Ivy Bridge sample sneaks out, is benchmarked

Intel’s 22-nm Ivy Bridge processors aren’t due out for a while still—knowing the chipmaker, we’ll probably see them pop up on retail shelves early next year—but the silicon is clearly functional already. We got a taste when Intel demoed an Ivy Bridge laptop two months ago, and we’ve now gotten another teaser courtesy of famed overclocker Coolaler.

Posting on his eponymous forum, Coolaler sated the curiosity of followers with photos and benchmarks of a purported Ivy Bridge engineering sample. His pictures show a fairly innocuous-looking LGA1155 processor package with "Intel confidential" and "ES 1.80GHz" emblazoned on the heatspreader. At the back, the capacitor and resistor arrangement resembles that of today’s dual-core Sandy Bridge CPUs. It’s not quite identical, though.

Coolaler’s BIOS and CPU-Z screenshots reveal that the CPU has two cores, four threads, 512KB of L2 cache, and 4MB of L3 cache. The images also confirm the 1.8GHz speed, which is presumably a ways below what we’ll see on the finished product. We are, after all, talking about one of the first CPUs based on Intel’s brand-new 22-nm fab process—or so it appears.

The benchmarks are somewhat anticlimactic, although they do give us a couple of noteworthy scores: 22.198 seconds for a SuperPI 1M run, and 1.81 points in a Cinebnch 11.5 test. Coolaler got higher numbers from a Core i5-2400 underclocked to 1.8GHz, but that chip has two extra cores and a couple megs of extra cache. (Thanks to VR-Zone for the link.)

Comments closed
    • DarkUltra
    • 8 years ago

    Aren’t you guys forgetting something? Overclocking! IB should have this new 3d tri-gate manufacturing. This will reduce the tdp and increase oc further.

    • ronch
    • 8 years ago

    The way things are going this is what Bulldozer will eventually compete with during its early iterations. I’m planning to get Bulldozer, but I don’t think I’d hesitate to pick this baby up, given Intel’s track record lately, especially if BD ends up giving us more mores but weak Single-Threaded performance just like Phenom II X6 does.

    • link626
    • 8 years ago

    if there are no major changes, maybe it will give amd time to catch up. maybe.

    • DavidC1
    • 8 years ago

    “Tick” chips aren’t always about performance increases, though that is usually bundled with rest of changes. Tick addresses deficiencies that Tock, new architecture had. If they were lacking some SKUs with the Tock chip, or the platform was a bit immature, they get that fixed. In other words, they round out the family.

    You’ll probably find out its similar for future architectures like Haswell. You might find that the features they promised isn’t there, or there isn’t a SKU that you want. The platform might turn out to be little unstable too. Which will be addressed with the shrink.

    • flip-mode
    • 8 years ago

    This is a “tick”, I think, so more about introducing a new process node than about architectural changes.

      • Game_boy
      • 8 years ago

      Intel says it’s a “tick plus”

      [url<]http://www.anandtech.com/show/4378/ivy-bridge-a-tick-with-configurable-tdp[/url<] You might as well throw out tick-tock for how much they have screwed with the naming, timing, and Penryn having IPC increases but Westmere not and there being about four generations of Second Generation Core (Core Duo, Core 2, Nehalem, Core 2010, 2nd gen Core)

        • [TR]
        • 8 years ago

        Now we just need Intel, AMD and nVidia to start naming their current products with the same numbers (i.e. 5xxx, 6xxx and so on) and you’ll be making PCs with GPUs for CPUs and vice-versa.
        “I just put together an nVidia 5387 machine with an AMD 5422 and 2x Intel 5199 SLI. I’m having some trouble with drivers, but it’s great!!”

          • swaaye
          • 8 years ago

          I’m all for it. I want my AMD 5422 already.

    • Arclight
    • 8 years ago

    [quote<]The benchmarks are somewhat anticlimactic, although they do give us a couple of noteworthy scores: 22.198 seconds for a SuperPI 1M run, and 1.81 points in a Cinebnch 11.5 test. Coolaler got higher numbers from a Core i5-2400 underclocked to 1.8GHz, but that chip has two extra cores and a couple megs of extra cache. (Thanks to VR-Zone for the link.)[/quote<] So.....arhitecture wise is it better....or? I mean we have dual core SB CPU, how does it stack against that?

      • Faiakes
      • 8 years ago

      It’s all about battery life for me. Was this a ULV range chip?

      • jorjxmckie
      • 8 years ago

      Ivy Bridge is just a process shrink of the existing Sandy Bridge architecture, don’t expect leaps and bounds of performance improvement. Most likely there will be some significant efficiency gains and modest performance improvements through minor architecture tweaks.

        • ImSpartacus
        • 8 years ago

        Yup, this is a battery life update. Intel is shrinking its process and debuting Ultrabooks. I expect to see some killer ULV products.

        • sweatshopking
        • 8 years ago

        and a major on die GPU refresh. this chip is the llano gpu competitor. dx11, supposedly MUCH faster, etc.

          • Goty
          • 8 years ago

          Oops, I just Trinity’d your Ivy Bridge!

          =P

            • NeelyCam
            • 8 years ago

            Maybe 12 months later.

            • DavidC1
            • 8 years ago

            Trinity is on the same 32nm process, and GPU is already taking a truckload of space on the die. I doubt it’ll bring anything significant on graphics. It’s the CPU that will be the bigger update. Llano is already 228mm2, and they are planning a big change! Imagine their financials with a 300mm2+ “mainstream value” chip that costs less than $150.

          • khands
          • 8 years ago

          Yeah it’s going to be close to Llano, though I don’t think they’ll have fixed their image quality issues. Trinity should be out shortly afterwards though.

        • swaaye
        • 8 years ago

        Well you know that their manufacturing edge is a huge competitive advantage not to be understated.

        The primary benefit to Intel will probably be lower costs which combined with their higher pricing and uber volume is a real win for them. For us the power consumption reduction and likely architecture improvements will be cool.

        From a power consumption standpoint, I see subnote CPU options much better than Llano and with similar GPU performance. I’m not sure that I’m in the market for them though.

        • UberGerbil
        • 8 years ago

        Well, IB is supposed to round out the AVX instruction subset with FMA. That’s not going to matter to many non-developers (yet), and even then it won’t matter as much as it should thanks to the FMA3/FMA4 Intel/AMD custerfluck, but it does make IB more than “just a process shrink.”: Notwithstanding that, the rest of your comment stands up just fine.

          • DavidC1
          • 8 years ago

          FMA=Haswell!

            • UberGerbil
            • 8 years ago

            Intel has been very cagey about this. AVX1 = Sandy Bridge. AVX2 = Haswell. FMA has its own cap bit in CPUID, suggesting it is independent of both / either; there used to be references in the AVX reference to “POST-32NM PROCESSOR INSTRUCTIONS” for various things, including the half-float instructions, that were open to interpretation since that applies to both IB and Haswell.

            This all may mean they’re just hedging their bets as far as delivery schedule and/or feature segmentation, or even to see what AMD does with Bulldozer. But AFAIK they haven’t made an absolute declaration either way. I remain hopeful (though probably in vain)

Pin It on Pinterest

Share This