamroller rolls out. Then again, Haswell should also be out by then. I hope I can still stick with AMD on my next CPU upgrade, but if PD/SR still turns out to be a bomb/s, well, I don’t know.

    • StashTheVampede
    • 8 years ago

    My next purchase will have this chip in it. Ivy Bridge is going to tier 1 customers first and a new iMac is in my future.

    • ssidbroadcast
    • 8 years ago

    Otellini ([i<]to AMD's new CTO[/i<]): "Oh, I [i<]do[/i<] believe the [s<]shields[/s<] chips will be fully functional before your friends arrive!"

      • Coran Fixx
      • 8 years ago

      +1 internet

      I think Intel was just waiting to see if they needed to rush Ivy Bridge out to crush the alliance .

        • NeelyCam
        • 8 years ago

        You got it wrong – Intel [i<]is[/i<] the Alliance, and Ivy Bridge is the White Star. BD is the dog slow Shadows ship that crashed into Mars.

          • mutarasector
          • 8 years ago

          Oooooo…. a Bab5 reference? *Kewl*…

    • lycium
    • 8 years ago

    Intel: shut up and take my money!

    • kamikaziechameleon
    • 8 years ago

    I’m holding out for IB I think. Its sad to see how behind SB-E is. I was hoping to pick up one of those.

    • ronch
    • 8 years ago

    Even die-hard AMD fans planning to buy a BD will not find this easy to pass up unless IB will be priced to take advantage of BD’s situation.

    • mcforce0208
    • 8 years ago

    Seeing as I was waiting so patiently for bulldozer, only to find well, they are crap. I shall be getting some IB goodness in my next PC.

    • HighTech4US2
    • 8 years ago

    Fixed the speculative quotes:

    [quote<]the latest whispers from the rumor mill suggest Ivy Bridge might not become commercially available until next spring.[/quote<] > Intel also stated that Ivy bridge is on target for a late Q4 "qualification for sale", which means that Intel will be begin shipping final products to its customers in the second half of the quarter. [url<]http://www.fudzilla.com/processors/item/24537-ivy-bridge-reading-for-christmas[/url<] Early Q1-2012 is much earlier than "next spring' [quote<] That ought to give AMD a little time to sort out its 32-nm manufacturing problems with GlobalFoundries,[/quote<] You would hope so. [quote<]and maybe crank out faster, cheaper FX-series chips that put up more of a fight.[/quote<] Now to use the quote from this article: Now, don't hold your breath

      • travbrad
      • 8 years ago

      [quote<]Early Q1-2012 is much earlier than "next spring'[/quote<] That same article also says "According to Otellini, first Ivy Bridge systems should become available in Spring 2012."

        • HighTech4US2
        • 8 years ago

        Systems usually means pre-built boxes from Dell, HP and others. OEMs take a little longer to get them out vs just buying the retail CPU.

        I do my own build and will pick up a retail boxed Ivy Bridge as soon at they hit the market which will be before spring.

    • chuckula
    • 8 years ago

    Both of the above stories from Xbitlabs and Fudzilla are likely accurate. It’s been rumored for some time that IB wouldn’t be available until later March or early April. Intel going into mass production now just gives the motherboard makers and OEMs plenty of time to validate systems and to make sure there will be plenty of chips available in a variety of SKUs at launch.

    It also means that Intel can technically say it shipped 22nm products in 2011, even though you can only buy one on the black market.

    Just as Anandtech gave a pretty thorough review of Sandy Bridge months before it was officially released, I’d strongly expect to see detailed leaks of fully-functional IB systems long before it goes on sale.

      • HighTech4US2
      • 8 years ago

      [quote<]Both of the above stories from Xbitlabs and Fudzilla are likely accurate. It's been rumored for some time that IB wouldn't be available until later March or early April.[/quote<] That time frame is for the notebook versions the desktop version is much earlier.

    • NeelyCam
    • 8 years ago

    I guess this means there’s no risk of a paper launch… but what saddens me is this:

    [quote<]whispers from the rumor mill suggest Ivy Bridge might not become commercially available until next spring[/quote<] To me it sounds like they could have launched in early January like the past two years, but they are choosing not to because of weakened competition... and if I recall correctly, they made this schedule adjustment way back some six months ago. Did Intel know that BD (or, worse, Trinity) is going to suck back then already?

      • chuckula
      • 8 years ago

      [quote<]Did Intel know that BD (or, worse, Trinity) is going to suck back then already?[/quote<] Bulldozer: Definitely yes. Both AMD and Intel get their hands on engineering samples from the competition and have quite a bit of fore-warning about what is coming down the pike. They make realistic assessments since they aren't relying on third-hand rumors on a website but have concrete results. For example: AMD likely knows exactly what to expect from Sandy Bridge-E and I wouldn't be surprised if they already have Ivy Bridge systems in house for evaluation. Trinity: A lot less likely simply because it has not been in development as long. Bulldozer had a *very* long gestation period going all the way back to mid-2010 with engineering samples. Intel likely has some Trinity chips in house now, but likely didn't when the rumors of IB's delay first surfaced. At the end of the day, it's good to know what the competition is doing, but since the warning you get is typically less than 1 year (Bulldozer being an exception) you really can't do a huge amount about it except trying to execute as well as you can at producing your product with good yields and maybe adjusting prices if necessary. At this point, Intel is not having to react to AMD very much while AMD clearly did the only thing it could with Bulldozer, which was to slash the price. Even though I'm sure Trinity will have a superior GPU to Ivy Bridge, Intel will still have enough of a market to sell as many IBs as it wants. The good news for AMD is that they will also be able to sell Trinity quite easily (assuming GloFo doesn't pull a Llano2 that is....)

      • smilingcrow
      • 8 years ago

      It seems unusually early to start volume production in Q3 for a chip that is allegedly not due to be released until Q2! That’s about 6 months of production!

        • Flying Fox
        • 8 years ago

        That’s just SOP for Intel. There are a few things to think about:
        1. There is some lag time between making the wafers and producing the dies in the fabs, and then they ship the dies to the packaging plants somewhere else in the world. You need to accumulate certain amount of dies first before sending them off to packaging.
        2. Then there is the inventory build up. It is not like they only wait for a few thousand chips fabbed and packaged before shipping to every OEM and retailers in the world. When it comes to Intel, we are talking hundreds of thousands, and may be millions before they are ready to ship out. So it takes some time as not all fabs are on the leading manufacturing process.
        3. Also it is likely that they are starting small, with just one production line doing very slowly (a few tens/hundreds wafers a day) to further test out the mass production process.

        That explains why Intel usually starts mass production a few months of actual launch. It is very different from say, GPUs where they are much softer launches.

      • OneArmedScissor
      • 8 years ago

      Sandy Bridge didn’t really take off until about April, which pushed things like Pentiums, Celerons, and ULV laptops back even further. Sandy Bridge E/EN/EX still isn’t here. They’re not exactly right on schedule.

        • chuckula
        • 8 years ago

        You know… the way you down-talk anything and everything that Intel does, especially when there’s no justification, you’d think that Intel was the company that came out with a major new micro-architecture over a year late and then had the microarchitecture fail to measure up to what it had already been selling for the past 4 years….

          • derFunkenstein
          • 8 years ago

          Well for a couple months you really did have a hard time getting any sort of SB motherboard because of the SATA bug. Why not give SB a while to mature, especially while they wait for the competition to at least look like they’re considering thinking about catching up someday down the road sometime.

    • Arclight
    • 8 years ago

    Surely AMD will be competitive by the time IB launches “shifty eyes”

      • HighTech4US2
      • 8 years ago

      Now to use the quote from this article: Now, don’t hold your breath

      • ybf
      • 8 years ago

      Bulldozer has put AMD on the mat.

      AMD will be getting up off the mat just in time for Intel to stop windmilling its arm.

      Ivy Bridge will meet AMD’s jaw like a fast-moving, hard thing hitting a woozy, breakable thing.

    • MrDigi
    • 8 years ago

    Is it possible the first IB volume is going to Apple for their product refresh and well ahead of the PC vendors?

      • Hrunga Zmuda
      • 8 years ago

      With reports a new MacBook Pro is coming next month, that very well could be the case. I sure hope it’s going to have USB 3.0, since it’s native to the chipset.

      • JMccovery
      • 8 years ago

      Didn’t Apple just recently update most of, or all of their lines to Sandy Bridge chips?

      • Farting Bob
      • 8 years ago

      I dont see it happening. They recently (by apple standards) upgraded their systems to SB flavours, and they tend to only update every now and then. They will be watching closely though and if IB seems such a big step up for them they might try and update some of their systems earlier than usual for apple.

    • UberGerbil
    • 8 years ago

    Tick-tock is relentless.

    For quite a while now right here on TR Asus has been advertising mobos that are PCIe 3.0 “ready” (and thus compatible with Sandy Bridge today while not hobbling Ivy Bridge tomorrow). But as far as I can tell they aren’t yet actually available for sale. Meanwhile, MSI and Gigabyte have their “Gen3” boards out but there are some question as to how functional they’ll actually be in the Ivy Bridge future. IB may be socket-compatible with SB but that’s certainly not all it takes for a clean upgrade path.

      • NeelyCam
      • 8 years ago

      [quote<]IB may be socket-compatible with SB but that's certainly not all it takes for a clean upgrade path.[/quote<] What do you mean by a clean upgrade path..? As far as I can tell, the only CPU-external differences between the platforms are USB3.0 in the chipset and PCIeG3-compatible traces on the board - both features are already included in those boards. To me it sounds like plug'n'play with a simple bios update - the easiest Intel upgrade in many years. I think they saw how much enthusiasts liked easy upgrades, and took a page off AMD's playbook. It's sort of ironic how Intel is now the "upgrader's preferred choice" while AMD is introducing new sockets across the board.

        • smilingcrow
        • 8 years ago

        There’s been a story floating around stating that there are issues with the BIOS updates required to run IB on an x6x chipset. It stated that Intel is unable to update at least some of its own boards due to this issue:

        [url<]http://vr-zone.com/articles/the-upgrade-path-to-ivy-bridge-might-be-blocked-by-changes-to-uefi/13513.html[/url<]

          • UberGerbil
          • 8 years ago

          Yes, this is precisely what I was referring to: [url=http://motherboardnews.com/2011/09/07/oh-snap-msi-lashes-out-at-gigabyte-over-pcie-gen3-claims/<]MSI[/url<]. [url=http://www.overclockers.com/gigabyte-attempts-debunk-msis-gen3-claims/<]Gigabyte[/url<], and now [url=http://promos.asus.com/US/Z68Gen3/index.htm<]Asus[/url<], and have been golden showering press releases on each other claiming their competitors' "Gen3" implementation is flawed (or, as MSI likes to put it, "[url=http://www.guru3d.com/fullimage.php?image=32726<]fake[/url<]") and only they have the One True Forward Compatible Motherboard (I was too rushed / couldn't be bothered to track down the various links this morning). The socket compatibility story is dictated by OEM / motherboard supplier concerns as much as any desire by Intel to cater to "enthusiasts" (Intel's idea of catering to enthusiasts is to give [s<]sucker[/s<] offer them an entirely separate platform at higher prices that mostly serves to beta-test their next Xeon platform, aka 1366 and SB-E). In the Nehalem era, Intel tried to do separate sockets for Lynnfield and Clarkdale (H1 and H2, 1155 and 1156) but their motherboard partners pushed back an we got one socket whether your chip had an IGP in the socket or not. Consequently, when Sandy Bridge rolled around Intel went a different route and achieved the segmentation they wanted through arbitrary chipset differentiation (H6x/P6x/Z6x) rather than different sockets. So with Ivy Bridge -- and its attendant chipsets -- Intel is really borrowing from its own recent playbook, not AMD's (AMD, after all, has been shuffling along with the same basic chipset for ages). Yes, there may be some nominal ability to drop IB CPUs into certain, selected, very late SB-era mobos; but the reality, it appears, is that for various reasons (mostly the mobo mfr's inability to test against IB and PCIe 3 peripherals as well as new features in the IB chipsets) most "enthusiasts" will be buying a new motherboard to go along with their new Intel CPU, just like always.

            • NeelyCam
            • 8 years ago

            I know I said a while ago that I’ve never seen a bad post from you, but to be honest, this rant is borderlining on one. You’re mixing reality with Intel evilness conspiracies here… I checked those links (thanks! – interesting reads). There is a lot of hype, marketing and fud here.

            Based on MSI’s slides, Gigabyte is skimping on functionality a bit. The key issue seems to be the inability to handle a mixture of PCIeG3 and G2 cards simultaneously – a case that’s (probably intentionally) missing from Gigabyte’s slides… according to MSI, Gigabyte PCIe switches will render both links to G2 – probably because of a limitation by the Intel CPU’s PCIe controller (it can’t handle having 8 PCIeG2 and PCIeG3 ports functional at the same time… although this might be a BIOS issue). Actually, MSI is pointing out that the Gigabyte board settled to G1 data rates… again, a BIOS issue (they are particularly giddy pointing out that “latest” F5 bios… maybe it means that a future Gigabyte bios version will fix that). Settling on G1 could also be a severe board-level wire issue, but I don’t think that’s likely..

            MSI is also pointing out that Gigabyte doesn’t meet all the PCIeG3 specs (e.g., capacitor/resistor values). In the strictest sense this is true. Gigabyte will [i<]probably[/i<] work, but it does not meet the specs, which to me as an engineer is a concern. Most importantly, I don't think Gigabyte has done a good job at addressing those concerns... their solution does seem like a hack, while MSI's seem like a solid, follow-the-spec approach. Meanwhile, Asus has taken the high road, not whining about either competitor, but simply stating that they meet all the requirements for full compliance and performance. The little video is cute, but only shows a single PCIeG3 lane - hardly a testament to a full x16 operation. Nonetheless, their website clearly states they support full x16 G3 link... (although so does MSI and Gigabyte.. at least in certain conditions). My take on all this is that MSI is in compliance and did a clean job - their switch chip is more capable than Gigabyte's, and the other components meet the actual PCIeG3 spec. ASUS might be there, but no solid proof of that is presented. Gigabyte, meanwhile, has been proven to be below the spec and lacking full support. MSI would (again) get my money, unless I went with Intel for lower idle power. Oh, and Intel has nothing to do with any of this mess (except maybe not supporting a split G2x8/G3x8 simultaneously). This is all on mobo manufacturers.

            • UberGerbil
            • 8 years ago

            How is this a rant? I was pointing out that the various mobo mfrs are pissing on one another, and the result was a muddy upgrade picture.. I wasn’t suggesting this was Intel’s fault [i<]per se[/i<], merely that the net effect may be that the "drop in" upgrade path for socket 1156 may not be as clean as it could be. It's not Intel's problem, but it might be a problem for Intel's customers. And that's the only point I was trying to make. Until Ivy Bridge (and PCIe 3.0 peripherals) are released, we really don't know. Some of the people who bought motherboards with the hope / anticipation that they'd be able to swap in Ivy for Sandy may be disappointed. That's always the case when you're banking on products being compatible with future standards; until the products exist to be tested against, you never know if some corner of the spec was more open to interpretation than the designers thought. (Conspiracies? Should I ask why you celebrate Intel's victories as if they're your own, or why you are so hyper-sensitive whenever Intel is mentioned in anything but a positive light? Perhaps you're objecting to my characterization of Intel's product segmentation strategies and their highly effective techniques at parting enthusiasts from their money. I don't see why; Intel has made no secret this. Whether it's introducing Nehalem from the high end down to maximize marginal revenue from the "enthusiasts" aka folks who can't wait -- and incidentally getting early adopters of Bloomfield to shake down the Gainestown server silicon -- or permuting all the various *t knobs, they should be justifiably proud of not leaving any money on the table. That they're a relentlessly effective business is hardly controversial. But we can admire their technical acumen and the efficiency with which they induce us to spend our dollars without feeling the need to treat them with excess deference or affection.)

            • smilingcrow
            • 8 years ago

            “Intel’s idea of catering to enthusiasts is to offer them an entirely separate platform at higher prices that mostly serves to beta-test their next Xeon platform”

            Intel are clearly moving towards lowering the TDP for both mainstream desktop and mobile CPUs which would leave out the enthusiast segment; i.e. IB having a max TDP of 77W and their announcement that mobile CPUs will have lower TDPs.
            This makes a lot of sense as computing power for many application types has outstripped what is required and there is a general trend in computing to smaller form factors and more power efficiency; Ultrabooks, Mac Air, tablets, smartphones.
            So better to address the mainstream market with your main platform and to address the much smaller enthusiast niche with a proper workstation class platform.
            AMD have tried using a Server/Workstation class CPU as a desktop part with BD and look how badly that has gone; mediocre performance overall and appalling power efficiency.
            BD is a better workstation CPU than a desktop CPU but the enthusiast market is more of a hybrid desktop/workstation platform where SB-E will surely kick its ass.

            • UberGerbil
            • 8 years ago

            No argument. The Bloomsfield experiment was highly successful, and clearly Intel plans to replicate it. But I doubt Intel would bother with a completely separate enthusiast / workstation platform if it wasn’t piggybacked on their server products. Lucky for them, that deep-pocketed niche exists and they can repackage their 1S Xeon to fill it — and in the bargain get a nice shakedown of the silicon long before the enterprise-level validation is finished for the real server versions. But the last time around, the high-end platform came out first and tempted all the enthusiasts to jump on it rather than wait and see if the more mainstream parts could satisfy them. That’s not the case this time: the itchy fingers have already pulled the trigger on a Sandy Bridge. There may be some Westmere owners who can’t bring themselves to step “down” to the mainstream platform, of course, but how many?

            Frankly, I think SB-E is going to be overkill for much of the enthusiast market. Professional workstations are another story, though these days even corporate workstation users don’t necessarily need to have all the horsepower sitting on their desk when they can tap into computing resources elsewhere in their org — though the high memory ceiling could be a big win for certain users. But to the extent that “enthusiast” is a synonym for “deep-pocketed gamer” we’ve kind of already reached the point of diminishing returns: between console ports and GPU bottlenecks, how much could additional memory channels and more cores really help? There’s bragging rights, I guess, but I suspect the “enthusiast” market for the most part is satisfied with seeing how high they can overclock a 2x00K.

          • NeelyCam
          • 8 years ago

          This is strange… doesn’t make sense to me. I did a quick google search on this; it seems that VR-Zone is the only source for this, and various forums say it’s BS.

          In my case, it doesn’t really matter, as I already have an Intel H67 system, but I guess this could deter people from buying an SB system..

Pin It on Pinterest

Share This