Google Partnering with Intel Is Completely Backwards

September 13, 2011

Today, at Intel’s IDF event, Andy Rubin announced they are making all future Android versions work on Intel’s Atom chips. I don’t like this move at all. It seems completely backwards to me. Microsoft is moving to ARM for a reason. Yes, they clearly don’t like having to move to ARM, because they are very vulnerable on ARM, having no apps that work on that chip architecture. Although, they will try to mask this, by downplaying the ARM version for now, and try to get developers to make apps (or web apps) that work on both chip architectures, and having them displayed in their new App Store, by the time they launch Windows 8 late 2012 or early 2013.

But even if they had to go through this pain, they still knew ARM is the future. And now Google does the opposite – they partner with Intel, just like they partnered with Adobe on Flash to spite Apple, even though it was the wrong thing to do. Instead of looking forward to HTML5, they looked backwards to Flash, just to have that “benefit” over Apple.

This move now seems similar. They probably want to say down the line that Android runs on “more powerful chips”, although more inefficient, more expensive, and still doubtful if Atoms will remain more powerful than high-end ARM chips. Kal-El probably has Intel’s most powerful Atom chips beat right now, and that’s for Atoms that have several times bigger TDP. I’m not expecting competitive performance from Intel’s watered down Atom chips next year, against chips like Kal-El+ and Wayne.

So when you look at it like that, it seems that Intel has nothing to offer Android from a technical point of view. It all seems political. Either it’s because Paul Otellini is on Google’s board of directors, or because Google is worried their favorite ARM chip makers will focus more on Windows 8 tablets, but it doesn’t seem like the platform was chosen on its merits.

Plus, adding a completely different chip architecture to Android’s on-going development, will probably mean significantly slower development of Android. It could mean that we’ll see fewer features in a new version, simply because they had to make sure all the ones they already have are working properly on Intel’s chips as well.

Google and Intel have partnered before with Google TV and Chromebooks, and in both occasions it has proven to be the wrong thing to do. Google TV was much more expensive than expected, and so were Chromebooks. Is it just a coincidence that in both cases they used Intel chips, the price was too high? I think not. Atom chips cost up to 5x more than a high-end ARM chip, and that translates to up to $200 at retail for the chip alone.

I really hope this fall we’ll get to see some ARM-based Google TV set top boxes and Chromebooks, priced as they should be. But these latest trends worry me, especially now that Google has a manufacturing arm in Motorola. Will they try pushing Intel into the mobile market by making an Atom based Nexus 4 next year? I find that thought pretty scary, especially since I don’t think if that came true, it would be because Atom has bested ARM in everything. It would be purely a political move, and I wouldn’t like to see Google go that way.

Comments

  • Anonymous

    I completely agree. But in the back of my mind, i think that google knows all this even better than us and this deal is about something else google is getting out of this. Clearly intel wants to be seen as running next gen mobile OSes so they asked google for android on x86, and in return, they offer xyz.

  • Anonymous

    I think GOOG is taking a long term view.
    Have you seen the performance specs of INTC Haswell
    http://www.theregister.co.uk/2011/09/13/haswell_at_adc_2011/
    20x power reduction with the performance of x86 but not available till 2013.
    Also it seems that GOOG wants to get more into the enterprise. INTC is probably key to
    this.
    Thirdly, INTC has a lot of useful patents that could shield Android to a certain degree.
    Fourthly, currrently Chrome native client only works with x86. I am sure they want chrome on Android eventually and that means native client as well.

    • https://plus.google.com/117702410245683101961/posts Lucian Armasu

      Look at the picture and what it says above. It says “Connected Stand-by Power”. That 20x reduction is only for stand-by. Intel won’t magically jump from a 40W TDP in their notebooks, to 2W. Your laptop might gain 30-60 minutes of battery life at most with that improvement. But everyone seems to have been misled into thinking that Intel’s chips will actually have a 20x smaller TDP in 2013. They won’t.

      As far as I know they’ve been working on adding ARM support to Native Client since earlier this year, and we might see ARM-based Chromebooks this fall.

  • http://www.typhonrt.org/ Michael Leahy

    I’m all for it as it opens the door to supporting existing x86 projects that would be horrendous to port to the ARM architecture. Regardless if Intel does reach that sweet spot where their SoCs will run well and efficiently for smartphones / small integrated devices this move opens the door to _wired_ x86 / Android based products. Also with Google fully supporting this path it gives OEMs outside of the smartphone sphere the peace of mind to invest in Android building a variety of hardware. For my particular use case I’m building a lowish cost audio DSP box using Android as the OS. Being able to run on x86 I can leverage a ton more DSP code that is available out there in fact I’ll likely be modifying SuperCollider (scsynth / the DSP server engine) to run on Android / x86. This providing a powerful synthesis environment that would be horrendous to port or rebuild something like it for ARM.

    http://en.wikipedia.org/wiki/SuperCollider

    Sure it may be a pain if x86 devices do get into the smartphone sphere of things regarding those leveraging native code, but the NDK will likely (err I think it already does) support easy cross-compiling and setting up multiple builds without too much trouble.

  • http://pulse.yahoo.com/_D72DSOQ4FGAX7HYK5NXUXSFHPE Tanisha Sayyad
  • David Wittenbrink

    I personally can’t share youre opinion on the adobe flash thing.. I’think no company has the right to block another out like apple does it with flash. Dont get me wrong, I love html5, in fact i learned it this summer mainly to write for android. Still, Im happy to be able to see all of the web :)
    On the intel thing i can only say that im not impressed.. android is an open platform and they want to give companys like htc, samsung, … The oportunity to have complete freedom in developing smartphones and tablets. I suppose as long as the java virtual machine runs on the intel chip applications shouldnt be that much of a problem :)
    Greets David

  • Anonymous

    I doubt that development will slow for Android. My understanding is that they only need to make changes to support the hardware (kernel level, hardware abstraction layer) they probably created a x86 hardware support team to work on a parallel level with the current ARM team. It would make sense that way.

    Partnering with Intel does have some major benefits as well! Can you imagine how the x86 support will improve the Android Emulator in the Windows SDK? Think of how more people like you and me will be able to do work on Android, test and publish with out having a dedicated testing device like the Nexus S.

    By partnering with Intel, Google will also get some influence on Intel’s future roadmaps, chip design and help steer the mobile market. Intel entering the Mobile space is inevitable. Anyone can see the potential profit to be made in hardware for mobile. Intel’s chip’s are also arguably more powerful than ARM’s. If Google didn’t partner with Intel then someone else would.

  • Anonymous

    Google’s goal is to have the highest amount of market saturation inorder to sell more ads. Being on x86 means more monies for google.

  • Lobotomik

    ARM is very nice, and very low power, but at the same MHz speed, x86 runs RINGS around it, even Atom, and MUCH more so Core iX. KalEl sounds nice, but it probably compares with a single-core Atom in raw horsepower, surely not better than dual-core Atom in any case.

    The advantage of ARM is that all that speed is not really necessary in many applications, and it can perform great in the limited apps that we use in pads and phones (and to a lesser extent, notebooks), with much less power dissipation.

  • http://amishbomb.com Tom

    I appreciate your concern for consumers wallets, but I disagree with you on the issue of Flash and HTML5 as far as Android goes. I’d rather have some backwards compatibility for the last 5 years of development on web apps than wait for HTML5 to develop or port projects that already exist in Flash.

    With regards to the news of the partnership, I think this move is more of something Intel needs vs. Google needs; if Android continues growth and computing becomes increasingly mobile, Intel could be missing a big piece of the pie if they miss out on Android. Hopefully this will create more competition in pricing (cheaper Intels since ARM is cheap already) and more niche markets for future devices, faster tablets? better battery life… Unless Google ditches ARM altogether kind of like Apple ditched PPC (think OSX Intel vs. OSX PPC) I think we’re going to be just fine.