Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
Microsoft Intel

'Welcome To the PC Malaise Era' (getwired.com) 271

Long-time research analyst Wes Miller, who previously worked at Microsoft, believes that Windows-powered PCs are now stuck in the same rut that American cars grappled with in 1973 to 1983. He writes: I've said before that Windows has never escaped x86. I'm still not sure if it ever can. So the challenges then come down to three things:

1. Can Intel succeed where they've failed for the last 5+ years, at building hybrid processors? The next year to two years should answer this question.

2. Can Microsoft succeed at finally getting application developers to write platform-optimized, energy-respectful, halo applications for the PC? I've been writing about the Windows Store for a long, long time. A long time. And I'm still not sure how Microsoft can light a fire under Windows application developers when they've lost that mindshare.

3. Can Microsoft begin pushing the Surface platform forward again? This one's completely up to Microsoft. I've seen the rumors of the next Surface Pro... and it's more of the same -- evolutionary, not revolutionary.

I guess we will see in the next 3-5 years whether Intel can cross this chasm; if they can't, then the future likely belongs to ARM, and that future will likely mean less and less to Microsoft, outside of running classic Win32 applications on x64/x86 Windows.

This discussion has been archived. No new comments can be posted.

'Welcome To the PC Malaise Era'

Comments Filter:
  • by chill ( 34294 ) on Monday November 30, 2020 @03:09PM (#60779490) Journal

    So, what does "revolutionary" actually look like for a PC? Would it be recognized as a PC?

    I don't know, but I'm fairly certain that it has almost nothing to do with the CPU hidden under the hood. The answer isn't Intel vs ARM vs ... anybody. That's a technical quibble that is being shown every day to mean less and less. Apple's evolution from Motorola 68x to PPC to Intel to ARM should have made that part obvious.

    • by MightyMartian ( 840721 ) on Monday November 30, 2020 @03:23PM (#60779558) Journal

      At the end of the day, there's a helluva lot of legacy software out there that outright breaks when Windows runs on a new architecture. Some of that is software developers' faults; they've been developing on the x86 ecosystem in one form or another for the better part of 40 years (even longer if you consider the continuum of 8080/Z80 to 8086). But it's worse than even that. I'm still forced to use a web application that was optimized for Internet Explorer in the early 2000s but is only partially usable even now on other browsers, and there's a huge amount of software built on top of the IE ecosystem that is likely to break in obvious or non-obvious ways; so that even Edge now has to have an "IE mode", meaning more cruft thrown into a major piece of Microsoft's modern software stack.

      The fact is the NT kernel was built from the ground up to be portable, and has run on a number of architectures over the last 30 years. But that doesn't mean much when a lot of the software is compiled to x86 binaries. Developers have never really bought into .NET, so even that hypothetical route to making software architecture independent hasn't been bought into by many developers.

      Apple's solutions of periods of being able to run old architecture binaries has been accomplished in no small part because Apple exerts a helluva lot more control over its architecture. It builds its computers, it builds its operating systems. Microsoft doesn't have that level of control. I can go on Amazon or down to Best Buy or whatever and buy one of dozens of PCs, some running Intel, some running AMD, all with different chipsets and video cards. Microsoft got huge by building an OS that could run on PC compatibles, and it has always been a moving target that I do give MS some credit for managing, somehow, to largely stay on top of, but that strength is also a weakness that means any transition to new architecture is going to come at far larger cost for developers.

      Java was developed to get over this, Microsoft has tried to sell .NET to the world to untether itself from x86, but so far not so good. It's a pity that Java never lived up to its promise, because CPU speed now is sufficiently powerful that you could hypothetically build an entire software ecosystem that had no idea whether it was running on x86 or ARM, or whatever new thing comes along next.

      • Re: (Score:2, Insightful)

        by brunes69 ( 86786 )

        "Java was developed to get over this, Microsoft has tried to sell .NET to the world to untether itself from x86, but so far not so good. It's a pity that Java never lived up to its promise, because CPU speed now is sufficiently powerful that you could hypothetically build an entire software ecosystem that had no idea whether it was running on x86 or ARM, or whatever new thing comes along next."

        That ecosystem exists... it's now called the web browser... and in the back-end, its running for the most part on a

      • Microsoft should have insisted on multiple architecture binaries back in the days of early Window's NT. That was a critical mistake and something Microsoft has not done anything to remedy in the interim.

        • The size of the binaries would be quite large back in the 90s. Remember that NT 4.0 was running on x86, PowerPC, Alpha, and MIPS. Apple's universal or "fat" binaries were always for two platforms max (68k and PPC, PPC and x86, and now x86 and ARM). Multiarch binaries make sense only if you have a plan to transition to something new and need a stopgap solution.

          I'm actually surprised that multiple ports of RISC NT lasted as long as they did as the number of non-x86 NT applications were near-zero and the mach
          • by jmauro ( 32523 )

            The Alpha ports of Windows lasted for as along as they did, because it was part of the agreement with DEC to keep supporting them after it was clear that the internals of VMS to implement NT, to the point where they probably violated some patents. To avoid lawsuits, they agreed to pay DEC and continue to support Alpha until HP, who eventually acquired the platform, decided to close it and PA-RISC down in favor of Itanium.

        • Maybe they should have, but then again, back in the early days of Windows NT 3.5, there was one dominant 32 bit archtecture; x86. As I recall, they did have ports to PPC and Alpha, but other than Apple, I don't recall a lot of PPC systems out there, and Alpha was promising, but never really went anywhere (I saw exactly one DEC Alpha workstation running NT back in the day, and it was $$$). They did have an opportunity at the very least, but they through DOS VDM support in there, as well as the WOW layer to r

        • by jmauro ( 32523 )

          They did. NT was originally built against MIPS and then ported to x86 and Alpha to prevent programmers from doing any x86 tricks. The whole Hardware Abstraction Layer (HAL) architecture at the core of NT is in part to allow them to support multiple archs without too much rewrite. The problem was the market had zero interest in anything other than x86 clones, since any MIPS or Alpha machines were orders or magnitude more expensive and none of the old DOS software ran on them at least at first. Sam

      • Apple's solutions of periods of being able to run old architecture binaries has been accomplished in no small part because Apple exerts a helluva lot more control over its architecture. It builds its computers, it builds its operating systems. Microsoft doesn't have that level of control. I can go on Amazon or down to Best Buy or whatever and buy one of dozens of PCs, some running Intel, some running AMD, all with different chipsets and video cards. Microsoft got huge by building an OS that could run on PC compatibles, and it has always been a moving target that I do give MS some credit for managing, somehow, to largely stay on top of, but that strength is also a weakness that means any transition to new architecture is going to come at far larger cost for developers.

        I don't think it matters that much that Microsoft doesn't build the hardware and Apple does. The biggest factor is the different philosophy: Apple are willing to abandon legacy APIs and architectures and Microsoft aren't. Switching processors is a little hypothetical, since alternatives to x86 have never been enough better to really force Microsoft to choose whether to support them. But if ARM leaves x86 in the dust, Microsoft could work with PC manufacturers to roll out ARM PCs with an ARM-based Windows th

        • by MightyMartian ( 840721 ) on Monday November 30, 2020 @06:57PM (#60780466) Journal

          Microsoft was writing operating systems for a pretty diversified set of hardware. In the early days of MS-DOS there were specialized variants of DOS for hardware that was a bit left field, but for the most part MS became THE x86 company, creating that somewhat unhealthy Microsoft-Intel hegemony which only really began to fall apart when ARM became the go to CPU for smart phones and related hardware. Microsoft made its fortune by writing an operating system and ecosystem that ran on x86, full stop. That success, particularly in the enterprise world, meant the customers buying a lot of licenses (like large corporations and governments) demanded that they not have to buy a new version of the software with each new release. In other words, the organizations that essentially bankrolled Microsoft wanted to make sure some crazy DOS app written in 1986 on an 8088 or 80286 would still run on a Pentium a decade later.

          This in and of itself isn't a terrible business model. IBM has been making continuous iterations of the 360 line for 55 years or so now, all with the idea that you can still run decades' old programs on new hardware, because when you're spending tens of millions of dollars writing software, you want to make damned good and sure that it's useful life is measured in decades, not years.

          Selling DOS and later Windows to consumers was, as much money as it made Redmond, a sideline. The real cash was selling hundreds of thousands of seats to enterprise purchasing departments, and to do that, they had to guarantee that the software would last. I think MS had hoped that the NT line would give them a similar development cycle to more specialized hardware manufacturers like IBM and Apple, but it never rolled out that way. It means a bit less now since 16 bit software is a thing of the distant past, and if you want to run it that bad, there's always software like DOSEMU. .NET has been around nearly 20 years now, and yet a helluva lot of software is still distributed as x86 binaries. That's momentum, and it's a hard thing to get around. Apple never had that kind of momentum, precisely because their Mac market was sufficiently small with sufficiently high levels of buy in that they could just go "Well, next week we're dumping PPC for Intel...", with some reasonable stop gaps, and moved on. If MS had, say, dumped Windows 3.1 in the mid-90s and went straight to NT without any legacy support, not thouands or tens of thousands, or even hundreds of thousands of users would have been disrupted, millions of consumers from Aunt Mavis just checking her email to major corporations would have had a major nightmare on their hands.

  • Ditch that abominable UI for windows. And stop the telemetry and tying all your crapware together. Then I'll look at what you have to offer.
     

    • by Merk42 ( 1906718 )
      Change the UI? But change is bad! Especially the UI!
      Oh wait, this is an article about Microsoft, so obviously nothing they do will be satisfactory to this crowd.
    • Ditch that abominable UI for windows

      Nooo! If MS attempts to "fix" it, they'll just screw it up in a different arbitrary bleeped-up way. Let's stick with the fuckage we know.

      Instead of reworking the Grand Tree of OS menus, they should instead focus on Bing-ifying options: make it easy to find options using search words. "Sound", "speaker", "can't hear", etc. should all list and link to the Windows "audio" settings. You wouldn't have to remember how to menu-tree-walk to it. Learn from customer feedback to tun

  • Does Microsoft care? (Score:5, Interesting)

    by brunes69 ( 86786 ) <slashdot@nOSpam.keirstead.org> on Monday November 30, 2020 @03:16PM (#60779524)

    The vast majority of Microsoft's revenue now comes cloud - from Azure, and Office subscriptions (which counts as cloud since it is backed by a SaaS in Office365).

    Office subscriptions are in no way tied to Windows either, as it is also the dominant commercial office suite on both Android and Mac OS.

    It's hard to break down exactly how much revenue comes from Windows OEM now, since they lump it into the "other" category along with XBox and Surface and a bunch of other stuff - but the total revenue of that grouping of things was only 33% of Microsoft's revenue in Q4.

    It's pretty clear that Microsoft cares a lot more about it's services growth than Windows. Nadella has actually been saying this publically for years and years, so it shouldn't surprise anyone.

     

    • That's a dangerous road to take.

      Apple's own software will always be better integrated with its own platforms, regardless of how much antitrust lawsuits you throw at them. They will leverage their platform advantage to bundle other software ... they haven't really gotten around to challenging Microsoft's office, but relying on that continuing is dangerous.

      Microsoft had it right with Windows Phone, it really was a necessity for them to stay relevant.

    • by Luthair ( 847766 )
      Microsofts cloud revenue is a fiction, they've just been slowly classifying various existing revenue streams as cloud.
  • by jfdavis668 ( 1414919 ) on Monday November 30, 2020 @03:20PM (#60779538)
    If Intel can't handle it. Are Windows users actually looking for efficiency and halo applications? (besides Halo, of course)
    • No, but Microsoft is. Users follow the applications, and if your platform doesn't have the application, you're fucked. See: Windows Phone.

    • Can Intel succeed where they've failed for the last 5+ years, at building hybrid processors? The next year to two years should answer this question.

      So why does it have to be Intel?
      The only reason I can think of is that they are so much bigger than AMD and they have the resources to turn things around for themselves. They did it before, with the aid of some behaviour where they were lucky to get away with only paying AMD a $1.25B settlement. Oh, by the way, that agreement [amd.com] expired recently.

  • MS should opensource Win32 and convert windows into a Linux Distribution and then cycle development down and eventually kill it off entirely. I really don't see any reason for Windows to be a thing in 2030.

    • This is kind of a "too many standards" fix to the problem. The reason Microsoft has such a hard time getting developers to write for the UWP is that, between 2005 and 2015 they rapidly cycled from .NET/WPF as the preferred development platform, to .NET/Silverlight, to Metro, to UWP. Within that time frame Microsoft has created and deprecated entire APIs and UI frameworks. Why would anyone develop a major product for UWP with that track record? And, if Microsoft does it yet again, why would anyone develop fo

      • Apple is guilty of this as well back when they were desperately trying to modernize classic MacOS. Anyone remember Game Sprockets, QuickDraw GX and OpenDoc? .NET sadly suffers from DLL hell, mostly because they broke a lot of things with the move to 2.0. A lot of early applications for 1.0/1.1 break on the newer run times.
    • Haha, no. They will milk it for whatever is worth for OEM licenses and then for support services to businesses, in a way that makes Unisys executives look to Microsoft with a simultaneous feeling of admiration and disgust.
  • "stuck in the same rut that American cars grappled with in 1973 to 1983."

    My PC definitively has a few Rust problems.

  • In my opinion Intel doesn't so much have an architectural problem, everyone has an EUV problem except TSMC. Apple used it's size and margins to buy pretty much all 5nm, but given it's node advantage it's architectural advantage is fairly minimal to non existent.

    Big/little seems to me to be screwing around in the margins, I think Apple mostly does it to keep architectures similar with iOS where the little cores really need to go way down in power consumption. Those extremes are not necessary for laptops, Int

    • Damn, I screwed up a lot of its there ...

    • Samsung is still making progress. Not the same progress as TSMC, but they aren't having problems like Intel is with their decidedly-not-EUV 10nm node. Jury's still out on Intel's EUV 7nm node but yeah, it isn't looking too good.

  • by spun ( 1352 ) <loverevolutionary&yahoo,com> on Monday November 30, 2020 @03:43PM (#60779680) Journal

    This author has done a very poor job of explaining exactly what his complaints mean, or what would fix them. I get that he has four main issues:

    1. The inability of Intel to deliver processors that deliver a great harmonization of energy efficiency and high performance
    2. The beginning push of ARM-based processors into the mainstream computing realm, in the form of Apple silicon.
    3. The lack of compelling energy efficient halo applications on Windows to pull consumer desire for the platform forward
    4. The lack of PC computing devices that push the platform forward overall.

    It basically sounds like he wants hardware and software developers to "wow" him in some undefined way, but he remains unwowed.

    And his answer to his own question, "So how do we escape this PC Malaise Era?" is just more handwaving. Intel needs to make new, better stuff! We need exciting apps that are energy efficient! We need more tablets!

    Christ what a load of hooey. Meaningless questions with insipid, vague answers.

  • by JudeanPeople'sFront ( 729601 ) on Monday November 30, 2020 @03:51PM (#60779718)

    The future for most users is web apps in the cloud. Not for the nerds, but we are a small and unprofitable market. If Intel fails, so what? Windows works fine on AMD. If ARM CPUs become better, or RISK-V, or whatever, Windows can be ported on them.

    The way things are going with Windows-Linux integration, Windows will become a Linux DE. It makes sense for everyone involved. Windows is not the cache cow for Microsoft anymore, so they will chuck it on top of Linux and save themselves a ton of development effort to keep a proprietary kernel.

    • That won't need PCs or Windows. Windows has far too large an attack surface to be a good terminal.

      Azure Sphere looks nice though.

  • 1. No. While Hybrid chips seems to be all the rage, it's not going to do anything but scale performance on a chip vertically instead of horizontally. You'll basically have a thicker CPU that's faster but eats more power, and it will still have the problem of cooling the lower parts of the chip since they don't have direct contact with the cooling die, and since silicon doesn't have great heat dissipation...

    2. The Windows Store is dead. The only company still taking it seriously is Microsoft, and even they c

    • Their second attempt at ARM is better but it's not gong to go anywhere when it only works for one chip manufacturer

      I'm curious about ARM's new parent company. If anyone can kickstart a post-x86 Windows era, surely it's nvidia with a Cortex-X1 gaming laptop - Surface Pro Xbox.

  • I guess point 2 addresses this somewhat, but do we really need any huge step function in Windows PCs at this moment? Anything you can buy right now is going to be overpowered for anything the majority of users are doing, with the exception of high end games and content creation, both of which are arguably edge cases.

    PCs are powerful, Windows 10 is fairly robust and not too annoying. Rather than mess about with any of that, I'd like to see something more interesting in applications. It's not what you have

    • Windows 10 robust? When updates break stuff? Nope. Microsoft quality control is shit, has 800 million member QA department that often gets ignored when it finds problems.

  • While Apples new chip may be fast for some things, it's not a AMD 3900x, once more, the 5950s are coming out. Chips with X86 legacy will be around for a while.
  • No more Surface junk (Score:4, Informative)

    by Scutter ( 18425 ) on Monday November 30, 2020 @04:18PM (#60779840) Journal

    My company went all-in on Surface awhile back and we've had tons of each iteration. They were junk. The failure rate was unreasonably high and even the cheapest repairs were never less than 50% of the cost of a brand new unit. You know, low enough to make you question the wisdom of repair vs replace but high enough to soak you for simple fixes. I finally convinced the bean counters to switch platforms to something that isn't a steaming pile of suck.

  • Any comparison with American cars of 1973-83 is nasty.

  • by timmyf2371 ( 586051 ) on Monday November 30, 2020 @04:22PM (#60779856)

    PCs reached a point some time ago where they are pretty much appliances with an interactive operating system.

    Think of the average user or organisation looking for a computer today; a laptop with 8GB RAM and an SSD which comes with Windows 10 & lifetime updates is neither high-end nor expensive. It will run a web browser, office and budgeting software and probably even some non-intensive games. Assuming no physical damage, this laptop will last a fair amount of time.

    There will continue to be developments in computer hardware and architecture. Immediate things which come to mind include advancement of graphic card technology as well as better power management (this includes so-called 'Hybrid chips' i.e. improvement in CPU architecture). There are users who will benefit from the latest and greatest improvements - people working with graphics/media, big data experts, those working on AI and ML.

    But the rest of us and the average user will simply pick up these enhancements when it's time to replace their appliance.

    The appliance nature of the modern PC is the exact cause of 'PC malaise'. Who gets excited about the incremental improvements in their washing machine or their fridge?

    • Apple will take everyone's lunch, being vertically integrated smother the market and will have to make extreme mistakes to be dislodged.

      IBM's mistake to create the PC was all our gain, this seems to be the end of the ride.

  • This is not the issue with PCs. The issue with PCs either are companies changing UI for little to no reason, making slow ass web apps that masquerade as full apps, and the constant stream of updates.
  • Hasn't this been said for like 10 years, the PC is dead or dying, and it still ticks along.
  • by JMZero ( 449047 ) on Monday November 30, 2020 @05:16PM (#60780096) Homepage

    I work as a programmer, primarily with .NET - and I generally get along just fine.

    My kid, though, is just learning programming - and that means writing games. He has done some stuff in Scratch, but his ideas are complicated enough that he's better off moving on from that. He did an online course with Python, and we started writing a text-y game with that - but he really (and naturally) wanted proper graphics. I couldn't immediately figure out how to get an image library working with the goofy install they'd done, so I figured "Hey, I can just get him going in .NET" - that'll be easy to support him with.

    Writing a simple 2d game in .NET is a bizarre mess.

    From past experience I knew using the built-in forms/GDI would be slow (performance wise) and unpleasant, so I figured I'd find out what MS recommended now. I searched.. and I don't know. There's like 10 generations of half-realized/extinct crap: DirectX.NET and XNA and at least a couple more, all with half-functional zombie documentation, dead links, and instructions that no longer work. There's a bunch of MS articles pointing you towards libraries that died long ago (SlimDX, SharpDX) and otherwise suggesting approaches that are broke.

    Or they suggest you use very heavy-weight libraries - things like Unity or Monogame - that have a whole network of other challenges to pick up, and/or require a ton of boiler-plate code/external tools to get going. These tools are probably great - but the point of this is to learn programming, not to learn how to use a bunch of resource editors.

    Eventually I found Win2d, which is so far "not insane"... like, you can load and draw an image to your form without a ton of code, without external tools, and without having to pay to remove a splash screen. But even Win2d has a bunch of baggage that makes it awkward for a new programmer. Like, it works with "Universal Windows Platform", which means a bit of config to get going and is generally odd. You have to edit XML to place your form on the page (which I'm OK with, but is pretty imposing when you're starting out). You have to use special URIs (rather than just hardcoding a path) to get to your assets, since you don't seem to have permission on the regular file system. There's a bunch of methods that are async only, which makes for some very intimidating looking await patterns when you're just starting out.

    Like, I can get a project going for him to work with, but it's all been way harder than it should be.

    MS should put a simple graphics/sound library in .NET: something that's easy to consume using straightforward, self-contained, procedural code. Ideally, you could compile it to a self-contained EXE, a web page, or an Android app.

    Doing that would capture a ton of interest from kids looking to program - and I feel like MS used to understand this (back in GORILLA.BAS days).

Business is a good game -- lots of competition and minimum of rules. You keep score with money. -- Nolan Bushnell, founder of Atari

Working...