Ivy Bridge mobile CPU specifications exposed

More details have leaked out on Intel’s upcoming Ivy Bridge CPUs. We learned about the desktop side of Sandy Bridge’s successor last week, and now the mobile part of the picture has been filled in. CPU World has summarized the specifications of no fewer than eight notebook-bound Ivy Bridge CPUs.

The first quad-core members of the mobile Ivy Bridge family will reportedly arrive in April. Three models will supposedly be available at that time, including a Core i7-3920XM with a 3.8GHz Turbo peak and a 55W thermal envelope. Obviously, that ain’t going in an ultrabook. Neither are the other two quad-core parts, which have 45W thermal envelopes and only slightly slower clock speeds. All of the quad-core chips support Hyper-Threading.

Intel’s threading magic will also infiltrate a collection of dual-core Ivy Bridge CPUs due out in May. According to CPU World, three of those chips have 35W thermal envelopes, while two have TDP ratings of just 17W. The fastest 17W chip is the Core i7-3667U, which purportedly boasts a 2GHz base clock speed and a 3.2GHz Turbo peak. For reference, the best Sandy Bridge part with a 17W thermal envelope is the Core i7-2677M, which is clocked at 1.8GHz with a 2.9GHz Turbo ceiling.

The May arrival of the first ultrabook-worthy Ivy Bridge CPUs should ensure that Computex is teeming with razor-thin notebooks. The trade show takes place from June 5-9 in Taipei, Taiwan, and we’ll have someone on the ground to cover all the gory details.

Comments closed
    • Abdulahad
    • 8 years ago

    DISAPPOINTING…. I EXPECTED BETTER!!!

      • forumics
      • 8 years ago

      what were you expecting then? Bulldozer?

        • Abdulahad
        • 8 years ago

        Nopes…. Celeron:-)

    • DavidC1
    • 8 years ago

    VR-Zone has the original article.

    One thing no one seemed to notice:

    GPU on regular parts are clocked 10-15% higher than the desktop ones
    GPU on 17W chips are clocked just like the desktop ones

    Remember they claimed 60% on GPU improvement for Ivy Bridge? That’s desktop comparisons. That means on Ivy Bridge, the graphics on the 17W Ultrabooks will perform 60% better than desktop Sandy Bridge’s GPU.

    Respect to the CPU: Remember how some articles claimed 20% improvement for Ivy Bridge over Sandy Bridge? That’s regarding mobiles. The better in-between Turbo frequency along with 5-6% better IPC will result in that 20% gain.

    • jdaven
    • 8 years ago

    I thought the rumors had the chips spec’ed more like this:

    2 cores 1.6-1.8 GHz 10W
    2 cores 1.8-2.2 GHz 17W
    2 cores 2.2-3.0 GHz 25W
    4 cores 2.4-3.0 GHz 35W

    This list doesn’t look very good.

    • ronch
    • 8 years ago

    AMD really has their work cut out for them to match these numbers. I imagine they could put in just one or two modules at ~2.0GHz to make up a mobile CPU, but it would probably still eat up more power and not have impressive performance. But who knows, maybe there are just some tweaks AMD has to do to make BD fly. As it is, it’s like two Pentium 4’s melded together with a shared front end.

    I really wish AMD the best. I really do.

      • khands
      • 8 years ago

      Yeah, I don’t think Trinity is going to cut it save possibly for low-budget gamers, and that’s not a position AMD wants to be in. Here’s hoping I’m wrong though.

        • forumics
        • 8 years ago

        the move from phenom to phenom 2 yielded a gain of less than 10% but even a 20% improvement to BD wouldn’t save it

    • OneArmedScissor
    • 8 years ago

    Still i7 dual-cores and where are the 35w quad-cores?!?

    FUUUUUUUUUUUUUUUUUUUUUUUU

    • adisor19
    • 8 years ago

    OK I was hoping they would arrive in the January – February time frame. Since the 17W mobile CPUs are scheduled for May, it looks like I’ll just go ahead and get my Mac Book Air now.

    Adi

      • chuckula
      • 8 years ago

      Oh come on and admit it Adi… you’ll just go out and buy the Sandy Bridge Air now, and then go out and buy the Ivy Bridge Air next year because the old Airs won’t be shiny anymore. You need to do your part to get Apple’s earnings back on track!

        • adisor19
        • 8 years ago

        If only I was that rich, I’d probably do it..

        Adi

        • crazybus
        • 8 years ago

        You could probably flip the 2011 MBA and not be out a lot of change, given their decent resale value.

          • adisor19
          • 8 years ago

          I’d still be loosing a decent amount of $ so it’s not worth it for me. I stuck with my MBP 2006 for 4+ years and i plan to do the same with the MBA.

          Adi

      • NeelyCam
      • 8 years ago

      So, you had your MBP for 4+ years, but you can’t wait three extra months to get IB-based MBA?

    • tone21705
    • 8 years ago

    I am looking at buying a laptop and was going to get a quad core Sandy Bridge (maybe the new HP ENVY coming out in the next few days), do you think waiting till ~April for Ivy Bridge is worth while?

    I currently have a 5-6 year old laptop that is slowing dying (internal wireless failed, among other things) and will probably use this one for just as long.

    If it was just the ~100mhz increase I would say forget it but the graphics improvements interest me.

    Thoughts?

      • OneArmedScissor
      • 8 years ago

      More important than Ivy Bridge itself is that it will come along with new lines of laptops from every OEM. Nobody can really predict what options that will bring, but I’m guessing that there will be a lot more lightweight laptops (just in general, not necessarily “ultrabooks”) and IPS screens.

      But there’s no real platform or pricing change, so as far as the CPU itself goes, there’s little reason to expect that to have much of an impact. There’s not going to be some paradigm shift towards $400 quad-core laptops or anything of that nature. If you want to play games at 1366×768, then it might mean something, but that’s about it.

    • Vulk
    • 8 years ago

    Umm, isn’t this somewhat disappointing news? I know this is supposed to be a ‘Tick’ or ‘Tock’ or whatever the marketing term is for moving an existing architecture to the new process node, and that I shouldn’t expect MUCH. Still, Intel has been talking it up with 10-15% improvements for IB. With the jump from 32nm to 22nm Intel should have been able to double the transistor count, decrease power draw significantly, or increase frequency by a healthy margin (or most likely some combination there of). What you’re telling me is that in the 17w range they’ve managed to… increase performance by a scant 100mhz base, and 300mhz in turbo (which is some indication of additional thermal headroom somewhere on the table), while doing little or nothing to improve power draw, and… I guess they’re using the rest of the shrink to offset some of the R&D cost and difficulty moving to the new process (i.e. binning, etc).

    That Graphics component better be AWESOME, because otherwise those aren’t good performance or power improvements. Not given the expectations they gave me when they announced the tri-gate process, lithography shrink, and other promises for IB.

    I know we won’t know the full story until we see samples. But I was expecting more low power, yet less crippled overall, CPUs based on their marketing.

    I’m just saying.

      • chuckula
      • 8 years ago

      I think the 17 watt chips from Ivy Bridge will perform substantially better than the 17 watt equivalents from the Sandy Bridge generation. You are looking at max turbo frequencies but forgetting another important part of the equation for using turbo: The proportion of the time that you can actually use turbo while still staying within the power envelope. I wouldn’t be surprised to see these IB chips spending a lot more time with the CPUs up around 3Ghz during heavy operations than the earlier SB chips were able to do.

      Additionally, don’t underestimate the graphics. The 17 watt parts have high-end HD 4000 series parts that have about the same transistor count as Llano’s graphics cores do. This is a BIG departure for Intel where they are finally starting to give serious real estate to the GPU. I’m not expecting these chips to go out and destroy the bigger Trinity chips, but in an ultraportable form factor with 17 Watt TDPs these graphics may be the best you get for the foreseeable future.

      • [+Duracell-]
      • 8 years ago

      This is a ‘tick’, which is just a die shrink of Sandy Bridge. Ticks are not meant for substantial performance improvements. That will come with Haswell, the next ‘tock’.

        • forumics
        • 8 years ago

        it doesn’t really matter actually, now that amd’s bulldozer has been bulldozed every tick feels like a tock and every tock, well feels like yet another tock too

      • Peldor
      • 8 years ago

      The thing to remember is these are TDP buckets. A 15W chip is in the same bucket as a 17W chip, even though the power draw is 12% lower.

      [quote<] Intel has been talking it up with 10-15% improvements for IB.[/quote<] The difference in those (supposed) specs is 200 MHz base, aka 11%, right where they said.

      • WillBach
      • 8 years ago

      I agree with what you’re saying in principle, that Ivy Bridge graphics need to be awesome in order for Ivy Bridge to be a large improvement over Sandy Bridge, but you are missing a few things.
      [list<] [*<]The clock speed improvement is 200 MHz, not 100 MHz, so that's 11% over 1.8 GHz, before IPC gains. The only reason I personally don't consider this to be a large improvement by itself is because Sandy Bridge is already so fast. [/*<][*<]IPC gains should be at least a few percent because of things like using register renaming to have register-to-register moves not enter the execution pipeline. [/*<][*<]Idle power should be substantially reduced, especially because of the lower leakage currents with 22 nm. [/*<][*<]Active power for equivalent workloads should be lower, partly because of IPC gains but mostly because of 22 nm. This could be the golden ticket. If TDP (which represents a maximum) stay the same, but actual power consumption falls, battery life for the same workload increases. [/*<][*<]Miscellaneous things like [s<]programmable TDP[/s<], increased integration, and maybe even lower prices.[/*<] [/list<] Edit: I was wrong about programmable TDP, I don't know where I heard that.

      • dashbarron
      • 8 years ago

      I originally thought along your lines but AnandTech’s article will help put things into perspective for you: [url<]http://www.anandtech.com/show/5174/why-ivy-bridge-is-still-quad-core[/url<] Besides or because of being a "tick" iteration, this processor is meant to refine, reduce power, and increase the integrated GPU which is the weakest point of the processor. Increase CPU power when they're already on top or improve the weakest link? What they are doing seems to make sense when you look at it that way.

    • sweatshopking
    • 8 years ago

    I’d Ike to see a return to core 2 culv power consumption. My wife’s 3810z gets 9 hours, is cool, quiet, and fast enough for almost everything. Focus there Intel!

      • Duck
      • 8 years ago

      My sister has a 5W 1.2GHz Pentium M laptop – fanless!

      I would have thought with all these advances they could bring the graphics on board, add an extra core and still fit it in that 5W thermal envelope. But then again, this laptop was of the era of $2000 ultra-portables.

        • willmore
        • 8 years ago

        Uhh, AMD did.

          • FuturePastNow
          • 8 years ago

          Thumbs up. AMD has demonstrated a 5W version of the C-50 APU.

            • willmore
            • 8 years ago

            Z-01

        • bcronce
        • 8 years ago

        To be fair, one of these dual core HT IB chips is probably 8-16 times faster than that Pentium M while consuming only 3x-4x more peak power. Also, that 17watt TDP includes a DX11 GPU.

        The idle power is probably less with the IB chip.

          • OneArmedScissor
          • 8 years ago

          [quote<]The idle power is probably less with the IB chip.[/quote<] I doubt it, unless they're actually using a separate manufacturing process for the ULV parts, as opposed to undervolting the standard 4 GHz-ish tuned chips. It may be more tightly integrated, but not [i<]that[/i<] much. All those parts that used to be spread out amongst a few more chips are still in there, but now with the addition of large L3 caches - and all designed to be cranked up a few extra GHz, [b<]including the memory controller, busses, caches, and GPU.[/b<] That makes for a leaky chip, and smaller transistors do not necessarily address this. They can make it worse, which is why we [i<]need[/i<] things like high-k gates, and soon, tri-gates. They aren't magic leakage stoppers, just compensators. Unfortunately, in this context, power gating doesn't yet do more than switch one core off while you're actually using the laptop, so that's another "advance" out the window. Intel has done a good job of keeping peak power down for very [b<]fast[/b<] laptops, but they never really made idle power for [b<]"fast enough"[/b<] a priority. Core 2 has yet to even be matched, and for all we know, some older platforms may have worked even better with LED backlit screens and other modernizations.

            • WillBach
            • 8 years ago

            [quote<]I doubt it, unless they're actually using a separate manufacturing process for the ULV parts, as opposed to undervolting the standard 4 GHz-ish tuned chips.[/quote<] They do use a separate manufacturing process. Still 22nm, but tuned for lower power consumption at lower clocks. Same for Sandy Bridge. [quote<]That makes for a leaky chip, and smaller transistors do not necessarily address this. They can make it worse, which is why we need things like high-k gates, and soon, tri-gates.[/quote<] It's my understanding that tri-gate at 22nm has much lower leakage than CULV Core2 at 65nm or 45nm. [quote<]Unfortunately, in this context, power gating doesn't yet do more than switch one core off while you're actually using the laptop, so that's another "advance" out the window.[/quote<] Power gating can turn off every core in Sandy Bridge and Ivy Bridge, and keep them off, and wait for an interrupt, even with the screen on. It's true that Core2 CULV was really, really good. But I don't think it was better at idle, just that it had a lower maximum power draw at the expense of performance.

            • OneArmedScissor
            • 8 years ago

            [quote<]They do use a separate manufacturing process. Still 22nm, but tuned for lower power consumption at lower clocks. Same for Sandy Bridge.[/quote<] They've stated there will be a low, mid, and high power process, but they haven't actually talked about the difference, or which CPUs use which, so that's wait and see if the impact is legitimate. For Sandy Bridge, it was not, and Ivy Bridge pushes the clocks higher. Point being, there's a distinct possibility that the ULV parts, though being the lowest power Ivy Bridge chips, will actually use the mid power process, while the low is reserved for things like Atom SoCs. [quote<]It's my understanding that tri-gate at 22nm has much lower leakage than CULV Core2 at 65nm or 45nm.[/quote<] Probably, but we have to keep everything in context. This isn't X transistor vs. Y transistor, it's an entire platform compared to another. Ivy Bridge has a gajillion transistors, for doing a gajillion things at a gajillion GHz. Core 2s CULV laptops were comparitively much simpler, but older laptops were even simpler yet, despite having an extra controller chip here and there. You use these new CPUs with laptops that have LED backlights, which is a significant power saver, but that didn't end up amounting to much with new vs. old laptops. I don't trust that the new CPUs from either Intel or AMD are really doing much to cut platform power. They have long since maintained the status quo. [quote<]Power gating can turn off every core in Sandy Bridge and Ivy Bridge, and keep them off, and wait for an interrupt, even with the screen on.[/quote<] Can, yes, but when do you actually see it happen? Much more than that [i<]can[/i<] be power gated, but the problem is that it's constantly coming on just to keep your background processes going. I'm sure it will help out a lot in the future, but we're just not there yet. The new ARM SoCs that use a special low power core just for idle processes are the direction things will have to go first. [quote<]It's true that Core2 CULV was really, really good. But I don't think it was better at idle, just that it had a lower maximum power draw at the expense of performance.[/quote<] The 3 GHz-ish Core 2 laptops certainly didn't do as well. The CPU design itself may have had very little to do with it. For any number of reasons, the laptops as a whole just did a better job. I think a lot of it had to do with LED backlights being standardized in the middle of Core 2's run. Everyone had to wait that out to implement a new way to blow the power savings. :p

            • WillBach
            • 8 years ago

            I guess we’ll see. I was wrong about programmable TDP in Ivy Bridge, I don’t know where I heard that. I got mixed up :-/

          • Duck
          • 8 years ago

          That may be true (especially platform idle power), but a 17W TDP is what it is. Over 3x the cooling requirements compared to the 5W Pentium M.

          Better to have performance that is just ‘fast enough’ – making all the gains you can while maintaining the power envelope. It would have been a much better fit for these slim, metal bodied ultrabooks that Intel has been pushing.

            • DavidC1
            • 8 years ago

            It’s not truly 3x, because Pentium M needs to run the ICH and the MCH. And don’t even compare the GPU.

            Atom performs at least as good as the 5W Pentium M back then, and yes that’s including single thread because Pentium M had to be clocked so miserably low. You get Hyperthreading on top of that, which gives you massive gains for this chip.

            • jensend
            • 8 years ago

            I call BS on Atom single-thread performance matching ULV Dothan, unless you’re talking about the 1.83 GHz Dxxx parts which have 10W TDPs and thus could be compared to LV Dothan instead. In both cases- ULV Dothan vs Nxxx and LV Dothan vs Dxxx – Dothan’s single-thread performance is quite significantly better.

            All kinds of reviews all over the web showed that Dothan only needed half the clock speed of a first-gen Atom (Silverthorne/Diamondville) to match it in single-threaded performance. The second generation Atoms gave about a 10% improvement. That’s not anywhere near enough to make Pineview at 1.66 GHz competitive with a 1.3GHz ULV Dothan (or to make 1.83 GHz Pineview competitive with 1.6GHz LV Dothans).

            For multi-threaded tests Atom will probably match equivalent-TDP Dothans, and any hyperthreading-enabled (i.e. non-nerfed, 4-threaded) versions of the upcoming dual-core Cedar View/Cedar Trail will probably beat equivalent-TDP Dothans in the all-important-but-intangible “everyday computing experience” benchmark. But that’s coming 7.5 years and 3 process generations later.

            • DavidC1
            • 8 years ago

            Ah, yes I forgot about the Dothan part. But its true in the case of Banias, and to make a point that you still have an option if you want <5W part.

            [url<]http://www.notebookreview.com/default.asp?newsID=1859[/url<] Dothan also brought tiny amount better battery life than Banias.

    • FuturePastNow
    • 8 years ago

    I note the absence of i3 processors on both the desktop and mobile lists that have leaked.

      • willmore
      • 8 years ago

      Sandy Bridge will stay there and lower for the time being–according to the articles I’ve read.

        • FuturePastNow
        • 8 years ago

        Well, I can’t argue that SB isn’t good enough for the low end, but I’d love to see the power savings and improved graphics of IB trickle down.

        I guess all of the first-run 2C/4T dies have been allocated to mobile chips.

          • willmore
          • 8 years ago

          I’m pretty happy with my B940 based laptop. Then agian, I’d probably just as happy with an A4 based laptop. Decent enough graphics, enough CPU to be useful, and reasonable battery life. Throw in IVB and that battery life’ll go up a bit and the graphics still won’t be good for too much, but that’s the nice thing about ‘good enough’.

Pin It on Pinterest

Share This