[HN Gopher] Microsoft confident Windows on Arm could beat Apple ...
___________________________________________________________________
Microsoft confident Windows on Arm could beat Apple with Snapdragon
X Elite
Author : top_sigrid
Score : 22 points
Date : 2024-04-08 17:12 UTC (5 hours ago)
(HTM) web link (www.theverge.com)
(TXT) w3m dump (www.theverge.com)
| top_sigrid wrote:
| From the Article: "Microsoft is so confident in these new
| Qualcomm chips that it's planning a number of demos that will
| show how these processors will be faster than an M3 MacBook Air
| for CPU tasks, AI acceleration, and even app emulation."
| top_sigrid wrote:
| https://archive.ph/cay2p
| retskrad wrote:
| It's great news to see competition to Macs. However, we need
| independent tests on it's efficiency. It doesn't matter if
| they're faster than Macs if they drain more battery. Also, at the
| end of the day, these chips will be inside Windows laptops.
| Windows OEM's ain't stealing Apple customers.
| forrestthewoods wrote:
| > Microsoft claims, in internal documents seen by The Verge, that
| these new Windows AI PCs will have "faster app emulation than
| Rosetta 2"
|
| I'm deeply fascinated by this. Rosetta 2 is crazy impressive. If
| the Qualcomm/Windows works as well then a Windows ARM laptop
| could be genuinely interesting.
| dmitrygr wrote:
| They are careful to promise better performance without mentioning
| battery life. And, separately, good battery life without
| mentioning performance. To me that creates the impression that
| they cannot do both. You'll either have good perf but only for a
| few hours (then just get an AMD CPU), or good battery life but
| shitty perf (then just get a rockchip-based Chromebook)
|
| The reason everyone is in love with Apple silicon chips is that
| you get amazing perf and still get twenty hours of battery life.
| kiririn wrote:
| x86 chips like 7840u can already trade blows with apple on
| efficiency under load, so I'd be surprised if this couldn't.
| With some 'little' cores to break through the 3w idle floor of
| the AMD chip, I reckon the Snapdragon stands a chance. But it's
| down to the OEMs to get idle power draw of everything else low
| and fit decent sized batteries
| bevekspldnw wrote:
| And I'm confident I don't want to use an operating system that
| forces me to disclose my email address to use the damn computer.
|
| I just bought a Microsoft branded laptop and to my horror it
| wouldn't let me even create a local account without giving it an
| email. I gave it a fake one and now it pesters me every couple
| minutes to validate it.
| pjmlp wrote:
| Even so, the greatest issue is getting Windows developer
| community to actually care to provide ARM binaries, which has
| been the main pain point from previous Windows on ARM hardware
| attempts.
|
| Managed languages will do just fine, their dependencies on native
| libraries, or typical compiled applications and libraries not so
| much.
|
| Breaking compatibility on Windows has always ended up badly for
| Microsoft, as it goes against the culture they themselves created
| on the PC.
|
| So it remains to be seen how much will keep running under
| emulation.
| mike_hearn wrote:
| Not necessarily even for managed languages. Sure, you can in
| theory just package an ARM runtime, but the devil is in the
| details.
|
| For example, it's typical that packaging tools only build
| packages for the platform they run on. So if you run the tool
| on Windows Intel, you get packages for Windows Intel. Want ARM?
| You need to run it on Windows ARM. Don't have such a machine?
| Tough cookies.
|
| This is one of the problems that motivated me to write
| Hydraulic Conveyor, and so it can build packages for every
| platform portably including Windows Intel and ARM regardless of
| what OS or CPU you run it on. It knows how to download all the
| different runtimes for all the different OS/CPU combinations
| and make and sign all the packages in parallel, without any
| native tooling. It knows how to strip out native binaries that
| appear in managed dependencies, but don't match the target. It
| makes download pages that detect what CPU you have (on Chrome,
| Safari lies unfortunately), or for Windows an installer that
| detects the user's CPU and downloads/installs the right arch
| for the user. So it handles all these details and lets you
| release for hardware you don't have. But that's a fairly unique
| party trick. For every other system you're going to be reliant
| on multi-platform CI, which may or may not support every target
| (ARM Macs took a long time to come to GitHub Actions), and in
| which adding extra platforms increases the cost of every build
| - sometimes by a lot!
|
| Apple relied not only on Rosetta but also their whole fat
| binary infrastructure, and they have sticks like notarization
| and carrots like the App Store to encourage devs to do the
| work. Also Mac apps tend to be well maintained and gaming is
| not so common. MS don't have the same advantages, so apps will
| presumably be running in emulation for a lot longer and have
| more compatibility warts.
| smackeyacky wrote:
| > Breaking compatibility on Windows has always ended up badly
| for Microsoft, as it goes against the culture they themselves
| created on the PC.
|
| I would say it was less of a culture and much more of a
| marketing/lock-in strategy that worked extremely well for
| decades. Firstly by being the most popular windowing
| environment that still ran MS-DOS apps, then being the only
| windowing game in town when they rug-pulled OS/2. That
| consolidation enabled Office to become dominant and _that_
| compatibility took over as the cudgel to keep the world locked
| to Windows.
|
| Microsoft, for the first time, is betting against their own OS
| with the web version of Office. I can rent Office on a monthly
| basis and run whatever OS I please to access it (mostly). But
| if I no longer absolutely have to have an Intel based Windows
| machine that lock-in has to shift to Office online.
|
| What processor you have will be irrelevent to them, they'll
| dump native app development as soon as they can because it's no
| longer part of their bulwark of defences, it's now an
| encircling trap they need to break out of.
| sylens wrote:
| Even if they were to match Apple Silicon pound-for-pound, they've
| been horrible stewards of the Windows OS in terms of nagware,
| bloat, and user hostile behavior (the default browser nonsense).
| If they actually made an operating system that respected their
| users, I'd be more inclined to take a serious look at these
| devices.
| freeone3000 wrote:
| I'm excited about the Windows release bringing standardization
| and solid hardware in this area. Apple Silicon is locked down, so
| having a traditional bootloader (which Windows Boot is) on ARM
| will be great for linux adoption in this space -- which is really
| what I see for these devices going forward.
| angulardragon03 wrote:
| I'm super confused - Microsoft has tried this both in 2012 and
| 2016 (with Surface RT, and then again with Windows 10 on ARM).
| Neither of these turned out to be great for Linux adoption in
| this space (to my knowledge) - what leads you to believe that
| this time will be different?
| blackeyeblitzar wrote:
| I've had issues on Windows 10 recently with windows updates not
| installing properly. After lots of research in their forums I
| found that I have to run scripts people have come up with to fix
| the issue, which just does not seem trustworthy or user friendly.
| I have just let these updates repeatedly fail because I am
| worried about messing up my installed OS somehow. On top of this
| drop in quality, I've seen more and more anti-competitive / dark
| pattern behavior with notifications and pop ups trying to push me
| towards Microsoft software and services. I feel like Microsoft is
| violating what a general purpose computer is supposed to be.
|
| Are newer Windows versions any better about software quality or
| these dark patterns? Given that MacOS already is free of such
| nagging and is already great functionally (like on battery life
| and performance), I am not sure what motivation there is (other
| than price?) to take a chance on future Snapdragon Elite laptops.
| throwaway5959 wrote:
| It's not the hardware that's the problem.
| nerdjon wrote:
| It's honestly meaningless until the entire Windows market moves
| over, which I doubt gaming will anytime soon.
|
| Apples transition worked since they could force it. Developers
| targeting Mac users had to work on Arm optimized versions or be
| stuck working through Rosetta.
|
| But if this doesn't sell well enough for developers to care, why
| would you buy a device that only runs software through emulation.
| Chazprime wrote:
| Are they magically reverting to Windows NT 4.0 for this build?
___________________________________________________________________
(page generated 2024-04-08 23:02 UTC)