Report: Clarkdale to ship in the fourth quarter

Will you be able to build a shiny new PC with a 32nm Clarkdale processor this year? Signs point to yes, according to DigiTimes. The site quotes sources at motherboard makers who say Intel has pulled Clarkdale’s mass-production schedule forward.

Reportedly, Clarkdale will account for a considerable 10% of Intel’s shipments to PC makers in the fourth quarter. For reference, DigiTimes says Lynnfield will only make up 2% of shipments, while Core 2 Duo processors will represent 35%. That means we may see plenty of affordable, 32nm dual-core processors with integrated graphics out in the wild before Christmas.

Mind you, Intel already dropped the biggest bombshell four months ago, when it said Clarkdale and Arrandale would hit production late this year. However, the company has stayed mum on its exact launch schedule for the two processors, suggesting they could slip into 2010.

Clarkdale should use the same LGA1156 socket as upcoming quad-core Lynnfield CPUs. However, we heard earlier this month at Computex that it might not work in the same boards, since Clarkdale mobos will have graphics ports hooked up to the CPU socket.

Comments closed
    • UberGerbil
    • 13 years ago

    I have heard rumors AMD actually prices their IGP chipsets at the same price, or sometimes cheaper, than their non-IGP chipsets. Why? To be able to claim graphics marketshare growth against nVidia and Intel.

    • tygrus
    • 13 years ago

    Another Intel CPU another series of chipsets/mainboards. Easy come, easy go. More sales, more dough.

    • wabbit
    • 13 years ago

    q[

    • Flying Fox
    • 13 years ago

    q[

    • wabbit
    • 13 years ago

    I see your point, but look at it this way: Most integrated motherboards (780G or G45 for example) have equal or nearly equal prices to their ‘disintegrated’ counterparts. Sure, there is still a matter of adding a discreete GPU to the mix if it isn’t on board but people are paying for IGP and a separate GPU in the case of (very good) 790GX chipset and no one is complaining. Furthermore, IGP is a nice feature to have if a discreete GPU fails for some reason, like that 4870 we all heard about here on TR. I would gladly pay more for the comfort of knowing that I would be able to use my computer in the meantime. We pay for dual BIOSes for that same reason, I don’t see why would anyone complain about 10 or 20 $ more a few traces would add to the street price of a motherboard. And it is a new way to advertise. I haven’t seen anyone say their motherboards offer graphical redundancy or something like that 🙂

    • UberGerbil
    • 13 years ago

    In theory, yeah, that makes sense. But eschewing FDI (and the traces to take the video out to the backplane) might allow motherboards to be made with fewer layers, making them cheaper to produce, and of course they wouldn’t need the extra video connectors (whether on a dongle or permanently attached) which is an expense also. That could be enough to offset the benefits of a common socket inventory. Or maybe not. I guess we’ll know soon enough how those trade-offs added up.

    • ssidbroadcast
    • 13 years ago

    flip-mode is on point. I’m getting pretty confused about intel’s socket roadmap.

    Also: Long live socket 754!! : )

    Edit: WOW a reply fail?? I swear to GOD ALMIGHTY that I /[

    • wabbit
    • 13 years ago

    So, why not use just one Socket1156 with FDI on every board? That way a CPU with IGP in it could use FDI and a version without FDI can just ignore it. As for the wasted IO panel space, I’m thinking dongles (like Gigabyte did eSATA). Then, move Socket1366 to server platform only. Nobody needs YXZ GB/s of memory transfers on desktop except for bragging rights. Servers can use it rather well on the other hand.
    I really like what AMD is doing with it’s sockets at the moment. Intel’s way is just confusing…

    • MadManOriginal
    • 13 years ago

    Well Intel had a chance to do what you said, and really it’s pretty close – LGA1156 for consumer and LGA1336 for servers. (Yes, LGA1336 is for servers even if there are ‘consumer’ boards that use it.) The lingering LGA775 is largely what mucks it up, they shouldn’t rename those to ‘Core i3’ because it might give people the false impression that they could upgrade to higher ‘Core i_’ CPUs but it looks like Intel is unwilling to use quad core dies all the way down the line and will wait until 32nm at least so they have a dual core die for the lower end. That LGA1156 is split in to chipsets for IGP and non-IGP CPUs is unfortuate, it seems like they ought to be able to make one chipset and socket for both and leave the feature up to motherboard makers. Maybe that’s how it actually is and I’m confused or lacking info.

    Here’s a chart I saw someone link: §[<http://img229.imageshack.us/img229/3016/corechart.jpg<]§ which actually shows it may not be a chipset issue but a motherboard one, except it's got LGA1155 and LGA1156 for Lynnfield :shrug: So it's not LGA1336 and LGA1156 split that's bad in general, it's the LGA775 'Core i_' branding that brings confusion for non-technical people and the possible LGA1155/1156 non-IGP/IGP CPU split that's bad for technical people.

    • UberGerbil
    • 13 years ago

    Intel never intended 1366 to be a mainstream socket. It’s purely a workstation/1P server socket, and might not even have existed if it didn’t offer a convenient way to get paid field tests of the much more important Gainestown die and platform.

    Having multiple sockets really doesn’t matter a whit to consumers. They upgrade their CPU along with the motherboard and the case and everything else. Only the enthusiasts make a big deal about it; the other 99 out of one hundred don’t know about CPU sockets and don’t care about it if you try to explain it.

    The OEMs, on the other hand, are a constituency that cares about sockets, and one that Intel has to listen to at least to some degree. But they’re focused on costs most of all, and if a new socket offers lower costs (as Clarkdale with its simplified motherboards should) that’s all they need to hear.

    • ClickClick5
    • 13 years ago

    l[

    • eitje
    • 13 years ago

    Maybe he meant 754.

    • JoHowdy123
    • 13 years ago

    you mean 775?

    • UberGerbil
    • 13 years ago

    There was a point when Intel was providing testing tools to its board partners for Sockets H1 and H2. H1 was identified as 1156 (ie Lynnfield), while H2 was identified as 1155 (Clarkdale). It wasn’t clear how interchangable the two processors would be with the two sockets — obviously a Clarkdale couldn’t output video on a motherboard without the FDI (Flexible Display Interface) link even if the same PCH chip was used for both platforms, but would it work at all? It may be that the extra pin was just there for product segmentation purposes and Intel has backed off on that. (It’s telling that both the mobile version of Lynnfield [Clarksfield] and the mobile version of Clarkdale [Arrandale] were always rumored to use the same mPGA-989 socket)

    Nevertheless you still have to distinguish between motherboards that have the FDI and related pieces and motherboards that don’t, at least as far as Clarkdale buyers are concerned, whether that’s through socket naming or some other means. Presumably there won’t be many folks trying to go from a Lynnfield quad to a Clarkdale dual (even if that takes you from 45nm to 32nm, unless that smaller node offers something pretty magical). There might be more people trying to go from a Clarkdale IGP system to a quad plus discrete GPU, particularly if a 32nm quad Lynnfield successor becomes available (Intel has said that won’t happen until the Sandy Bridge era, but I wouldn’t be surprised if that changes). But some clarity on this whole interoperability question would certainly be helpful. Perhaps the whole “CPU+mobo as a unit” thinking is the best way to go. At least they’re all DDR3….

    (This schematic references the now-cancelled Havendale/Auburndale, but Clarkdale/Arrandale are expected to use the same arrangement, just with a 32nm CPU die in the MCM. The key things are the FDI link from the CPU to the PCH, the display controller unit in the PCH, and the output circuitry to the actual video outputs. The display controller may be in every PCH but dormant in Lynnfield systems; the other circuitry presumably isn’t present at all in motherboards aimed at Lynnfield)
    §[<http://pc.watch.impress.co.jp/docs/2008/0922/kaigai_3l.gif<]§

    • Hattig
    • 13 years ago

    ARGH!

    Will we have Core i3 1156A, Core i3 1156B, Core i5 1156A, Core i5 1156B, Core i7 1156A, Core i7 1366 (I forget the exact number), Core i9 1366, …

    • flip-mode
    • 13 years ago

    The socket situation on the Intel side is getting pretty messy. I don’t see 755 going away any time soon due to low end CPU continuing to ride that train, so that means four sockets for the next little while.

    • wabbit
    • 13 years ago

    If that is the case, then we will have three sockets from Intel… It will be worse than Slot1/Socket370 situation….

    • ryko
    • 13 years ago

    yeah, from what i understand there will be 1156-a and 1156-b sockets. one will have a video-out port on the rear cluster, the other will not. theoretically, you should be able to use any clarkdale cpu in either mobo, but you will obviously need a discrete video card if you choose the one without video-out.

    at least let’s hope this is the case!

Pin It on Pinterest

Share This

Share this post with your friends!