Sign in with
Sign up | Sign in

Opinion: Intel is Underestimating ARM

By - Source: Tom's Hardware US | B 51 comments

We have been talking for more than two years about the next great processor war, a confrontation between Intel and ARM.

We haven't seen much more than saber rattling and scatter shooting so far and it appears that both sides are still preparing their first targeted shot at the other side, while prepping their defenses against a likely retaliatory move. Intel leaked some strategic information during the most recent earnings call. Still, I am wondering if Intel is taking the ARM threat seriously enough?

I would be lying if I said that I am not disappointed that a much more dramatic confrontation between ARM and Intel has not taken place yet. It's a battle between two incredibly strong and established ecosystems that will draw blood on both sides and has the potential to be drawn out over years. The benefit goes to the consumer, who will see unprecedented innovation in the product segments Intel and ARM will be fighting over. One of the reasons why this fight has not been very visible yet is because it is unlikely that the companies will be going into battle without a proper plan. Both are guarding their crown jewels and are positioning their strongest assets on the field.

If we were to compare ARM's and Intel's key assets, you'd be dealing with a long list of accomplishments and capabilities, but I believe that the fight will come down to ARM's vendor ecosystem as well as Intel's manufacturing capabilities. The big question on Intel's side has always been, "How will they deal with an armada of ARM vendors, a vast array of tailored processors, which not include powerful vendors such as Qualcomm, Samsung and Texas Instruments, but especially Nvidia (which is just waiting to throw another can of whoopass into Intel's face)?" In the end, ARM has the luxury that its vendors will go to battle, not ARM itself.

On the other side, Intel is one of the very few remaining independent chip manufacturers. It has, by far, the most advanced production process and immense capability to produce and adjust the production of microchips. Based on revenue, Samsung is the only other semiconductor company that may be able to afford $5 billion fabs down the road - and upgrade them every two years to accommodate a new shrink. How will ARM vendors compete with Intel?

We are still waiting for an answer about ARM production capability and flexibility, but Intel has given some answers about its strategy.

CEO Paul Otellini just confirmed once more that there will be cell phone chips from Intel next year. He acknowledged that there are many ARM vendors and his idea is to counter that competition by offering multiple flavors of Intel processors. "Just like there's no one ARM, there's no one version of Atom going into these devices," he told an analyst over the phone. "Intel has tailored Atom for low-end PCs, it's now tailoring it for tablets. We're tailoring different versions of it for handsets and cell phones, and other versions for embedded and automotive implementations."

While this is a clear scenario on how Intel intends to compete initially, the hope is that the cell phone processor will move upmarket and move closer to what Intel is really good at, while it is moving away from the generally lower performance of ARM processors. In other words, Intel believes that achieving higher performance will be more difficult for ARM vendors than lower power consumption for Intel.

"As the need for computing performance goes up, both the Intel architecture and the ARM architectures face the same fundamental physics problems, which is more performance requires more transistors. So at the end of the day, to deliver multi-core performance, better graphics performance in a battery-constrained environment is going to be a function of the transistors more than the micro-architecture," Otellini explained. Translation: Intel's capability to design processors and put them into an advanced production process will decide the battle.

There have also been some notes that AMD may be moving into the ARM market, which could make a lot of sense for the green team. However, it was interesting to hear that Intel believes that it can sell its products for higher prices because of a certain perception value, which includes "more performance," "features, "reliability", and "high quality levels," according to CFO Stacy Smith. In the particular example from the earnings call, Smith mentioned a $20 to $30 price advantage for a processor kit for a low-end target market.

Smith went on to say that those values would allow Intel to "shine" in the ARM market as well. He got carried away a bit, in my personal opinion, when he compared AMD with ARM and described Intel's advantages as the main weapons against the new rival: "It's going to be the same issue with ARM," he said. "All of those advantages will be even more so against ARM. […] As Paul [Otellini] said, at the end, it all comes down to we're solving these problems of physics two years ahead of the rest of the industry, that gives us a cause-to-performance a future advantage that become very difficult for people to match."

The problem, of course, is that some of Intel's advantages only work in the PC market, not in the smartphone and tablet market, where Intel is the new kid on the block. Performance of Intel processors has not been proven in the smartphone/tablet market yet, and reliability has not been proven and features have not been proven either. Then there are issues about power consumption and the right balance between power consumption and performance. Intel is the unknown factor in this game and would be a risk for a smartphone designer, not ARM. Plus, there have not been Intel products that have been especially compelling in this space, even if Intel has been talking about the development of smartphone chips since the release of the first Atom processor more than three years ago. Intel will need a processor to back up its claims to create a foundation for future opportunity in the smartphone/tablet space. I cannot help but say that the current claims are not really convincing.

Even if we assume that an Intel processor is faster than a certain ARM processor from a certain vendor, ARM will remain a moving target with quite a bit of design capability at several large players out there. Intel has tremendous design resources at hand, but it would be naïve to believe that the ARM armada can easily be beaten on the perception of performance, reliability and quality in a different market. I am wondering: Could Intel be underestimating ARM, or were Otellini's and Smith's claims simply part of an evolving marketing phrase?

Discuss
Ask a Category Expert

Create a new thread in the News comments forum about this subject

Example: Notebook, Android, SSD hard drive

This thread is closed for comments
Top Comments
  • 14 Hide
    fellskrazykayaker , October 20, 2011 6:36 PM
    I've been pretty disappointed that Intel doesn't have better low powered CPUs in the mobile sector. It's been good to see ARM chips give Intel some much needed motivation.

    I've been pretty pleased with how ARM is doing in the mobile sector. It's forcing Microsoft and other app developers to properly write code for ARM starting from the ground up rather than trying to retrofit existing x86 code onto Intel's low power offerings (which have been pretty poor so far).
  • 10 Hide
    digiex , October 20, 2011 7:59 PM
    CISC vs RISC all over again.
Other Comments
    Display all 51 comments.
  • 14 Hide
    fellskrazykayaker , October 20, 2011 6:36 PM
    I've been pretty disappointed that Intel doesn't have better low powered CPUs in the mobile sector. It's been good to see ARM chips give Intel some much needed motivation.

    I've been pretty pleased with how ARM is doing in the mobile sector. It's forcing Microsoft and other app developers to properly write code for ARM starting from the ground up rather than trying to retrofit existing x86 code onto Intel's low power offerings (which have been pretty poor so far).
  • 6 Hide
    deweycd , October 20, 2011 6:40 PM
    It shall be an interesting war.
  • 5 Hide
    Onus , October 20, 2011 6:41 PM
    Oh, I think Intel will do well enough, unless their performance in the Smartphone market ends up sucking. Advanced purchasers will think "Hey, my Intel-based PC is smoking fast, why not an Intel-based smartphone?" and will go to it, unless / until reviews show that Intel is resting on its PC laurels and doesn't have the same performance in the smartphone segment.
  • -1 Hide
    DjEaZy , October 20, 2011 6:43 PM
    Intel is Underestimating ARM... while nVidia [tegra] and AMD [OpenCL] is cooperating... and... because intel haz no good GPU... ...
  • 6 Hide
    de5_Roy , October 20, 2011 6:45 PM
    since amd hasn't been able to provide much (if any) competition for intel in the cpu field lately, so why not intel vs arm? anyone who forces intel to compete and drop prices is welcome as long as users win. :) 
  • 3 Hide
    memadmax , October 20, 2011 6:45 PM
    Intel is a company that is not to be underestimated, nor "prodded" by short term trends.

    If they don't move hardcore on it, then short sell it.

    Think about it, do smartphones really have a bright future now? I think the market has been tapped and you can't really do much more to them except merge them into "PsP" and call it good.... Get the gaming market into the smartphone market and not much else.... you just can't do much more with these things, and human nature is already running its course.
  • 6 Hide
    zanny , October 20, 2011 6:46 PM
    Its power intensive to decode variable length instructions. When the instruction set is simple, its easy to interpret them. There is power overhead in x86 that has given ARM the edge.
  • -2 Hide
    memadmax , October 20, 2011 6:46 PM
    And to add to my last comment, look at the core processor segment, essentially unchanged, but powers everything.
  • 4 Hide
    GreaseMonkey_62 , October 20, 2011 6:47 PM
    Unlesss Intel uses an ARM based architecture, they probably won't get anywhere. If they're phone processors are still x86 based, that means the current phone operating systems won't work unless they are redesigned, which may or may not happen. I wouldn't mind seeing an AMD based fusion type ARM processor. That would be cool.
  • 1 Hide
    Onus , October 20, 2011 6:58 PM
    I agree, GreaseMonkey, as long as it doesn't suck down too much power. The batteries in smartphones are considerably smaller than the ones in tablets, which are pretty small as it is.
  • 6 Hide
    amk-aka-Phantom , October 20, 2011 6:59 PM
    GreaseMonkey_62Unlesss Intel uses an ARM based architecture, they probably won't get anywhere. If they're phone processors are still x86 based, that means the current phone operating systems won't work unless they are redesigned, which may or may not happen. I wouldn't mind seeing an AMD based fusion type ARM processor. That would be cool.


    Windows 8 will work. Linux will work. And I want a full-blast OS on my phone, too, so I won't mind if it's x86. But they'll have to deal with the power consumption.
  • 0 Hide
    LuckyDucky7 , October 20, 2011 7:04 PM
    I'm not sure why nobody's noticed this:

    You know when you go to do something on a tablet, and you're playing catch-up while the screen decides how to tilt, then does it in a very jerky fashion?

    Or how about when the screen decides to tilt all by itself- locking up the tablet for at least 5 seconds while it tips to the side, and then tips back?

    Or how about the fact that ads slow mobile browsers to a crawl? (Granted this happens on x86 also but not as dramatically.)

    That's because the ARM processor inside your device isn't fast enough to keep up with a slow pace of use on a device.

    So even if it uses 1W at 100% load, it really needs to be faster to get something that people want to use- people who see the device respond instantly will be partial to it. To make it faster, you need to have a higher-powered CPU.

    Now, the lowest-end CPU, which actually is the Z-01 from AMD (Intel doesn't have any good parts right now- their Atom sucks compared to the C-60 and E-450 mainly because their integrated graphics are awful).
    It runs at 6W idle. It's a 40 nm part, runs at 1 GHz, and is a dual-core part. It's basically a lower-clocked E-350/C-50.

    Now, Windows 7 is too heavy for this application. But Linux isn't- especially if you trim it down to the basics. So the machine will swap orientations and launch applications faster.

    So if you could compile Android for x86, or introduce a compatibility layer (though for the developers it shouldn't be too hard) for ARM, it would be a good-sized gain.

    Because if your tablet loads things instantly, if it swaps orientations instantly (shoudn't be hard with 80 graphics cores), people WILL buy it. The iPad 2 will look like a dinosaur.


    The only issue is battery life. We don't seem to have any good batteries (which the chemistry department really needs to work on). So for extended use of a device it must be charged or the battery changed out for a fresh one.



    Otherwise- I don't think Intel will enter the phone and tablet CPU market until either a.) ARM has a design that's actually fast or b.) Intel comes up with a design that can do x86 at faster-than-ARM speeds.
  • 5 Hide
    timeofdoom , October 20, 2011 7:24 PM
    de5_roysince amd hasn't been able to provide much (if any) competition for intel in the cpu field lately, so why not intel vs arm? anyone who forces intel to compete and drop prices is welcome as long as users win.


    While i agree that competition IS a good thing for us users... I have to disagree with the first part of your statement. APU's are gobbling up the market from Intel, ( at least they are in my country - DK, btw.) so i'd say AMD is pretty competitive ATM.
    Right, they didn't do a good job at BD. SB ate it completely. But that is only the ever-so-shrinking desktop market.
    For laptops, which are on the rise, AMD's APU's are much more attractive, since they are so cheap - adn they get the job done.

    Sure, their CPU power is weaker than the SB - by a mile - but buying a laptop with a weaker CPU, but a decent/good GPU versus buying the SB with a strong CPU, but a crap GPU...The choice should be quite easy to make.

    So - right, don't start the: "But you can buy the SB paired with another GPU in a laptop"...That's not really attractive for the average joe, because of the even higher price.

    If you are gaming on a laptop, (NOTE: YOU ARE DOING IT WRONG! :p  ) then sure, go ahead.

    But that is such a small niché.

    No, AMD have done a good job by exploiting Intels weakness right there - and as a little bonus, they also hit nVidia at the same time.

    Cheap APU's, which can multitask and play HD-content and smaller games for the average joe.
  • 0 Hide
    Jeteroll , October 20, 2011 7:28 PM
    Is this talking about ARM possibly trying to compete in the desktop/laptop scene? also how do you guys think this affects AMD?
  • 3 Hide
    pug_s , October 20, 2011 7:38 PM
    The Nvidia Kal-El and the recently announced ARM Cortex A7/A15 big/little strategy is going propel ARM to be competitive against Intel in the next year or 2. Not to mention that ARM enabled windows 8 will compete with Intel also.
  • 0 Hide
    RipperjackAU , October 20, 2011 7:44 PM
    If you are gaming on a laptop, (NOTE: YOU ARE DOING IT WRONG!) then sure, go ahead.


    Funny thing you said that, because RAGE runs better on my nVidia equiped laptop, then it does with my ATi/AMD CrossFire gaming rig!

  • 7 Hide
    burnley14 , October 20, 2011 7:52 PM
    Quote:
    Intel believes that achieving higher performance will be more difficult for ARM vendors than lower power consumption for Intel.


    I think he nailed it right there. And Intel's 22nm shrink with tri-gate transistors agrees.
  • 10 Hide
    digiex , October 20, 2011 7:59 PM
    CISC vs RISC all over again.
  • 6 Hide
    Cazalan , October 20, 2011 8:13 PM
    Clearly Intel is taking ARM seriously. They've already started volume production of 22nm Tri-Gate designs. When that technology is perfected and applied to Atom they will have significant power savings. While TSMC/GF are struggling with 32/28nm.

    ARM has a low power advantage now because their processors can't do a fraction of what a Sandy Bridge can. ARM is scaling up with A15 and Intel is scaling down. Performance/capability will cross as some point yet Intel will keep a 2+ year technology advantage in FAB technology.

    The problem ARM developers have is the tremendous competition. Samsung, Apple, Nvidia, TI, Qualcom and more. All of these competing against each other with razor thin margins on the processors themselves.
Display more comments