How Does Intel’s Mobile Chipset Stack Up Against ARM?

If you own a smartphone, there’s a very great chance that it uses a chipset designed by ARM Holdings. Its architecture is uniquely tailored to handle the compact nature of mobile devices – while still providing a large amount of processing power. For years, this has been the standard, and everyone was happy with the results. Conventionally, ARM designed mobile processors while Intel was busy designing and building the latest desktop and laptop chips. However, the immense explosion of mobile devices on the market forced Intel to rethink its strategy and expand its horizons, creating its own mobile chipset. Will the veteran desktop hardware manufacturer actually be able to beat ARM at its own game?

In order to dive deeper into this problem, we must first ask ourselves what helped ARM solidify its place in the mobile world. Answer this question, and you’ll realize what Intel has to compete with in the first place.

First of all, ARM was always chiefly focused on making processors that require very little power to operate. The GHz/watt ratio of an ARM processor far surpasses anything you can buy from Intel for your desktop. Also, ARM’s business model is quite different from Intel’s. It makes no processors of its own, but designs them and retains the intellectual property rights for each one it makes. Companies like Qualcomm have purchased rights to use ARM’s intellectual property and later manufactured processors like the Snapdragon series. When people talk about processors having ARM architecture, they’re saying that these CPUs literally are derivatives of ARM’s designs.

Right now, ARM has it easy. Smartphone users aren’t doing things that require the immense amount of power that a desktop needs to draw. Many of the apps are simply scaled-down minimalistic versions of their PC counterparts. However, the market is quickly evolving to outgrow the need for simple applications.

armintel-transistors

There’s a new demand appearing on the horizon for more intelligent high-performance processors that operate with a 64-bit register. Intel has been known to deliver exactly this sort of thing. Because of its experience in high-performance computing, coupled with recent pushes the company has made to drive down the amount of power consumption in their newer chips, Intel may have a chance to compete with ARM on this battlefield.

Until now, ARM has been designing 32-bit processors that performed just fine on mid-range and high-end devices. That’s going to be old news at some point, though. It will either need to step up its game or be left behind in favor of Intel’s new 14 nm Airmont-architecture processors. Compare that to ARM’s architecture, whose smallest transistors measure 28 nm.

If you’re confused about these numbers, let me explain them: Each transistor in a CPU helps it perform tasks and process data. A transistor is measured in nanometers, and the less one measures, the more of them you can pack into a CPU. In a given space, Intel will be able to pack twice as many transistors into its CPUs than ARM, meaning that it will be able to deliver twice the processing power and performance in a processor of the same size.

armintel-failing

Not necessarily. Even if Intel produces the miracle processor of the millenium, ARM will still have a place in the market for low- to mid-range devices. Also, the company is already making significant strides in getting a 64-bit processor out there. Its first one, the Cortex A53, isn’t a chip to boast about, but it does put ARM on the map as far as 64-bit chips are concerned. The processing power delivered by the A53 is more comparable with the Cortex A9 than any of its later and better-performing chips, so you’ll only feel a performance boost when using applications that require a 64-bit architecture.

This all translates to the fact that ARM might lose its grip to Intel, but it won’t be obliterated. Its focus on delivering low-cost computing will be a saving grace in the long run.

Is Intel going to be the next near-monopoly in high-end smartphone and tablet SoCs? Or is ARM going to find a way to wiggle out of the predicament it faces? Let us know your thoughts in a comment below!

8 comments

  1. Intel has a habit of missing the boat as markets change. They actually owned Marvell, a company that built processors on ARM designs, and sold it because licensing from ARM Holdings hurt their corporate pride. And we all know what cometh before a fall.

    This piece was obviously written at the behest of Intel. ARM has an ARMy of companies working on denser chip technology – one or more of them probably right now has something at <28 nm ready for release.

    wk

    • There’s just too many opportunities for Intel to slip in through the cracks for it to not get some traction in the mobile market. I understand that it can lose its grip just as easily as it got it, since until recently, it has been producing desktop chips. Its lack of experience in the mobile world, however, shouldn’t be a token of impending failure. Intel chips will appear on mobile phones. The packaging, however, will matter more than anything. If it can’t get at least as much attention and coverage as Qualcomm’s chips, it won’t have even a smidgen of a chance of breaking into the mobile market. What I mean is that phone manufacturers have to pay attention to Intel if Intel wants any chance of success. What could save ARM from competition is its already-established loyal customer base.

  2. 1.) Carelessly written article. 14nm squared is FOUR times the density of 28nm squared.
    2.) Intel is contracted to fab Altera’s ARM 64 chips. http://www.eetimes.com/document.asp?doc_id=1319919‎
    http://www.forbes.com/sites/…/exclusive-intel-opens-fabs-to-arm-chips
    3.) ARM Holding website indicates that current ARM 64 is still a 32-bit architecture as in the instruction set still only operates on 32-bit data, however internal registers can be 64-bits (like the Pentiums of the 1990’s) and Addressing can be 64-bit (but are there even any 3GB (32bit addressable) ARM Tablets even in development? Re: low power, RAM in an Intel desktop/laptop uses less power than the CPU, opposite of ARM.
    4.) “Mobile Devices” originally had 4-bit CPUs to control the complex radio used in wireless protocols. The 8 then 16-bit CPUs. A simple (but wasteful) 32-bit CPU could do most of what a complex 8-bit design. ARM adapted by dropping 32bit instructions to operate in 16 bit “Thumb” mode (now ARM Coretex CPUs). At each step in the evolution the ARM was “good enough”.
    5.) The term ARM is as generic as Linux. Linux is a kernel that noone sees. ARM is a base CPU surounded by a SOC (System on a Chip). It isn’t smart, it has hardware drones do what many CPUs do in software. Example Tegra, with a graphics hardware accellerator 20x-50x more powerful than the “dual core” ARM it is attached to.. (q.v. HSA).
    6.) PDA/Smartphone competitors of ARM (Renasas, MIPS, QUICC, …) hit 64bit a decade ago. Cost/licensing and die size = cost let ARM lead. Again, it was “good enough”.
    7.) Did I mention cheap? Intel’s Galileo Development kit has high (desktop CPU performance) but $150 “at cost”, but one can buy a ARM Raspberry Pi for $45. Engadget teardown notes the parts in a $699 iPhone cost $54 and $74 for an $899 iPad. Chinese “house brand tablets are $25-$30 manufactured cost.
    8.) I had a superior non-Intel non-ARM tablet in 2007, a closeout because (remember the Newton) because People didn’t buy tablets. Then Apple built a successful tablet on ARM (
    9.) (for the silly Marvell note) Intel acquired the ARM rights in a patent fight with Digital Equipment (DEC, absorbed into Compaq, which was absorbed into HP). Did I mention that Apple owned 35% of ARM Holdings?

    • 1) My apologies.
      2) Samsung made/still makes (not sure which one is more accurate) panels for Apple. They sue the living daylights out of each other and no one can say they’re not competing. It’s just that sometimes being “frenemies” is a better arrangement than either sticking to friendship or enmity.

      3) I do consider the push to 64 to be a bit premature and silly, if you’d ask me. However, people will still demand it and they’ll soon realize it’s superfluous for the moment. They won’t care. If Intel comes up with powerful 64-bit chips that get cheap enough for the mid-high range of phones, you’ll definitely see it putting some pressure on ARM Holdings and the like.

      4) I cannot possibly contradict the fact that ARM was good enough. In fact, it’s managed to exceed expectations in many of its implementations. However, we can’t naively ignore that Intel is going to pose a threat to the company in the long run.

      5) Correct. Most of the awesome smartphone chips are *based* on ARM Holdings’ designs (I think I’ve mentioned that). They add to those designs significantly, creating chips like the Qualcomm Snapdragon 808, currently one of the most powerful little pieces of awesomeness you can have in your phone. What I *am* talking about, though, is “xeroxed” versions of ARM’s chips, which appear in low- to mid-range devices already (phones that advertise having Cortex chips and nothing more). One of my phones, which is a low-ish phone, has a Cortex A9. I actually quite like it, since it does whatever I actually want it to do.

      6) No denying that. I wouldn’t say that 64-bit was fully understood a decade ago, though. I don’t think that consumers fully understand it even now. They just see another extra that will lead them to buy a “better” phone.

      7) Yes, and Intel’s middle-range performance CPUs are pretty cheap by my book. They probably have either the same performance or higher performance than anything else I can buy at the same price. I’ve seen my Intel CPU, which was big news at the time, go down in price from $900 the day I bought it (and it was *on sale*) to around $200, and it’s still a beast. Intel’s most expensive desktop CPU two years ago cost in the $1000 range. Today, we’re looking at around $500 for even more processing power. The pressure is on Intel to make sure it starts from the get-go with a low price. This might mean that the company will have to subsidize its chips, like another reader mentioned here.

      8) As much as I love ARM (just as I’m sure you do), I don’t think Apple was successful *because* it built a tablet on ARM. It was, however, very successful because of how it marketed its tablet and how the software was packaged. It was a very harmonious device and I think that Apple deserves a lot of the credit for having the stones to push something that Microsoft and many other companies failed to sell.

      9) And round and round it goes. Companies always do things like this. They’ll always step on each other to get ahead of their competition. It’s a dirty game, but I’m concerned with the innovation, not the drama, behind the chips that we use.

  3. A53? Haven’ t you heard of the A57, released at the same time?
    “Compare that to ARM’s architecture, whose smallest transistors measure 28 nm.”
    The density is nothing to do with the architecture. It’s entirely up to the merchant fabbers TSMC and Global Foundries, and Samsung, which is both a merchant and in-house chipmaker. They are running about a year behind Intel on the shift to 16nm. TSMC is already in volume production at 20nm.

    Is a one-year process advantage enough to get Intel a real SOC performance advantage? You also need to allow for the fact that Intel is poor at at least one crucial SOC component independent of the processor, the modem.

    The article doesn’t mention Intel’s biggest disadvantage: cost. ARM processors have historically been much cheaper than Intel’s – Apple reportedly buys its cutting-edge SOCs from Samsung for under $25. Intel has been slashing prices to buy market share in mobile. As a result it’s been hemorrhaging money.

    Intel’s fab leadership reflects its history of fat monopoly profits in the static desktop/ laptop markets and in faster-growing servers, where ARM is also attacking. At some point Intel, facing eroding margins, will have to cut back on fab investment, and its lead over the merchants will fade.

  4. I tjink the one thing that will make intel succeed is that it is k own for it reliability speed and also being abme tk out perform its competetors. Amd use to be a chalanger for intel can thread better than any other cpu on the market and thats what makes it a choice above the rest expecially for computer gamers and lets not forget that they build a very good low end cpu called the atom which does not perform to bad either so once they get there foot in the mobile market they could possibly just take over

Comments are closed.

Sponsored Stories