Intel says ARM Windows 8 won’t run x86 apps

If you were hoping for some sort of Rosetta-like emulation layer to enable Windows 8 to run x86 applications when installed on an ARM system, well, prepare to be disappointed. EE Times quotes Renee James, the head of Intel’s Software and Services business, as saying the ARM version of Windows 8 “will not be running legacy apps not now or ever.”

The x86 version of Windows 8, on the other hand, will reportedly have a “Windows 7 compatibility mode.” That means, going forward, software companies that want their Windows apps to run on ARM machines might be forced to release new versions compiled specifically for the architecture. Such a requirement wouldn’t rule out the possibility of fat executables containing machine code for both x86 and ARM systems, however.

Then again, that might just be the tip of the iceberg. EE Times says Intel CEO Paul Otellini went so far as to claim, “The ARM guys are getting a port to Windows, but it’s really four ports [because] every OS has to be written to a chip so Microsoft is really doing four ports of Windows to ARM.” Otellini didn’t elaborate on what exactly those four versions are, though EE Times says he “showed slides suggesting they could be targeted to specific versions of the ARM core or SoC implementations of the cores from ARM licensees.”

I suspect we’ll soon hear some sort of clarification from Microsoft about all this. You’ve got to remember Intel has quite a big stake in seeing Windows on ARM fail, so seeing the chipmaker spread a little fear, uncertainty, and doubt around the subject seems natural.

Comments closed
    • ET3D
    • 8 years ago

    Okay, let’s try to cut through the FUD.

    Four core versions of ARM will be supported. How many x86 variations are supported by Windows? There are two distinct versions of Windows (x86, x64), and multiple cores with different extensions supported. Windows has specific support in the code for many of these extensions and variations (SSE, …). It doesn’t look to me like the ARM situation could be much worse, and my guess is that it’s actually better.

    As for legacy apps, first of all, Windows is typically pretty bad about that. I have a pretty large stack of games (and some other software) which haven’t survived the move from 98->2000->XP->Vista (I’m still stuck at Vista, but I understand that Vista->7 wasn’t a real problem for games).

    It makes sense that there’s truth to Intel’s claim that “legacy applications” will not be supported. The question is, how bad will this affect users.

    I imagine that Windows on ARM will support .NET. It would make no sense otherwise (considering that .NET is supported on Zune, Xbox and Windows Phone). Which would mean that a pretty large body of software should run as is on the ARM. Sure, that’s just a certain segment of software, but it’s not that small, and is pretty prevalent in the business sector. The move to a cross architecture OS should strengthen .NET even more.

    Some of the native x86 software is already multi-platform and will likely make the move quickly. Web browsers are the obvious example, OpenOffice (and its variations) probably also. Microsoft software will surely be released for the new platform. So users who do mostly web and office should be mostly covered.

    It’s true that a lot of the software for enthusiasts will not survive the move, in particular games (not that ARM systems are likely to be great games machines in the short run). It will be interesting how things turn out when it comes to such apps. However if (as I argued above) the common users and many businesses are already covered, then Windows for ARM should have no problem gaining some traction.

    Edit: I forgot cloud apps, which should run fine on the ARM, and I imagine that Remote Desktop will also run on that version of Windows. So yet more use cases where the ARM machine could serve as well as an x86 one.

      • indeego
      • 8 years ago

      According to wiki:
      [i<]While the standards that make up .NET are inherently cross platform, Microsoft's full implementation of .NET is only supported on Windows. Microsoft does provide limited .NET subsets for other platforms such as XNA for Windows, XBOX 360 and Windows Phone 7, Silverlight for Windows and Mac OS X. Alternative implementations of the CLR, base class libraries, and compilers also exist (sometimes from other vendors). While all of these implementations are based on the same standards, they are still different implementations with varying levels of completeness in comparison to the full .NET version Microsoft ships for Windows and are on occasion incompatible.[/i<] Let's not get into all the freaking versions of .net, and their respective patches and dependencies. Would like to see your stats on .NET apps, I have serious doubts that it's in wide use. In our business, it doesn't even make up 20% of our apps. And let's just say that zero one of these apps will work on ARM until mass testing has gone on for many, many years. You don't just migrate to a new platform because it's a new platform, especially in business. We still have a few LOB applications that work on XP that do not work on 7 64-bit (many or 16-bit, hah). These applications do not exist in any smartphone world, and certainly won't be ported to ARM unless demand gets there first.

        • Flatland_Spider
        • 8 years ago

        What makes you think that MS won’t release a full .Net implementation for Windows ARM? It will be a full Windows environment, and if I remember correctly, most of the missing components are related to creating Windows GUIs. Besides, XNA, XBox, Phone 7, and Silverlight are all targeted variations, so it really doesn’t make sense to use them as a counter-point.

        Mono the patented GUI stuff MS has in .Net. Mono probably could reverse engineer WinForms, but it wouldn’t do them any good, since it is mainly targeted at Linux/Unix, and it would be nothing but patent troll bait.

          • mesyn191
          • 8 years ago

          MS has tons of issues with .NET on x86 as is, you really think they’re gonna have a decent much less good or even fully working .NET ARM port in Win8? Hell ever? If they do get .NET working on ARM it’ll probably be .NET 3.5 or 4 or whatever and later only and it’ll probably be half assed and buggy at that.

      • mesyn191
      • 8 years ago

      Are you being paid to say this stuff or what?

      Most of the different “versions” of x86, AMD64 and Intel64 in particular, are very very similar and were designed to be relatively easy to port older x86 apps too. So that is at very best an apples to oranges comparison to the difficulties in store for a ARM port for any x86 software or OS.

      Also SSE2,4, etc. are SIMD which is a whole different ball of wax. And there is a metric fucking shitton of stuff that doesn’t work with .NET so I have no clue why you’re handwaving that around as an example of cross platform software development being easy or even “easy”.

      Win8 on ARM will probably be for cellphones and MID’s and stuff. Maaaaaaaaybe for servers if ARM really takes off there, which is very doubtful too. But on desktops and laptops its going to be a long shot to say the least.

    • ronch
    • 8 years ago

    I’ve been using Ubuntu for about 1.5 years and I’ve been ‘peddling’ it to some family and friends, citing some advantages and disadvantages. Main reason why people stay off Ubuntu/Linux? If it won’t run their Windows apps, they won’t switch.

    Whatever MS calls their next OS for the desktop, if it won’t run legacy Windows apps, it would be hard selling it to the world even if MS calls it Windows. It would be just another OS fighting an uphill battle like Linux, except it’s governed by a single company everyone is afraid of getting too fat. They’d better figure out a way for it to run x86 apps on ARM.

      • Flatland_Spider
      • 8 years ago

      MS has started to use virtualization a way to bridge the gap between their older OSes and their new OSes. Windows 7 has XP Mode, which is nothing more then a seamless XP environment running in Virtual PC.

    • PRIME1
    • 8 years ago

    Intel says you can take my monopoly from my cold dead hands!

    [url<]http://i.imgur.com/fG5v0.jpg[/url<]

    • HighTech4US2
    • 8 years ago

    Microsoft calls out Intel’s FUD

    [url<]http://www.theregister.co.uk/2011/05/19/microsoft_contradicts_renee_james_on_windows_8[/url<] "Intel’s statements during yesterday’s Intel Investor Meeting about Microsoft’s plans for the next version of Windows were factually inaccurate and unfortunately misleading

    • albundy
    • 8 years ago

    if thats the case, why not just use the windows phone 7 os on tablets? complete and utter FAIL if no x86 software is supported! especially directx games. might as well go to ipad at this point.

    • Corrado
    • 8 years ago

    You guys also realize that iOS is a ‘stripped down OS X’? Is it possible that Windows 8 for ARM is a similar type of thing? Have they said you’ll get the exact same experience between x86 and ARM?

      • sweatshopking
      • 8 years ago

      no. they’ve said that there will be at least 2 different UI’s though. It will likely be the same kernel, but not the same OS.

    • mutarasector
    • 8 years ago

    re: “Intel says ARM Windows 8 won’t run x86 apps”

    Ok Chipzilla, yer >point< is? No x86 VM for ARM? BFD… ARM doesn’t do Windows 7 apps very well either, does it?

    As for the 4 ports, my guess would be the Cortex ‘A’ variants, the A9 and A15 for certain.

    • willmore
    • 8 years ago

    As a long time Linux user, I’m getting a lot of perverse enjoyment from this. Port to a new platform? Oh, noes. Crap, I’ve lost count of what all Linux runs on. x86, x86_64, ARM, ALPHA, SPARC, HP/PA, toasters, refrigerators……

      • boing
      • 8 years ago

      I don’t understand the enjoyment. NetBSD runs on even more platforms, does this mean I’ll get to feel even more perverse enjoyment if I install NetBSD?

        • indeego
        • 8 years ago

        Mostly an embedded network-only OS. While it definitely has its perks, it doesn’t have the GUI foundations Linux has.

        • willmore
        • 8 years ago

        No, running NetBSD just means that you’re perverse, nothing more.

      • ronch
      • 8 years ago

      Just a question from a relatively new Linux user. Does this mean that say, a particular software package runs on a Linux distro, and that Linux distro supports many different architectures (x86, amd64, PowerPC, etc.), can that package run on those different Linux distros for other architectures too?

        • Flatland_Spider
        • 8 years ago

        It depends on language the source code was written in.

        If the program was written in C, C++, or some other compiled language, the program will need to be ported to the platform, compiled using a compiler for the target architecture and debugged.

        If the program was written in Python, Ruby, Java, or some other interpreted language, the program will run with out having to port it, assuming the interpreter or VM has been ported and the program doesn’t rely on code outside of the interpreter or VM.

        Additionally, each package is created by someone who takes the time to make sure the program runs on that particular architecture. That’s why you’ll see one platform with 10,000 packages and another with only 5,000.

          • willmore
          • 8 years ago

          Quite so. To add to that, most code under UNIX like systems has seen a migration from one arch (or even sub architecture) to another which means that *some* of the architecture dependent stuff has been rubbed off. Not everything will port nicely, but it’s more likely than not that code will run well on multiple architectures.

          Some of the incompatability is bad programming practice–if you only ever write C to run on x86, you can learn some bad habbits like doing things with pointers that aren’t safe. And, some of it is due to the architecture being strict or ‘odd’ in certain ways.

          A lot of GNU and other code used under Linux (and the BSDs) has also ported from 32 bit to 64 bit sytems and that’s helped clean up a lot of latent bugs.

          Even if a package doesn’t compile right out of the box, it’s generally not that hard to get it working–unless it relies on foibles of a particular architecture to a great extent. Usually all it takes is munging around some variable types or cleaning up some crufty pointer manipulation. Oh, endianess is a big one–especially for architectures which are bi-endian. Poor MIPS.

          • Anonymous Coward
          • 8 years ago

          [quote<]If the program was written in C, C++, or some other compiled language, the program will need to be ported to the platform, compiled using a compiler for the target architecture and debugged.[/quote<] I think thats over-dramatic for most applications. Linux/unix developers have never been under the impression that x86 was the only instruction set. Mostly we're just talking about a compile.

    • ShadowTiger
    • 8 years ago

    First they should make a stripped down version of Windows that is light weight with modular add-ons for all the non-core features.

    Once they have that done it makes sense to target lower power platforms.

    • Krogoth
    • 8 years ago

    Why does this matter?

    ARM and X86 are different animals, after two different markets.

    Windows 8 ARM is just MS’s way of cashing on the whole smartphone, tablet crazy “a.k.a bubble”.

      • Meadows
      • 8 years ago

      [quote<]Windows 8 ARM is just MS's way of cashing on the whole smartphone, tablet crazy "a.k.a bubble".[/quote<] Wow, I'm having a field day with you today. What's a "tablet crazy-also-known-as-bubble"?

    • hansmuff
    • 8 years ago

    This seems to only make sense. Please do not make the mistake of mixing ARM with “desktop”. The ARM ports are for portable devices, not for your Desktop or Laptop machine.

    Current Intel desktop chips are very, very powerful compared to anything in a mobile device. If and how things converge at some point is another story, but we’re quite a ways away from it.

    There will be a Windows 8 for x86 (x86_64) and that’s what your desktop and laptop will run, along with the near and dear “old” x86 apps.

    It isn’t surprising in the least that a portable ARM machine won’t run or emulate x86. It would be entirely pointless, it’s for a target segment that doesn’t care about, or require, x86 software.

      • HighTech4US2
      • 8 years ago

      Windows 8 for ARM will support Desktops, Laptops, Netbooks and Servers.

      Haven’t you heard of Project Denver from Nvidia?

      [url<]http://pressroom.nvidia.com/easyir/customrel.do?easyirid=A0D622CE9F579F09&version=live&releasejsp=release_157&xhtml=true&prid=705184[/url<]

    • Xylker
    • 8 years ago

    If you don’t code, who cares? The result should be the same.

    Obviously, it will be necessary to label software in a way that average consumers know whether that particular program will run on their particular system, but outside of that it won’t really matter.

    As a marketer, I would want to ship my software with full compatibility; and design an installer which was able to determine the architecture on which the OS was running so that no one needed to know the difference.

      • Flatland_Spider
      • 8 years ago

      The solution is simple. Software for Win 8 ARM comes from the Windows Store only, or the software is .Net only.

      MS is working on the AppX package technology, which could do exactly that.

    • The Dark One
    • 8 years ago

    The guy whose company is renowned for spreading FUD is predicting doom for any user who dares to use Windows on a non-X86 architecture? Good Lord!

    • jcw122
    • 8 years ago

    No one cares about x86, Intel.

      • Meadows
      • 8 years ago

      I do, all my software uses it.

        • sweatshopking
        • 8 years ago

        then get new software. Mines all amd64

          • Meadows
          • 8 years ago

          Which is x86.

            • sweatshopking
            • 8 years ago

            i know. i was teasing. silly boy.

            • Meadows
            • 8 years ago

            At least do it in a way such that people don’t think you’re undereducated.

            • sweatshopking
            • 8 years ago

            undereducated? i’m not ANY educated.

    • mcnabney
    • 8 years ago

    What benefit is there to run a Windows variant if you have no access to legacy apps?

    If I didn’t need/want to run legacy apps and games on a computing device I would certainly make use of a Linux-based OS. Why absorb the WinBloat without the WinStuff?

      • Ushio01
      • 8 years ago

      Familiarity the reason linux failed and xp won with netbooks will also apply to ARM powered netbooks/laptops.

        • mcnabney
        • 8 years ago

        But those XP netbooks really could run all of those old Windows programs. My netbook exists solely to play movies for my kids and Starcraft/Civ2&3 for me when I travel.

      • Anonymous Coward
      • 8 years ago

      I don’t see much remaining value in legacy apps, especially not on any new age computing form factor. For corporate crap and for old games, x86 computers continue as always.

    • designerfx
    • 8 years ago

    the “4” are not hard to figure out.

    X86, X64, server, desktop/workstation.

      • bhtooefr
      • 8 years ago

      No, the 4 are TI OMAP, Qualcomm Snapdragon, Nvidia Tegra, and an unknown partner.

      Outside of x86-land, there’s no solidly defined layout of a system, so different chip makers use different layouts. (In x86-land, IBM defined the initial layout with the PC AT, and subsequent layouts have been defined by Intel, primarily.)

      Therefore, you have to target your OS to both the CPU architecture, and to the rest of the system.

      However, this is a non-issue – Windows NT has had support for using different Hardware Abstraction Layers for different motherboards since the beginning.

        • TaBoVilla
        • 8 years ago

        nay, you are both wrong.. Microsoft is not addressing manufacturer specific ARM designs, they are talking most likely about ARM architecture and instruction sets: such as A8 and A9 (ARMv7) for current devices, legacy ARM11 cores for older devices and compatibility with future stuff based on ARM15 designs including 64bit ARM processors, etc.

      • Meadows
      • 8 years ago

      The difference between a fool and a clown is confidence. Hello, clown.

    • jensend
    • 8 years ago

    Definitely FUD. If Microsoft doesn’t build the capability into the OS you’d better believe 3rd parties will provide it. The only questions are what fraction of native x86 performance will be achieved and correspondingly how many of those using x86 would get sufficient performance from an ARM solution.

    • Kurotetsu
    • 8 years ago

    This doesn’t seem like a terribly huge issue? Besides Windows itself, Microsoft’s other most significant product, Office, is already being ported to the web (like many, many other apps out there). And from what I understand most anything written using the .NET framework should be able to run on ARM (though I’m not sure how many “legacy” applications this applies to).

    • odizzido
    • 8 years ago

    No x86? Well that just got rid of any reason why I would want this.

    • Da_Boss
    • 8 years ago

    Anyone else get the sense that Microsoft and Intel are nothing more than relics of an old age, convinced that the only way to become relevant again in this new and rapidly growing consumer computing sector, they need to polish off their old turds, and re-sell them to a market that has long since moved on to more suitable products?

      • holophrastic
      • 8 years ago

      How can you think that the only two companies pushing the industry forward are relics? You can’t name three other companies in the entire industry who steadily innovate new technologies to push computing power forward. Sure Apple makes things look better, and sure AMD produces excellent equipment, but Microsoft is the only one driving business to spend money on better hardware, and Intel’s the only one designing better hardware, with any degree of consistency.

      Just look at the Vista launch — how many people do you know who bought new computers to support vista? And look at what’s pushed them into intel chips.

      No one else has been able to force the market to move forward in a performance or reliability direction.

      But hey, the recent chrome pc concept says it all — “people shouldn’t be able to manage their own machines”. That’s the world of consumerism. You think Intel and Microsoft are relics because you used to use business machines, and now you’re content with consumer machines (which didn’t really exist before). But you forget that consumer machines come out of business machines. The innovation in technology comes at the business level — where it has financial value to someone. Your consumer-grade carp only innovates at the marketing level — which is why you see it as fore-front.

        • Corrado
        • 8 years ago

        The ‘people shouldn’t be able to manage their own machines’ is exactly what MOST people want. Its the same with a dishwasher, a fridge, a washing machine, a car, a lawnmower. TO MOST people, a computer is an appliance/tool to get something done, not a hobby, or toy. They want to push the button, do what they want to do, have the computer do it as fast and easy as possible, and then walk away from it.

        50-60 years ago, you were EXPECTED to be able to work on your own car and fix things that broke and manage them. Nowadays, its all done by a dealer or not at all.

        • Da_Boss
        • 8 years ago

        Firstly, I don’t doubt that the technology that is driven by intel at the high-end does slowly make its way down to the lower end and consumer products. What makes me laugh is the form that it takes.

        I mean, as if Microsoft thought it was a better idea to make 4 different ARM ports of Windows, a relatively sprawling OS built on the foundations of decades old technologies and UI design in order to cater to backwards compatibility. The unintended result of how Windows has evolved to this point, however, is that Windows is becoming increasingly unsuitable for this new generation of cheap, low-power, touch-driven devices.

        As for Intel, they’ve slowly been trying to make their x86 processors smaller and more efficient for years now, but still cannot match the performance per watt of current ARM CPUs. Not that they’re incapable of doing it–Intel has the biggest budgets, the best engineers, and the best chip manufacturing in the world. Look how they’ve been able to do with x86, even after 20 years of backwards compatibility and numerous specialized instruction sets. But the game is different in this new consumer space. It’s all about how much performance you can get out of a ~1W CPU, and it’s probably going to require them to let go of x86 for CPUs catered to that sector. I don’t think they’ve realized it yet, though.

        Again, I don’t doubt their contributions to the current consumer computing revolution indirectly, I just don’t think they’ll be the major driving force they once were, mostly due to their inability to make more radical changes to their more traditional business models.

          • Hattig
          • 8 years ago

          I’m sure that in reality Microsoft has said that their ARM Windows will run on any ARMv7 architecture core with NEON FPU. Four ports is highly unlikely, more like one single ARM kernel with various drivers supporting the other functions. When ARM64 comes out, I guess that will be the second ARM kernel.

          Why else has NEON suddenly appeared in Tegra 3, when previously NVIDIA were saying it was a waste of die space and that the functions could run on the integrated GPU, or via the less powerful ARM FPU?

          Other hardware on the SoCs is merely a matter of writing the drivers, or integrating the manufacturer supplied ones. This really isn’t a major issue.

          Of course there will be some SoC specific hardware like TI’s DSPs, or AES/Security cores, that will need special support – but the same goes for x86 platforms.

            • Flatland_Spider
            • 8 years ago

            Everyone forgets the NT kernel is designed to be ported. MS may have different HALs for different ARM implementations, but they don’t have four different NT ARM kernels.

            I would assume the Intel exec wouldn’t have any knowledge about MS OS design, and as such, anything running the Windows userland and booted as “Windows” would be Windows to him. It could be running the NT kernel or something else entirely.

            Nvidia engineers figured out they would overtax the GPU trying to run FPU and graphics code at the same time, and relying on the less powerful ARM FPU as a back up would make them look like chumps, maybe?

      • TaBoVilla
      • 8 years ago

      sorta, yeah

      • Jahooba
      • 8 years ago

      Your mean-spirited demeanor towards Intel and Windows is simply childish, and no way relevant to the subject matter.

        • Da_Boss
        • 8 years ago

        It’s not mean spirited to call a spade a spade. This whole Windows on ARM/ARM W8 won’t run x86 apps just reeks of desperation to me.

        For two companies so capable, I’d expect better.

          • sweatshopking
          • 8 years ago

          I guess you’d say the same about the car companies making electric or hybrid cars.

            • Da_Boss
            • 8 years ago

            Yes and no. There’s so much more involved in why those haven’t taken off yet.

            I see what you were going for, though…

      • WaltC
      • 8 years ago

      Anyone get the sense that most people these days are so steeped in marketing lore and crass jingoism that they wouldn’t recognize a fact if it reared up and with shiny sharpened teeth took a huge pound of flesh from their exposed buttocks? I mean, does anyone even understand how much slower, how much incredibly slower, an ARM cpu is at processing data compared to a similarly clocked AMD or Intel x86 cpu? Does anyone even understand these days that the purpose of x86 cpu development is gobs and gobs of performance per clock, while the purpose of ARM is gobs and gobs of low power consumption per clock, performance be damned? Does anyone even understand the difference anymore?

      Does anyone understand that ARM cannot–let me rinse and repeat lest there be any doubt–that ARM cannot, [i<]simply cannot[/i<] take the place of x86 in today's or tomorrow's market? Better yet, does anyone *not* understand that very, very clearly? Does anyone not understand that contrary to being a new and shiny and sparkling technology in 2011, ARM saw its first incarnation [b<]1983[/b<]? [url<]http://www.ot1.com/arm/armchap1.html[/url<] , and that new incarnations of ARM are pretty much just taking that "old ARM turd" and polishing it? (Sorry for that slight--but I think the owner of the post above deserved it.) Does anyone not know that the highest and best use for ARM processors is that of [i<]low processing power[/i<] (contrasted with x86) for use in [i<]mobile devices[/i<] where the low computational requirements meant low power designs were feasible, which meant in turn that longer battery life could be expected? Does anybody reading this thread know that although Apple has adopted the "old turds of embedded ARM cpus" for use in its low-powered, mobile, high-battery-life devices, that Apple has no plans--ever--not in a million years--of removing Intel x86 from its Mac lineup of personal computers, to turn them into a lineup of products little better than [i<]souped-up pocket calculators[/i<] running ARM cpus--but with *gre-e-e-e-e-at* battery life!? I think everybody knows that this isn't going to happen at Apple or anywhere else in the world for that matter. Yes, everybody knows that ARM is nowhere in the Mac's future--but the more troubling question is "Does everyone know why?" Apparently not. To these people a cpu is a cpu, and a MHz is a MHz, and "architecture" be damned--what's that mean anyway--"'R'keetekture"?--Sounds fishy, don't it?" (Right? Ri-i-i-i-i-ight.) Sigh...why is it so predictable that you can count on it like a sundial on a clear day...why is it that every time Apple decides to use a product that nobody who has used an Apple product before has ever heard of--that suddenly that product is reinvented to become brand-new again and is sure and certain to sweep everything else on earth away in a technological tsunami? How could Apple proponents constantly keep getting these things wrong, over and over again? Tell you what, though--let's make it fair to these folks. Let's at least wager that when Apple takes the x86 out of the Mac than ARM probably isn't too far behind--uh, but that's assuming that when Apple takes the x86 out of the Mac that Apple has decided to [i<]keep on making[/i<] a Mac...;) And talking about "polished turds"--well, Mac OS is as polished as any of them, having supported MC68xxx cpus, then PPC's of several different flavors, and the the true forerunner of OS X--the NeXT OS which was strictly an x86 operating system--for years before it was ever known as OS X and sold in a Mac. Indeed, it should therefore come as no surprise to anyone that OS X today is--drumroll--why nothing less than that polished turd of x86, all over again. The fact of the matter is that if Microsoft and Intel's current products are "polished turds," then Apple sells as many polished turds as anyone else...;) But one more set of facts before I blessedly go: Today's Intel/AMD x86 cpu designs bear such scant resemblance to their for-bearers that they are literally "x86" in name only, and today's Windows7 OS bears such little resemblance to DOS/Win 1.x that the two versions of Windows are, practically speaking, [i<]incompatible with each other.[/i<] And last but not least, Apple has no plans whatever to replace its x86 Macs with ARM-powered Macs, because that would completely remove the Mac from the pack as a competitive consumer p0erformance-based computer. The ARM cpu devices are niche devices filling a specific niche segment concerned primarily with mobility and long battery life. That niche will never change or go away. But x86 cpus are becoming much faster, too, as time goes on, and if anything they are steadfastly increasing the performance gap they currently hold over Arm in the current market. IE, if we say that the general performance of Arm = x; the general performance of x86 = x (squared) * Y ...;) Besides the overwhelming performance advantage x86 holds over ARM, economies of scale and manufacturing muscle must not be forgotten, either. Today, with both AMD and Intel approaching incredibly, nearly unbelievable process nodes (.22n and <), and with the strides both companies are making in controlling leakage and gate efficiency, x86 is primed to eventually be nearly as energy efficient as ARM while delivering dramatically greater processing power at the same time. Increasingly, ARM will have its work cut out for it just trying to stay in the game. So if you want to call all of this "turd polishing," be my guest. It's the kind of "turd polishing" that keeps the world right on spinning ,though...

        • Flatland_Spider
        • 8 years ago

        One little point…

        NeXTSTEP started out on Motorola RISC chips. NeXTSTEP was later ported to x86 and a host of other processors, so it still wasn’t pure x86. At that time, NeXT had plans to release a new NeXT station using a PPC chip, but they decided to shutter their hardware division.

        NeXT only ported NeXTSTEP to x86 because they were desperate to get sales.

      • travbrad
      • 8 years ago

      Intel is such a relic that they just had their highest profits ever. Poor Intel.

        • Firestarter
        • 8 years ago

        High profits don’t necessarily imply a healthy future, mind you. If I recall correctly, Blackberry posted record profits the last year or two because of their sudden popularity in the consumer market. Their future in that market is unclear at best though, others say they’re pretty much doomed because of competing smartphones.

        Not to say Intel is doomed though, I just assembled a computer with 63% of its budget invested in Intel parts 😮

      • Anonymous Coward
      • 8 years ago

      Microsoft is huge and relevant in business applications, the way things are going Windows is going to be a side business.

    • flip-mode
    • 8 years ago

    Meh, this could be a positive in the long run. I have no idea how. Perhaps apps will end up getting ported to virtual environments like .Net and JVM or something.

    Perhaps Windows 8 ARM applications won’t overlap with Windows 8 desktop applications at all – maybe they’ll be just for light weight devices or for low power server environments.

    My feelings would not be hurt if x86 eventually bit the big one.

    • boing
    • 8 years ago

    Sounds like a marketing decision more than a technical reason. Couldn’t they easily (relatively speaking) port the emulator-layer used in old NT for Alpha/MIPS/PowerPC? Or what they use in Windows for IA-64?

      • Scrotos
      • 8 years ago

      No, you’re thinking of the HAL, hardware abstraction layer. Programs still had to be compiled for the specific architecture, if memory serves. And it doesn’t matter anyway, MS killed HAL in XP/2K because in NT4 it also abstracted the video which at the time made the OS really slow for playing games. You’ll find that DEC had a dynamic translator ala Rosetta on the Mac called FX!32 which allowed WinNT for Alpha to run, say, Microsoft Office and used some recompile/optimize/store the code for later so it would perform better on subsequent runs. There were some non-x86 native versions of Win2K beta but I think in the final release only Alpha was supported.

      So then video drivers could write directly to the video hardware and bad drivers could crash the OS. Fast forward to Vista/Win7 and I think MS once again did some HAL-like thing with the video. I’m not sure of the extent of it, though.

      It would probably make better sense to adopt something like Rosetta from Apple which is actually some licensed IBM product, I think. Apple did pretty good with the PPC/x86 platform support with “fat” binaries/packages and Rosetta for non-native binaries. Too bad they are ditching that for OS 10.7 as I would have thought they’d keep it in some version for ARM/x86 at the very least. Honestly I am not too familiar with how Apple supports ARM and x86 for one program, if at all.

        • Flatland_Spider
        • 8 years ago

        The HAL is still there.

        “Hardware extraction layer (HAL) independence. Retail versions of Windows Vista are HAL-independent.”
        [url<]http://technet.microsoft.com/en-us/library/cc507856.aspx[/url<] Windows Vista split the graphics stack into a kernel component and userland component. "At a technical level, WDDM display drivers have two components, a kernel mode driver (KMD) that is very streamlined, and a user-mode driver that does most of the intense computations. With this model, most of the code is moved out of kernel mode. That is, the kernel mode piece is now solely responsible for lower-level functionality and the user mode piece takes on heavier functionality such as facilitating the translation from higher-level API constructs to direct GPU commands while maintaining application compatibility. This greatly reduces the chance of a fatal blue screen and most graphics driver-related problems result in at worst one application being affected. WDDM also provides fault-tolerance against display driver hangs. This enables Windows Vista to detect system hangs and restart the display driver again without the need of a system reboot. Additionally, display drivers in Windows Vista have been significantly simplified by eliminating the need to include code for the support of various device driver interfaces introduced over many years. Thus, Windows Vista implements only a single interface while ensuring that all the older drivers are recognized and function optimally. " [url<]http://msdn.microsoft.com/en-us/library/aa480220.aspx[/url<]

        • boing
        • 8 years ago

        Actually, reading what you wrote I realize it was FX!32 I had in mind. 🙂

    • bdwilcox
    • 8 years ago

    Oh noes! Most users’ applications like GMail, Twitter, Facebook, MySpace, et al. can ONLY run on x86 chips. Plus, most people’s software like iTunes, Office, Quicken, et al. are only good in their x86 versions and people will never upgrade, even if it’s free!

    Intel is scared and should be. The cloud has disarmed much of their FUD and few consumers run legacy apps anymore. Sure, the corporate marketplace will be tied to x86, but the consumer space couldn’t care less about ‘Intel inside” anymore.

      • Jahooba
      • 8 years ago

      Intel is not scared and will do fine. Tablets and smartphones may trample on the toes of desktops, but for those that know/ care what they’re buying, Intel is trusted and proven.

      Your perceived ambivalence that consumers have towards what’s in their machines only highlights the dumb consumer perspective that has no idea what they’re buying and only look at the price tag. While these people do exist, that represents only one demographic. Smart consumers/ companies/ organizations that know what they’re buying will absolutely choose Intel for a quality computing experience.

        • Corrado
        • 8 years ago

        Most corporations don’t blindly choose Intel. When a company spends millions of dollars on equipment, you think they just choose Intel by default? There are a LOT of cost analysis studies done on LOTS of different solutions before one is chosen. If ARM is competitive in performance, and superior in cost, maintenance and power consumption, who do you think they’re going to go with? Hell, being ARM doesn’t really mean anything in the server space if theres a proper virtualization hypervisor for it. Our servers don’t know they’re being run on what they’re being run on. They’re just VMWare machines on an ESXi cluster. If ESXi ran on ARM and offered enough performance, we’d certainly consider it.

          • sweatshopking
          • 8 years ago

          I don’t agree with you. I’ve worked for a number of MAJOR corporations and companies that bought Intel servers, cause that’s what the sticker was on the old ones.

            • bdwilcox
            • 8 years ago

            And I was talking about desktops and laptops anyway. Try porting 200 validated apps off of native x86 and onto ARM and then tell me about cost savings. And even if you use an emulator to run the native x86 apps on an ARM machine, the emulator and app have to be validated anyway.

            • sweatshopking
            • 8 years ago

            even with desktops and laptops. we’ve had no “anything we’ve never used before, and we’ve only ever used intel” policies. it’s all the same.

            • Corrado
            • 8 years ago

            Then those people should be at least reprimanded if they didn’t do proper research before buying. I’ve built out projects for Comcast, Accenture, the Pennsylvania State Government, Salary.com, BrassRing, Carpenter Technologies, and a few other companies. If you purchase Comcast Business Class email, thats all Opteron. Accentures email servers are all Opteron. Salary.com is all new Xeon’s in its own ESXi cluster virtualized. Every place I’ve done work for, they all want a complete justification of why you are purchasing what you are purchasing and that includes power, heat, performance, price and form factor in the decision.

            • sweatshopking
            • 8 years ago

            they did the research they were “supposed” to do. what worked in the past? nobody wants to risk a change in the slightest. I’ve worked with leon’s furniture, subway, etc. and it’s not only them. they’d rather pay more, and get what they know HAS worked.

        • mutarasector
        • 8 years ago

        re: “Your perceived ambivalence that consumers have towards what’s in their machines only highlights the dumb consumer perspective that has no idea what they’re buying and only look at the price tag. While these people do exist, that represents only one demographic. Smart consumers/ companies/ organizations that know what they’re buying will absolutely choose Intel for a quality computing experience.”

        The $64K question is just what the >ratio< of smart/dumb, cheap customers will be. On that score, I guess I have to go with the old addage that says ‘the masses are asses’.

        Translation: ‘Wintel’ may become ‘WARM’.

        As strange as this may sound, x86 Windows platforms may become ‘esoteric’, and jump onto OS X’s greener lawn.

          • OneArmedScissor
          • 8 years ago

          That assumes anyone who buys an ARM-based Windows 8 computer is dumb and cheap…in 2013, when Windows 8 will show up paired with over 9,000 core Tegra 5s with frickin’ laser beams attached to their system bus.

            • NeelyCam
            • 8 years ago

            [quote<]frickin' laser beams attached to their system bus.[/quote<] Yeah, like ARM doesn't have enough problems keeping power consumption in check while trying to up the performance...

            • OneArmedScissor
            • 8 years ago

            Hook, line, and sinker!

    • bthylafh
    • 8 years ago

    I’m more interested in this “windows 7 compatibility mode”. Are they finally pulling an Apple and going with a new API/ABI?

      • bhtooefr
      • 8 years ago

      From the rumors I’ve seen elsewhere, they’re basically doing a new .NET-based API/ABI. So, no native programs any more, it all runs in the VM, and runs on any architecture.

    • dpaus
    • 8 years ago

    A Windows OS with no apps fighting multiple well-entrenched competitors?

    Winning!! (in the Charlie Sheen sense, that is)

Pin It on Pinterest

Share This