Report: An ARM version of Windows is in the works

Right now, Microsoft is making no excuses about shoving Windows 7 into the tablet market. In a couple of years, though, we may see a fresh version of Windows tailored for the very ARM-powered slates Microsoft is trying to beat right now. So suggest separate reports by the Wall Street Journal and Bloomberg, both of which quote anonymous sources "familiar" with the software giant’s plans.

Based on those reports, Microsoft will announce the upcoming, ARM-compatible operating system at the Consumer Electronics Show early next month. The final product "isn’t expected to be available for two years," the WSJ adds. By then, it’s anyone’s guess what kind of touch-friendly operating systems will be lurking on next-generation tablets—or how much market share Intel’s next-gen Atom chips will have managed to snatch away from ARM.

Speaking of Intel, the WSJ suggests that Microsoft’s mystery OS will support both ARM processors and x86 chips from Intel and AMD. Word is that the addition of ARM compatibility is "part of a broader push at Microsoft to make Windows more ‘modular’ so that pieces of the operating system that are unnecessary for smaller, low-power devices . . . can be easily stripped away."

I expect the real challenge for Microsoft here may be software compatibility. Windows’ main selling point is its support for a wealth of software compiled exclusively for x86 processors. An ARM-compatible release would compromise that, unless Microsoft could persuade a majority of software developers to release "fat" binaries with both ARM and Intel support. But even then, different user interfaces would presumably be required for desktop and tablet versions of the same apps.

Comments closed
    • Palek
    • 12 years ago

    Oooh, truly cringe-worthy. You deserve a trophy!

    • ronch
    • 12 years ago

    The reason why I think this MAY be possible/feasible is Apple. Apple always kept changing their underlying platform. First, Motorola, then PowerPC, and now x86. I’m not an Apple owner and am not really familiar with Apple’s development history, but those old programs written for old Macs based on Motorola must have run on PowerPC-based Macs. It’s just the OS that talks to the PowerPC CPU and takes care of the app written for Motorola CPUs. And likewise apps written for PowerPC-based Macs would probably run on today’s Intel-based Macs. Can’t Windows do that as well? Heck, even today’s x86 CPUs aren’t really x86, but have RISC execution engines with x86 front ends. They just do the translation really fast in hardware to feed RISC execution units that have entirely different registers and instruction sets. x86 code running on a CPU that’s actually not x86 presents some interesting possibilities. If they can just do the translation in a different manner to feed x86 code to a different architecture, then bingo.

    • sweatshopking
    • 12 years ago

    this just seem like windows 8 to me. With 8, it shouldn’t matter what the underlying hardware is, as everything should be virtualized. when they announce that windows 8 will support arm, you can say “man, that ssk guy knows his stuff!”

    • sweatshopking
    • 12 years ago

    right. but there are people who want windows on their ipad

    • LauRoman
    • 12 years ago

    So this isn’t a version of Windows 7 CE Series OS 7 Series OS Windows 7 Series, WIndow CE being the central part? Will it actually be ARM compiled code from the Windows codebase.

    • HisDivineShadow
    • 12 years ago

    This seems to segue well with all the talk of MS getting into virtualization in an even bigger way than they already have been. I expect them to make the type of CPU (ARM vs x86) less important than performance in a given application/workload with a modular OS that allows you to buy certain “components” that you need module by module. By then, they’ll probably be leaning toward a service model to help sell it.

    I think, though, that this will de-emphasize Windows and open the door for competitors to snatch the crown jewels out from under MS. That is, if Chrome OS can last long enough to have its features integrated into Android to make a super combo of Android+ChromeOS, letting it step into a full desktop OS role in a future where the client PC matters only as much as the cloud (not less than like Google would like, not more than like MS/Apple would like).

    I expect Apple to continue the slow migration of Mac users from OS X to a very iOS-like future. The OSX app store is just the beginning. As iPad’s increase in performance, Jobs will begin questioning why they even sell low-end Macbooks when the iPad soon does everything it can do, just somewhat more slowly. Remember that Jobs is not one to let function keep him from making form that much more chic…

    • Grigory
    • 12 years ago

    Well said.

    • moog
    • 12 years ago

    Yes, legacy stuff is virtualized now.

    • BobbinThreadbare
    • 12 years ago

    What about the fact that no one has been able to make a product which is competitive in performance and cost in the desktop marketplace?

    x86 will just adapt, the modern 64 bit chips don’t have a great deal in common with the 8086 any more anyways.

    • Beomagi
    • 12 years ago

    Isn’t that called Windows CE?

    The big issue isn’t just getting windows to run basing 32 bit apps, but getting all the libraries as well. For that reason, I doubt we’ll see the desktop OS brought to ARM.

    • Chrispy_
    • 12 years ago

    l[<"so that pieces of the operating system that are unnecessary for smaller, low-power devices . . . can be easily stripped away."<]l Or maybe pieces of the operating system that are unnecessary for, well, /[

    • thermistor
    • 12 years ago

    #34…I don’t know enough to disagree. However, Apple had to ‘break’ things to move away from Motorola to Intel; not all of their customer base was happy about it.

    • blastdoor
    • 12 years ago

    This is not such a far fetched statement as people might imagine. The success of x86 is entirely dependent on two things:

    1. customers’ need for backwards compatibility
    2. the economies of scale that give Intel an advantage in process tech

    Trouble is, #1 is increasingly an artificial restriction. Apple and other Unix vendors don’t allow themselves to be constrained by it. If Microsoft makes moving software from x86 to ARM as easy as Apple made moving from PowerPC to Intel to ARM, then #1 fades fast. As soon as #1 is gone then #2 is at risk.

    It’s really not hard to imagine a tipping point coming at about the same time as, say, Windows 9, when x86’s world comes crashing down very dramatically. The future could be POWER/SPARC/IA64 at the high end and ARM everywhere else. Or heck, it could just be ARM everywhere.

    • UberGerbil
    • 12 years ago

    Sure, through emulation (like Parallels used to do for Windows on PowerPC-based Macs). But running x86 in emulation on ARM is going to be s-l-o-w. There are tricks to improve that — transmeta / FX!32-style on-the-fly recompilation, paravirtualization, etc — and it’s possible they’ll figure out how to get acceptable performance (though in many cases you end up trading off size for speed, and execution / address / storage space is also at a premium on mobile devices), but that’s a much bigger challenge than getting Windows ported to ARM (given that the core NT codebase is, or at least was, actually pretty portable).

    You’re right though that it would be the right thing to do from a competitive standpoint, because the only thing Windows really has going for it is the huge library of existing binaries, and so the only way to leverage that asset is to get them running unmodified on the new hardware. The question is whether it’s possible to do that in a performant way, and to do so in a timeframe where it is still relevant.

    The other possibility is for MS to do something sneaky with “the cloud” — where they actually run the x86 code for you on an x86 machine out on the internet, and give you a Remote Desktop view of it on your ARM device.. all done silently and transparently in a way that makes you think you’re running that x86 program locally. That runs up against bandwidth and latency issues, of course. (And licensing possibly as well, though that’s the kind of thing that doesn’t require a technical solution, just money and lawyers)

    • bhtooefr
    • 12 years ago

    To run Windows, it must be x86.

    80386 isn’t protected by patents any more, though. Problem with that is, an 80386 can only run Windows NT 4 or so. IIRC, Windows 2000 requires an 80486, Windows XP requires a Pentium, Vista and 7 require a K6-2 or Pentium II.

    (And by “requires,” I mean, it WILL NOT run on anything older. Microsoft won’t support it on something that old, of course.)

    • UberGerbil
    • 12 years ago

    The MIPS (and DEC Alpha) versions of Windows were full Windows NT (as it was at the time) builds; they were not the relatively separate codebase that is Windows CE / Mobile / Phone 7. The IA-64 (Itanium) version is the last vestige of that portable “full-blown” kernel, and demonstrates it remains reasonably portable.

    This sounds like a continuation of those full versions, not the mobile versions, so yes it’s like the MIPS version but no it’s not like the Windows CE versions.

    • axeman
    • 12 years ago

    There’s a lot of bloated messes that don’t use .NET either… what boggles my mind, though, is that it’s a huge runtime (full download of 3.5SP1 is 231MB) to run code that isn’t really portable. Supposedly you can in theory, but does this actually exist?Any programmers here know what the bulk of this is? .Net 2.0 wasn’t that big IIRC. Maybe it is WPF that’s responsible for all those bits? WPF apps are just hideous IMO. If WPF is the future of Win x86, ick.

    • SonicSilicon
    • 12 years ago

    That’s “through” an iPad, not “on” one. It does make for a sleek thin client, but that still means tethering to an actual x86 system by some means.

    • albundy
    • 12 years ago

    direct 3d on non x86? i knew pigs could fly! but i dont want the world to freeze over!

    • srg86
    • 12 years ago

    Actually, you’ll find relatively little uses the .NET framework, not even Microsoft Office for example. .NET is a bloated mess.

    • sweatshopking
    • 12 years ago
    • nagashi
    • 12 years ago

    “An ARM-compatible release would compromise that, unless Microsoft could persuade a majority of software developers to release “fat” binaries with both ARM and Intel support.”

    …or they could release .net binaries. Like most windows software made in the last 4 years. Very few new applications use C/C++ these days, mostly things that really need the perf (games, math/science softwares). This transition will go quite smoothly I predict.

    • jalex3
    • 12 years ago

    would this not just be windows ce7 out of beta?…

    • jdaven
    • 12 years ago

    Of all the websites on all the web, I haven’t seen the following:

    “Boot Camp” for the iPad.

    Enough said.

    • ronch
    • 12 years ago

    Just a thought.

    Is it possible for Microsoft to make an OS that runs all our Windows apps while running on a different underlying hardware platform? Intel’s been very selfish licensing x86 to anyone. If a group of chip makers will band together and create a new (if not a TOTALLY NEW) CPU IA architecture which Windows can run on and which is open, we can scrap Intel altogether. More players mean more innovation. This will undoubtedly lead to a shake out, but at least no one holds a firm grasp on IA licenses and anyone who wants to create an implementation of this new architecture may do so at no expense except its own R&D. It should be open. And with more players innovating, the better it is for the consumer.

    Also, SPARC, ARM and MIPS are licensable but royalties must be paid and/or only part of the architecture is open.

    • SonicSilicon
    • 12 years ago

    I suppose, though Windows for MIPS should be added, even if it hasn’t been true for the greater part of a decade.

    • jstern
    • 12 years ago

    I’m enjoying my new Android phone, but it’s a Windows world, and I hate not being able to run and do certain things that I’m able to on Windows. So I’m hoping for a phone with an x86 processor. If I had money to waste and decided to get a tablet, then I would get a Windows 7 one. Just being able to install a codec pack, or use a million other programs would make it a better choice for me.

    • OneArmedScissor
    • 12 years ago

    Vista should have set off an even bigger alarm than iPhones, as far as the, “part of a broader push at Microsoft to make Windows more ‘modular’ so that pieces of the operating system that are unnecessary for smaller, low-power devices . . . can be easily stripped away,” thing goes.

    It drives me nuts that you can’t just pick out what you want in your Windows installation when you are setting it up. They sell so many different versions, but they’re all running junk just about nobody is going to need.

    • Grigory
    • 12 years ago

    And one more thing: It’s about time that it gets warm after wintel. 😉

    • Grigory
    • 12 years ago

    I know, that was actually my point. I tried to make fun of teja’s not-so-very-brilliant comment that way. 😉

    • esterhasz
    • 12 years ago

    It’s a fresh start for application programming, the issue at hand. WinMO 6.5 applications no longer run on WP7.

    • dpaus
    • 12 years ago

    In name only!

    • Corrado
    • 12 years ago

    Isn’t Windows Phone 7 and every iteration of Windows CE/Windows Mobile already ‘Windows for ARM’?

    • FireGryphon
    • 12 years ago

    There’s no new technology here, just a new product. How, in tech industry, does a product that’s two years off qualify as news? It’s already obsolete!

    • khands
    • 12 years ago

    true, if a little difficult to follow.

    • tay
    • 12 years ago

    WP7 isn’t a fresh start rewrite from scratch. They simply added a good layer on top of what was Winmob 6.5 along with a good runtime.

    • Jigar
    • 12 years ago

    And you fed the troll… Leave him alone or else he will continue…

    • esterhasz
    • 12 years ago

    The question of how to get software to work with different CPUs has become less difficult over the last years as programmers increasingly use higher level APIs that hide the hardware behind comfy software interfaces. Microsoft has been pushing .NET (and its already very portable offspring, Silverlight) for quite a while and in a similar fashion to JAVA, you only have to write a new CLR to get a piece of code running on a different architecture.

    The real problems come with accommodating different form factors, input methods, quality control, easy of use, etc. while keeping compatibility with current software. The only real possibility I see here would be to make a fresh start (like they did with WP7) and push the legacy stuff into a VM…

    • TaBoVilla
    • 12 years ago

    dude, the beginning of the end for everything has already begun when talking about decades into the future =)

    • dpaus
    • 12 years ago

    No, they should have started working on it as a contingency plan about 3 years ago, when the ‘current generation’ of smartphones was ushered in by the iPhone and the writing was clearly on the wall. Then, they might actually have it right now, and the competitive landscape for tablets would – at a minimum – be very different. But actually creating /[

    • Grigory
    • 12 years ago

    Why? He is right! Just give it a couple of decades.

    • geekl33tgamer
    • 12 years ago

    What have you been smoking, exactly?

    • tejas84
    • 12 years ago

    The beginning of the end for x86…

    • blastdoor
    • 12 years ago

    Yeah, two years is way too late. In fact, if they really aren’t going to have it for another two years, then they shouldn’t even announce it. Instead, they should move forward with the project as a secret back-up plan in case Intel doesn’t get its act together. At the same time, they should make any updates to their developer tools to make it easier for developers to make ARM versions of mobile apps if/when they do make the change. Note this is not an original idea — it’s the Apple playbook.

    • TaBoVilla
    • 12 years ago

    oh cyril:
    “..or how much market share Intel’s next-gen Atom chips will have managed to snatch away from ARM..”

    really, really dunno about this one. I think it has been proven that manufacturers are not THAT into x86 environments for mobile markets, due to lack of competitive platforms and mobile OSes, applications etc.

    ARM is happily eating that cake alone and no one will dare challenge it. I think AMD’s recently revealed strategy (thanks to platform previews and demos for upcoming Bobcat processors) showed that AMD considers 1-2W power APUs to become their lowest end parts for these designs, so, correct me if I’m mistaken, AMD does not plan to go head to head with ARM on sub-watt segments, vital for mobiles, UMPCs and handsets, in the near future

    and intel should follow suit, quit trying to stuff atom into mobiles and make decent tablet and ultra portables CPU+GPU. They don’t stand a chance against ARM system on chips. Microsoft probably realized this already, therefore, Windows for ARM.

    • dpaus
    • 12 years ago

    Two years?!? And without a compelling advantage over WebOS/iOS/Android/QNX platforms?

    YATL/TLMP: Yet Another Too-Little/Too-Late Microsoft Product…

    EDIT: Oh, crap, I accidentally posted first. Can one of the admins change my post number to 6 or 7 or so…?

Pin It on Pinterest

Share This

Share this post with your friends!