Become a fan of Slashdot on Facebook

 



Forgot your password?
typodupeerror
Emulation (Games) Microsoft Intel Windows Hardware

Intel Fires Warning Shot At Qualcomm and Microsoft Over Windows 10 ARM Emulation (hothardware.com) 197

MojoKid quotes a report from HotHardware: Qualcomm and Microsoft are on the verge of ushering in a new class of always-connected mobile devices that run full-blown Windows 10. The two are enabling ARM-based Snapdragon 835 processors to run Windows 10 with full x86 emulation, meaning that devices will be capable of not only running Universal Windows Platform (UWP) apps from the Windows Store, but legacy win32 apps as well. There is little question, Intel is likely none too pleased with it and PC OEM heavyweights Lenovo, Hewlett-Packard and ASUS have also signed-on to deliver Windows 10 notebooks and 2-in-1 convertibles powered by Qualcomm. Until now, Intel sat by quietly while all of this unfolded, but the company today took the opportunity to get a bit passive-aggressive while announcing the fast-approaching 40th anniversary of the world's first x86 microprocessor. The majority of the press release reads like a trip down memory lane. However, Intel shifts into serious mama bear mode, with significant legal posturing, touting its willingness to protect its "x86 innovations." Intel goes on to say that Transmeta tried and ultimately failed in the marketplace, and has been dead and buried for a decade. The company then pivots, almost daring Microsoft and Qualcomm to challenge it by making Windows on ARM devices commercially available. "Only time will tell if new attempts to emulate Intel's x86 ISA will meet a different fate. Intel welcomes lawful competition... However, we do not welcome unlawful infringement of our patents, and we fully expect other companies to continue to respect Intel's intellectual property rights."

Intel Fires Warning Shot At Qualcomm and Microsoft Over Windows 10 ARM Emulation

Comments Filter:
  • Can go either way (Score:5, Insightful)

    by Luckyo ( 1726890 ) on Friday June 09, 2017 @06:21AM (#54583661)

    Between the fact that current ISA is actually AMD64 (which is x86 compatible, but not intel-designed) and the fact that many key patents should have expired by now, it's going to be interesting if intel has the legal bass to actually stop this from happening.

    • Re:Can go either way (Score:5, Interesting)

      by gweihir ( 88907 ) on Friday June 09, 2017 @06:25AM (#54583671)

      As far as I remember, emulation is not covered by the patents anyways, as an instruction-set is basically an API. Does anybody know whether this is accurate?

      • It really depends on how they implemented the emulation. There is more to emulating a processor than just re-implementing the instruction set. If they took the instruction set and re-implemented it from scratch, adding new implementations of caches, memory management etc and didn't copy the actual internal processes of the chip there shouldn't be a problem with patents however there may be issues with compatibility as there will likely be issues where the emulation differs from the hardware (for example o

    • Re:Can go either way (Score:4, Informative)

      by Carewolf ( 581105 ) on Friday June 09, 2017 @07:18AM (#54583829) Homepage

      Between the fact that current ISA is actually AMD64 (which is x86 compatible, but not intel-designed) and the fact that many key patents should have expired by now, it's going to be interesting if intel has the legal bass to actually stop this from happening.

      Yeah. Even AMD64 is almost 15 years old by now. They might have patents on some newer additions (maybe no AVX emulation?), but what can still be patented about x86?

      • by Guspaz ( 556486 )

        Microsoft/Qualcomm aren't emulating AMD64, though, they're only supporting 32-bit x86 apps. If they want to avoid any patents at all and went back 20 years, they might run into some compatibility issues, because 1997 puts them before every single marketed instruction set extension: MMX was introduced in 1997, so getting in just before that, you're basically back to the 486 or early pre-MMX Pentium 1.

        The question is, how much modern Windows software is ABI compatible with the 486, and doesn't assume MMX or S

        • The question is, how much modern Windows software is ABI compatible with the 486, and doesn't assume MMX or SSE support?

          Probably not much. In fact, Windows 10 will not run on any x86 processor that doesn't support PAE, NX and SSE2, and that's been the case since Windows 8.

        • The absolute minimum requirement for Windows XP was a plain old Pentium, with neither MMX or SSE.

          But you asked 'modern', so one might assume that software written in the last 15 years would target something newer.

        • by Agripa ( 139780 )

          The question is, how much modern Windows software is ABI compatible with the 486, and doesn't assume MMX or SSE support?

          A lot now requires at least SSE2 whether it uses it or not so the minimum emulation target is the Pentium 4. This has become a major problem with my legacy Pentium 3 systems which have plenty of performance but are no longer supported.

      • In the medical world companies file similar enough patents to confuse government examiner's basically getting a 25 year extension. Not sure if Intel has done this or not.
    • Re: (Score:3, Informative)

      by Anonymous Coward

      The other hilarious part of this is that Intel shipped ARM emulation code for x86-based Android devices for a few years before totally failing to gain any traction and giving up.

      • Well, that situation is different - Intel is a licensee of ARM. Remember the StrongARM / Xscale chips? When they bought DEC, they got DEC's license.

    • by PPalmgren ( 1009823 ) on Friday June 09, 2017 @08:01AM (#54584001)

      Honestly? I'm more interested in what Intel's legal bass tastes like. Is it more like a sea bass, or one of those skanky lake bass? This is important stuff.

    • Legal? Nah. FUD is the message here, didn't see that? Basically they're saying "This cannot fly. Look at Transmeta. See? We left them in the dust. And this, this HAS to go the way of the Dodo too!"

      They're trying to badmouth it, not fight it with patents.

    • by alexhs ( 877055 )

      it's going to be interesting if intel has the legal bass to actually stop this from happening.

      If they don't, they can always attempt the shark with frickin' laser beams attached to its head.

    • Between the fact that current ISA is actually AMD64 (which is x86 compatible, but not intel-designed) and the fact that many key patents should have expired by now, it's going to be interesting if intel has the legal bass to actually stop this from happening.

      While that's true, what is being emulated is 32-bit x86, not (just) the 64-bit x64. And that is something Intel owns. Of course, if they've expired, that's another story.

      But I see this move as Intel punching downwards. An ARM has no chance of doing a good job emulating x86. What might happen is that x86 software will be emulated on ARM, but then software writers - at least the ones whose software is popular on this platform - would port their software to be ARM native: it can still sleep in Microsoft'

      • No. No, they won't. If I have to port software that old to ARM, I might as well port it to Linux or Android.

        • by caseih ( 160668 )

          Don't know why you think that. Nearly all Windows applications are "legacy" applications right now. Meaning Win32 api. If devs could just select ARM from the Visual Studio target drop-down box and hit go, they'd be all over that I think. Maybe not things like Adobe products, unless they see a big enough market. But for things like office applications such as MS Office, Quicken, Chrome, Firefox, etc (the apps people actually use vs Windows Store), you can be that if VS supports the ARM target, they will

  • How fast can Snapdragon processors run Windows software? I'm sure that productivity software - Word, Excel, PowerPoint, OpenOffice Acrobat Reader, Edge, Firefox, Chrome et cetera - will run just fine. That stuff doesn't need huge CPU power to begin with. What about something more CPU-heavy like Adobe Photoshop, AfterEffects, AutoCAD, 3D Max,Blender, Handbrake? How fast will that software run on Snapdragon? Of course this is no big problem - if ARM can't run it today, you can always run it on an Intel or AMD
    • by Anonymous Coward on Friday June 09, 2017 @06:37AM (#54583703)

      Well, here are two demos of x86 Win32 emulation in Windows 10 on ARM:

      1. A shorter, more marketing orientated demo: https://www.youtube.com/watch?v=A_GlGglbu1U [youtube.com]
      2. A longer, more "show me stuff" orientated demo: https://www.youtube.com/watch?v=oSXUDKpkbx4 [youtube.com]

    • How fast can Snapdragon processors run Windows software? I'm sure that productivity software - Word, Excel, PowerPoint, OpenOffice Acrobat Reader, Edge, Firefox, Chrome et cetera - will run just fine. That stuff doesn't need huge CPU power to begin with. What about something more CPU-heavy like Adobe Photoshop, AfterEffects, AutoCAD, 3D Max,Blender, Handbrake? How fast will that software run on Snapdragon? Of course this is no big problem - if ARM can't run it today, you can always run it on an Intel or AMD box. But the question remains - how fast is emulated Windows on ARM?

      It won't be winning any benchmarks. The base processor starts off slower than most x86 processors, and then has an overhead for emulation. I am guessing it is mostly just for old applications, and not for games or anything CPU intensive, and for that it might be enough.

      • LLVM can compile ARM to x86. CIL is compiled to the current platform ISA. Why wouldn't you JIT, instrument, optimize?

        • LLVM can compile ARM to x86. CIL is compiled to the current platform ISA. Why wouldn't you JIT, instrument, optimize?

          You do, but it works about as well as translating through a third language in google translate. Not always bad, but not great either.

          • You do know all modern compilers convert program source code into a static single assignment tree which is then optimized by simplifying complex graph traversals with weighted edges based on CPU architecture variations in speed of certain operations, right?

            You can turn CPU instruction code into the same static single assignment tree, perform the same optimizations, and emit the code as if it were compiled from source code to target a different ISA.

            • You do know all modern compilers convert program source code into a static single assignment tree which is then optimized by simplifying complex graph traversals with weighted edges based on CPU architecture variations in speed of certain operations, right?

              You can turn CPU instruction code into the same static single assignment tree, perform the same optimizations, and emit the code as if it were compiled from source code to target a different ISA.

              Which works great if you have that intermediate code, but once it becomes instructions they often do a lot more than what is necessary, and unless you know the original intention or can analyse it from the entire program you might have to fully emulate every nuance of the instruction instead of just using the one simple instruction in the other ABI that perform the same intermediate operation but has slightly different nuances in the other ABI.

              • Which works great if you have that intermediate code, but once it becomes instructions they often do a lot more than what is necessary

                No, that's not how this works.

                The compiler takes the source code, produces a single-assignment transformation tree, and then discards the original source code. All further analysis is performed on the single-assignment transformation tree. That means 100% of the information a compiler uses to compile a C, C#, Java, Ada, Fortran, JavaScript, etc. program to CPU ISA output in any setting is represented by a tree derivable in reverse from and functionally-equivalent to a tree derived from the output of com

                • Actually, analysis of the compiled output will tell you things you can derive into a static single assignment tree. For example: if you run a subtraction on x86, it sets the Flags register, which affects branches. If another operation affecting the Flags register occurs before any sort of branch instruction, this is irrelevant, and thus not recorded into the tree. By extension, knowing how such instructions are affected by prior instructions allows you to create a full listing of the logic at each node, such that you can easily understand the conditions for a jump (e.g. if you do SUB %eax,4 and then JLE @addresss, you don't need to encode into the SSA tree that SUB may set the zero flag, signed flag, and overflow flags; you only need to encode that, under the condition %eax is less than or equal to 4, the program takes a branch).

                  Can you do that in real-time, in an interpreter?

                  Additionally unless you are sure you can guess which jumps are function calls and returns and you are certain no link-time optimizations have been performed and that the calling convension is still obeyed, you might not even know which registers need to be alive unless you do full analysis of everywhere a jump can go or return to. Of course you can often derive that data from a full analysis (though not always, since you can't cover every possibly state withou

    • by stooo ( 2202012 ) on Friday June 09, 2017 @07:49AM (#54583939)

      >> What about something more CPU-heavy like ... Blender, Handbrake ...

      For those ones, you don't need x86 emulation, as you can use an ARM compiled version, the source is available.

      Fot the other proprietary stuff, perhaps some will be released in an ARM binary form, especially if ARM laptops and even desktops become popular in the future ....

    • If someone buys a low power device and expects to run AutoCAD on it, then they get what they deserve. Anyone who actually uses those high-end software titles will know better, and steer clear of these low-power devices that are "yet-another-iPad-killer."

      The only thing that has proven to be an iPad killer, is the iPad itself - either because new versions don't offer anything compelling over versions already purchased, or because the entire class of device has proven to not be useful outside of a somewhat na

      • by Khyber ( 864651 )

        "If someone buys a low power device and expects to run AutoCAD on it, then they get what they deserve."

        I remember using AutoCAD on a 266MHz PII back in high school drafting class.

        What's the excuse for it running like dogshit now?

        • Today's AutoCAD does many more things that require vastly more computational power than the old version you used in high school?

    • If the OS maker also supplies the emulation they can have all their code native, including apps. Emulation is at best half speed, and ARM chips are not exactly top of the line to start with, so things are going to be slow. Current desktop systems are much faster than is required by most software, so the slowdown is unlikely to be noticed for most programs. You'll still want your games running on Intel or AMD though. Most SW shops will release dual ported (both processors) software just like apple did wh
    • by AHuxley ( 892839 )
      Depends on the task. For everyday OS and simple productivity apps? Fine.
      For a new 4K, 8K ready computer game?
      Can a developer spread a new game's tasks over a new cpu and have it still work as well as expected?
      Can every task be split up and shared over a new average cpu design? Or do some tasks need a fast cpu?
      New programming language? New ways of working with content? New ways of working with a cpu, gpu?
      If a game developer gets all that for free they may change.
    • All the microsoft software, like Office, Edge, et al would have been recompiled. I'm sure the same would be true about FireFox and Chromium (if not Chrome), since those are FOSS. Question is whether those other titles that you listed will be recompiled, and whether Microsoft would allow them if they're not available via the app store.

      In Windows 8, if one wanted to install Microsoft Office and went to the app store, it redirected them to the website and handled things from there. Microsoft could do t

    • Also, has anybody tried running Windows 10 IoT on Raspberry Pi? Or is that not out as yet?

  • by OpenSourced ( 323149 ) on Friday June 09, 2017 @06:40AM (#54583711) Journal

    ...the writing on the wall?

    The future is mobile, and in mobile RISC wins (for now). First you emulate, then you go native. Microsoft has seen the new Samsung S8 working as a desktop replacement with a dock, and is sweating cold. If that trend goes on (and why shouldn't it, as phones become more powerful), and mobile apps adapt to the "desktop mode", soon Windows will have a real competitor. I can see plane stewardesses distributing keyboards to the passengers, so as to use the entertainment screens with your "desktop mobile". I can see "laptops" that are just a screen, a keyboard, a humongous battery and a dock bay for your phone.

    You cannot fight the tide. In three years smartphones will ship with 1 Tb storage, 16 Gb memory, and 16 cores CPU. All of them itching to do something more demanding than displaying your last photo of your cat. If Microsoft doesn't emulate in firmware, VMWare will emulate in software, and soon you won't care in which OS is your app working. You will have one and only one computer, that will incidentally have the capacity of making phone calls. Congratulations everybody, we are just now entering the era of the PC.

    • by Dwedit ( 232252 )

      And they won't be able to do anything beyond showing photos of your cat with that horrible touchscreen.

    • I can see "laptops" that are just a screen, a keyboard, a humongous battery and a dock bay for your phone.

      Meet the superbook [kickstarter.com]

    • It's not at all clear that RISC wins in the mobile space. Intel's offerings win in performance per Watt benchmarks by a factor of 2-5, they just haven't hit the minimum Watts that ARM processors have.

      The problem for Intel is price. Current ARM processors have about 2-3 billion transistors and sell for about $10-$20. Kaby Lake has about the same number of transistors, but sells for $100-$300. Intel has enjoyed that huge price per transistor for so long that they simply don't know how to compete at a l
    • No. No matter how much storage you put on the phone, you are still dealing with a device that can, at max, use 25 watts. Mobile is not going to replace workstations, ever.
  • by Zuriel ( 1760072 ) on Friday June 09, 2017 @06:41AM (#54583721)
    I have heard that Intel is delighted about AMD's new 32 core server CPU and 16 core desktop CPU. If I remember correctly, the last time Intel was this delighted with AMD they started bribing system manufacturers into not offering any AMD-based products. [wikipedia.org]
  • by ChunderDownunder ( 709234 ) on Friday June 09, 2017 @07:12AM (#54583799)

    Proposition: Apple are planning to release low end macOS products based on their ARM64 SoCs.

    Imagine having a common ARM-based hardware spec for Watch, Mac Mini, iPhone, iPad, MacBook and iMac - only select 'Pro' models would require Intel Inside.

    Such a transition would require checking an extra box in XCode for fat-binaries. Optionally they could develop a Rosetta-style translation layer for 'legacy' amd64 only binaries.

    • This is actually a viable suggestion. The only people that truly need processing power are video gamers and content producers (e.g. artists, developers).

      Considering there are a lot of people out there that still think the internet is that E icon (or fox, or yellow/red/green circle) and only really need to check email, facebook, youtube and cat pictures... ARM tablets have been gaining steam, why wouldn't they gain some traction in more traditional forms like a laptop a laptop-like phone dock?

    • Apple customers have already played the "buy a new version of the same software at full price or eat a performance penalty" game twice now (680x0 -> PPC, PPC -> x64). While Apple has magnificently pulled off these transitions, their software partners have in large part been amazingly douchey about it. Especially the software vendors that most Mac shops depend on the most: the creative and business titles like Adobe, Microsoft; and back in the day, Quark.

      Quark was the WORST, and quite frankly I'm sur

  • Lawyers are the last argument of technology companies that have lost the power to innovate. Intel wouldn't be saying these things if they had invested in technology rather than idiotic purchases of horrid anti-virus companies, bizarre offshore strategies, and generally letting wall st. run them rather than engineers and technologists.

    • by Zocalo ( 252965 )
      No, there's one more step, it's just that we haven't got that far yet:

      Soap box - marketing lies about which is the best product.
      Ballot box - the market chooses the best product.
      Jury box - lawyers bicker over which product has the biggest dick^W patent portfolio.
      Ammo box - two products enter...

      I hope to get tickets but, failing that, I'm sure it'll be livestreamed in 8K HDR Augmented Reality to any suitable mobile computing device.
  • Intel welcomes lawful competition, and we are confident that Intel's microprocessors, which have been specifically optimized to implement Intel's x86 ISA for almost four decades, will deliver amazing experiences, consistency across applications, and a full breadth of consumer offerings, full manageability and IT integration for the enterprise. However, we do not welcome unlawful infringement of our patents, and we fully expect other companies to continue to respect Intel's intellectual property rights.

    I r

  • The problem here is that Intel has no legal basis to threaten anyone over x86, so instead they are just blowing smoke to scare away investors using the press. Both side know that Intel's x86 patents have expired and AMD owns all the patents for x86_64 which is also known by it's original name AMD64. Since Intel is now powerless, they are reverting back to their old anti-competitive habits.

  • 1. Intel wants into the mobile market. 2. Intel's attempts with Atom to get into the mobile market go nowhere. 3. Microsoft wants into the mobile market. 4. Microsoft's attempts with Windows RT to get onto ARM architecture go nowhere. 5. Intel can't beat ARM on cost. 6. The Wintel Monopoly. 7. The Wintel Monopoly... no more! 8. Shrinking PC shipments. 9. Android surpasses Windows as world's most installed OS. 10. Windows on ARM. 11. Will that work? Battery life? Performance? Experience? 12. What will Intel

  • If this catches on, I'm looking forward to the flood of cheap Windows 10 tablets that will have support dropped after two years because the BSP for their SoC is no longer being updated or maintained.

    Granted, I expect Microsoft to do marginally better with Windows on ARM than Google does with Android because Microsoft has much tighter control over the Windows ecosystem. But at the end of the day, if you want to play in the ARM playground, you're going to get burned by short chip life cycles.

    On the other hand

  • There is little question, Intel is likely none too pleased with it and PC OEM heavyweights Lenovo, Hewlett-Packard and ASUS have also signed-on to deliver Windows 10 notebooks and 2-in-1 convertibles powered by Qualcomm.

    Take that, every English teacher you've ever had from grade 3 onwards.

    In another strip, after stating that he could not identify Plymouth Rock, lest it "compromise our agents in the field," Calvin cheerfully remarks, "I understand my tests are popular reading in the teacher's lounge".

    There's

  • "In the early days of our microprocessor business, Intel needed to enforce its patent rights against various companies including United Microelectronics Corporation, Advanced Micro Devices, Cyrix Corporation, Chips and Technologies, Via Technologies, and, most recently, Transmeta Corporation. Enforcement actions have been unnecessary in recent years because other companies have respected Intel’s intellectual property rights."

    Actually, enforcement became unnecessary as most were driven out of the chip

1 1 was a race-horse, 2 2 was 1 2. When 1 1 1 1 race, 2 2 1 1 2.

Working...