Microsoft is Confident Windows on Arm Could Finally Beat Apple Silicon-Powered Macs 85
An anonymous reader shares a report: Microsoft is getting ready to fully unveil its vision for "AI PCs" next month at an event in Seattle. Sources familiar with Microsoft's plans tell The Verge that Microsoft is confident that a round of new Arm-powered Windows laptops will beat Apple's M3-powered MacBook Air both in CPU performance and AI-accelerated tasks. After years of failed promises from Qualcomm, Microsoft believes the upcoming Snapdragon X Elite processors will finally offer the performance it has been looking for to push Windows on Arm much more aggressively. Microsoft is now betting big on Qualcomm's upcoming Snapdragon X Elite processors, which will ship in a variety of Windows laptops this year and Microsoft's latest consumer-focused Surface hardware.
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. Microsoft claims, in internal documents seen by The Verge, that these new Windows AI PCs will have "faster app emulation than Rosetta 2" -- the application compatibility layer that Apple uses on its Apple Silicon Macs to translate apps compiled for 64-bit Intel processors to Apple's own processors.
App emulation has been a big problem for Windows on Arm over the past decade, but Microsoft did deliver x64 app emulation for Windows 11 more than two years ago. This helps ensure apps can run on Windows on Arm devices when there isn't a native ARM64 version. Native Arm apps are key for improved performance on upcoming Windows on Arm laptops, and Google has just recently released its own ARM64 version of Chrome ready for these upcoming devices.
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. Microsoft claims, in internal documents seen by The Verge, that these new Windows AI PCs will have "faster app emulation than Rosetta 2" -- the application compatibility layer that Apple uses on its Apple Silicon Macs to translate apps compiled for 64-bit Intel processors to Apple's own processors.
App emulation has been a big problem for Windows on Arm over the past decade, but Microsoft did deliver x64 app emulation for Windows 11 more than two years ago. This helps ensure apps can run on Windows on Arm devices when there isn't a native ARM64 version. Native Arm apps are key for improved performance on upcoming Windows on Arm laptops, and Google has just recently released its own ARM64 version of Chrome ready for these upcoming devices.
Need emulation for drivers! (Score:5, Interesting)
I suppose it doesn't matter as much for average office users, but for power users/tech folks, the huge lack of device drivers re-compiled for ARM windows is a big problem.
It's an even bigger problem if you want to use any hardware that isn't the latest and greatest. How many companies are going to recompile and release drivers for ARM for hardware that isn't current generation?
MS provides drivers for the laptop, like Apple (Score:2, Redundant)
I suppose it doesn't matter as much for average office users, but for power users/tech folks, the huge lack of device drivers re-compiled for ARM windows is a big problem. It's an even bigger problem if you want to use any hardware that isn't the latest and greatest. How many companies are going to recompile and release drivers for ARM for hardware that isn't current generation?
Emulation is for 3rd party apps, not Windows and its drivers itself, nor key MS apps like Office.
Also we are talking laptops, not an ARM desktop where you can plug in random x86 era PCIe cards. These Microsoft laptops will be like Apple's, pretty much the only hardware they need to support is what it leaves the Apple/Microsoft factory with. Any user add-ons will strictly be through standardized interfaces like thunderbolt, usb-c, etc.
Re: (Score:3)
Re: (Score:2)
Aren't most printer drivers running in userspace, so they can be emulated anyway?
There's also standards like Postscript, PCL etc which don't need printer specific drivers.
Re: (Score:3)
These days most printers are driverless which is to say they have a standard protocol so the driver is built into the OS. Basically with ipp you can fling some standard image formats at them (a subset of CUPS, PNG), or any additional formats like PDF that they may offer.
They also present the options in a standardized way. It's quite similar to PPD, except that they speak bitmaps not postscript.
Re: Need emulation for drivers! (Score:3)
I don't think it's that big of a deal like it used to be. USB HID classes have largely made custom drivers for most peripherals pointless. It's only really a concern for certain types of PCI devices, like GPUs. There are some little extras that you may want vendor supplied software for, but it's generally userland software, making it suitable for emulation.
Re:Need emulation for drivers! (Score:4, Funny)
For me a bigger problem is that I'd have to use Windows. I'd rather eat a broom soaked in old, cold hot dog water.
Re: (Score:2)
While what you say was the downfall of early ARM attempts, I honestly don't think this is relevant in 2024. The number of devices which require some kind of special driver these days is vanishingly small even among power users. Most devices on the market these days including some more advanced devices such as e.g. colour calibrators for photography, audio interfaces for music production, or even my damn oscilloscope (to mention three device types I've had no end of driver problems with back in the days) all
Re: Need emulation for drivers! (Score:2)
If the divers are user space, then it should be relatively trivial. Kernel level ones, which make certain architecture assumptions or are timing critical, might be a little more challenging?
Now Boarding: The Hype Train (Score:5, Insightful)
Seems highly qualified comparison, M3 Air only (Score:3)
Re: (Score:3)
its nice to see Microsoft enter the game
No, it's actually not. Microsoft-controlled hardware is a fantastic way to end up with a pile of EOL'd devices which have otherwise decent hardware after only a couple months.
Knowing Microsoft, there will likely be call-home functionality in the hardware itself, and booting anything other than approved, licensed OSes will be impossible. That's not good for anyone except Microsoft, particularly when you consider how good Microsoft is at security (they aren't) and the m
Re: (Score:2)
Agree. They are comparing their future unannounced date computers with M3 air which is over a year old. Besides, ALL surface laptops released so far have been released at a price point higher than MacBook Air. So what they are saying is that, "out future offering will beat the past offering of Apple at a price point higher than Apple's".
The second thing is that ARM computers are first class citizens in Apple's eco system and will be second class citizens in MS world. Check their history of how they supporte
Re: (Score:2)
Eh. I'll be one of the first to be critical of Microsoft (because fuck'em, that's why), but their historic platform support really doesn't factor in here. It's a different scenario in 2024 than when Itanium and MIPS architectures were around and relevant: they were niche.
We're in a world where x86-64 is quickly sunsetting in relevance, and the CPU core race is effectively over. Intel is consistently on the losing side for TDP and overall performance against AMD, and it doesn't much matter at this point beca
Re: (Score:2)
Re: (Score:2)
The fact that they're spotlighting the M3 Air tells me there's at the very least some cherry picking going on.
Yes and no. On the one hand, yeah the Air is the least powerful of Apple's devices, but on the other hand this is the market in which Microsoft competes directly. They aren't producing full sized powerbooks. Even their full sized Surface Laptops emphasise size and form factor over performance, but to date all of Microsoft's ARM efforts have been almost universally slate devices, so it stands to reason they would compare themselves to Apple's thinnest most mobile device.
Re: (Score:2)
By the time the new devices from MS are available, it's likely that Apple will have moved on to the M4.
Most Mac software has already been ported to ARM so it runs natively, whereas very few Windows apps have native ARM versions. As fast as their emulation might be, it's going to be slower and more resource hungry than native code. And likely part of the reason why Apple have not made huge efforts to improve Rosetta is that for them it's a temporary migration aid that becomes less relevant over time.
When i g
Re: (Score:2)
I'll reserve final judgement until I see the actual benchmarks reproduced by a reputable third party. The fact that they're spotlighting the M3 Air tells me there's at the very least some cherry picking going on. As of now, this reads like marketing hype.
Not only does MS carefully avoid mentioning the M3 Pro or Max (nevermind the soon-to-be-released M3 Ultra); but you can be sure that by now, Apple is already well into testing Pre-Production Samples of the M4. . .
Performance (Score:4, Insightful)
We don't give two shits about "performance" - laptops are fast enough already.
Apple massively wins in the battery life department. It also wins in the stability department. In the general "getting shit done" (that isn't gaming), Apple destroys any Windows laptop. If you need to performance, offload it to a workstation or server somewhere.
Per usual, Microsoft is chasing benchmarks for headlines, but the reliability of the ecosystem as a whole is trash in comparison.
I love how my docked Windows laptop will randomly bluescreen a few times a week because Thunderbolt devices don't know how to properly sleep on a Windows machine.
Note: I have and use multiple Windows laptops, desktops, workstations, and servers. I'm not an Apple shill, I just know and love that my MacBook Air can last more than an an hour or two on battery, its an all-day device just like a phone. I wish many of my actual useful Windows desktop utilities had a Mac variant. Its a struggle!
the fonts are too small. (Score:5, Interesting)
The fonts on OSX are (still!!) too small for some people, like me, and there's nothing meaningful that can be done about it, still to this day! Ubuntu/Gnome puts OSX to shame in the department. Windows fares only slightly better in this regard. To scale fonts in OSX in any meaningful way, the entire screen must be enlarged.
With every new release of OSX I go check at the Apple store and I'm always disappointed, (and a little smug knowing I'm better off).
Me loves Ubuntu/Gnome.
Re: (Score:2)
This totally. Totally baffles me why such fundamental usability features are missing from macOS. And to a lesser extent, Windows also. I suspect it is because the UI toolkits on both platforms are very inflexible and rely on fixed layouts.
Even zooming the whole screen is difficult in macOS. Since the monitor is HiDPI, macOS presents very limited choices for "screen resolution." I found a third-party utility that creates a virtual screen of any size and then you mirror that into your primary screen which
Re: (Score:2)
Maybe you want fine-grained control over all your fonts, which could be tricky. But if all you want is larger text, the Displays control panel is your friend. There's even an icon labeled that.
Re: (Score:2)
Thank you for your reply, however I find the Display control panel magnifies the entire display when all I desire is larger fonts to read from, not less functional desk space. This deficiency is especially pronounced on something like a large 4K display.
Re: (Score:2)
You can't adjust the DPI?
Re: (Score:2)
No. Not easily, anyway. And that has the same effect as the feature I described. Lower DPI = less screen real estate, not just larger system fonts.
Re: (Score:2)
If you just make the fonts bigger and don't scale other things too, then you end up throwing out assumptions made when UIs etc were developed. Often a button for instance will be just big enough to contain its label, if the font becomes bigger it no longer fits inside the space that has been reserved for the label. Things end up ugly or even broken.
It's a very hard problem to solve.
DPI is a separate thing, sizes are supposed to be based on real world lengths not numbers of pixels - so higher DPI should just
Re: (Score:2)
There's also the Text Size slider under the Accessibility control panel.
Re: (Score:2)
Agreed. I almost took the MacBook Air back until I found the magnify setting in Displays. It's still not a perfect solution, but it's usable.
Cook is not a young man. How does he not notice this?
Re: (Score:2)
Cook is not a young man. How does he not notice this?
I dunno, macOS has a pretty extensive Accessibility control panel. Have you checked it out? There's a Text Size slider in there, among many other things.
Re: (Score:2)
Re: (Score:3, Insightful)
There are only two reasons to run Windows: insane corpo software, and proprietary games. Both are thoroughly x86-only. So why would anyone wants to run Windows on ARM? Are they so used to blue screens, insta-reboots, boots failures, botched upgrades, and random program crashes left and right? All on perfectly sound hardware.
Re: (Score:2)
Gosh, Microsoft will optimize just like they did for the ARM Surface!
Oh, wait....
Windows: Now trailing edge technology.
Sorry. Forgot for a moment.
Re: (Score:2)
Way back in the day, I was volunteered to run windows on DEC Alpha for my department.
I asked the same question. Why would anyone want to run windows on Alpha?
Re: (Score:2)
I don't know Microsoft might win this one. I am using 86x64 build Firefox on my m1 (long story) right now.
A few required plugins that don't exist for ARM work with it. The performance nearly indistinguishable (in human terms) from the ARM build on other sites. Even sites like youtube, playing back full screen video. It does use the battery faster. i am sure if i compared with some benchmarks suites the native build would show faster too but that isn't the point.
The point is at least on MacOS highly relia
Re: (Score:2)
I'm for one allergic to laptops, but for when I need to use one, I have a Pinebook Pro. It's several years old by now, and it was a cheap tinkerer device even when it was new; it's RK3399 based. And yet, there are no performance problems when it comes to client tasks: it needs to run a browser, terminals, ssh, maybe occasionally Gimp or something of that kind -- all of which it does fully adequately.
Meanwhile, both my desktops are 64-way (amd64, riscv64), and even the AMD one costed way less than a new fa
Re: (Score:1)
We get that you haven't run Windows since "Windows ME" but, that was an aberration, grandpa.
Re: (Score:2)
I love how my docked Windows laptop will randomly bluescreen a few times a week because Thunderbolt devices don't know how to properly sleep on a Windows machine
For me it's my docked M1 Pro which randomly crashes about once a fortnight. It was hopeless monitor+ethernet on initial third-party docking station so I bought the expensive one listed on the Apple store, CalDigit. This has never been particularly happy with sleep or its external monitor but at least is a bit better. To this day, iMessages on my mac crashes every single time I open one of the chat threads with my kids (which is populated with 100s of screen-time requests which I guess is too hard for iMessa
Re: (Score:2)
I have an OWC TBolt hub, and I've never had a crash on my M1 Max Macbook Pro. Seems to be a problem with your hub. Now I would agree that a hub should not crash the laptop, even if it's driving video from the laptop.
And I've never had a crash from iMessage. So my mileage definitely varies...
Re: (Score:2)
Re: (Score:2)
I actually use the battery in my laptop as a work clock, now. I very rarely work tethered anymore. If my battery is dead or dying on a fresh charge at the beginning of the day, that means I need to get up and get out of the office to do something else. Diminished returns.
Re: (Score:2)
Windows used to be the fastest. At memory leaks. Maybe still is.
Re: (Score:1)
Reading about "stability" in 2024 feels awkward.
The only crash I can recall was with aging company laptop tortured by antivurus and similar crap, with me compiling some big ass Java nonsense and crash being vmost likely due ot overheating.
4800u (AMD's oldish CPU) beat M1 at power/perf (dependent on task, but was largely in the same ballpark no matter what anyhow) per anand's tests. Did things REALLY change drastically since that time? I doubt it.
We have x86 laptops that can last a day for what, at least ha
AI PCs (Score:1)
That will still complain about missing drivers and dlls.
This after 30+ years of PCs that can be almost completely disabled by unlinking a font.
I'll pass.
any thing can beat apples ssd pricing! (Score:1)
any thing can beat apples ssd pricing!
If MS can finally give us a standard Arm platform (Score:4, Interesting)
Arm's biggest problem for decades has been the general lack of a standardized platform. Every implementation (mainly SoCs) so far is different and incompatible with each other as far as hardware trees, booting, and so forth. Linux support is awful, frankly, because vendors fork distros and also the kernel, release it with some binary driver blobs for a while, then loose interest long before anything gets merged into the mainline kernel. Honestly at the moment, Linux on Apple M1 and M2 is probably your best Linux experience out there, partly because Apple has defined a consistent platform.
If MS can define a standard Arm platform that all vendors must follow, complete with something like UEFI, then hopefully we'll have something that Linux distros can build on as well. Then finally I can just download an install image from Fedora, Debian, Mint, etc, and install the same distro from the same source on all Arm hardware. Then there will be almost zero reason to buy x86 hardware (or a Mac M series).
Re: (Score:3)
I do not understand what you are describing. ARM licenses their designs to others to make chips. There is standardization when it comes to ARM like the current ARM v8 APIs. The difference is that companies who license ARM can customize the chips the way they want. Does the chip need to be able to hardware decode Google VP8, h265, etc. or use 2.5Gb Ethernet? A smartphone chip is going to be different than a laptop or tablet chip.
Windows on ARM has been a thing for over a decade and it has been a miserable fa
Re: (Score:2)
You've never tried to use one of the many Arm SoCs with Linux? Windows on Arm is nonexistent thus far. Linux on Arm is a horrible experience frankly except for a very few number of devices such as the Pi, but even there, there are proprietary bits. Every device requires a distro and kernel fork for the particular chipset. Like I said, no standardization of the platform at all. I cannot just download a standard, generic distro and run it on any Arm device. This is quite different from the Intel world.
If
Re: (Score:2)
Re: (Score:3)
No not really. A standard platform is a requirement for a user modifiable device. And while you can take my desktop with it's upgradable GPU, RAM, CPU, ... every component out of my cold dead hands, the reality is the computer industry does not run on user modifiable systems, it runs on integrated OEM provided solutions.
There's no reason to standardise anything. Users run apps. If the apps run on Windows and Windows runs on the hardware then that's all the platform that 99.9% of people care about. While you
Re: (Score:2)
Depends on the measure of success. In the wasteland that is the Android phone, yes Arm is truly successful, and that's really all they care about, honestly. And if MS wants to do it that way, they probably will find some success, but not a lot. Windows RT was their last foray into this world, and it was majorly unsuccessful.
If we ever have general-purpose Arm laptops and desktops, it will absolutely require some standardization. Linux on Arm totally sucks rocks. Which is sad, because a lot of Arm systems,
Re: (Score:2)
You are banking on MS allowing you to run Linux on their box?
delusional suit thinking (Score:1, Redundant)
Re: (Score:1)
windows cannot beat osx on any metric whatsoever.
OMG fish in a barrel with this shill
https://en.wikipedia.org/wiki/Usage_share_of_operating_systems#Desktop_and_laptop_computers [wikipedia.org]
Re: (Score:2)
better
nobody said that. gp provided one metric where "windows beats osx" quite clearly to someone who literally wrote "windows cannot beat osx on any metric whatsoever", and ipso facto made that person look really dumb, but you now look even dumber.
Yeah but... (Score:5, Insightful)
This is assuming Apple is sitting still. Which they are not. They are claiming to beat the M3 Air (fanless) which is a shipping chip, and not even their highest performing variant; but it's not a stretch to assume Apple is working on an M4 that will outperform the M3 and therefore the Qualcomm. But hey, I'm good with these two companies taking turns leapfrogging each other -- consumers win.
Re: (Score:2)
Exactly. They're saying they'll beat the most basic chip that Apple offers and do so a year after Apple first began shipping it.
Totally down for the competition. It's just funny they're so proud that they THINK they might be able to beat a year-old processor.
Re: (Score:2)
Bad, bad, bad, bad, no good, terrible idea (Score:3)
One of the reasons that windows is popular is backward compatibility
For users who simply browse the web and read email, this is not a problem
For others who have expensive packages that only run on x86, it's a BIG problem
A lot of older software is what is is. It will never be updated, either because the maker lost interest or they are out of business
Fantastic idea, youre short sighted. (Score:2)
One of the reasons that windows is popular is backward compatibility For users who simply browse the web and read email, this is not a problem For others who have expensive packages that only run on x86, it's a BIG problem A lot of older software is what is is. It will never be updated, either because the maker lost interest or they are out of business
You gotta start somewhere. The loss of multi-CPU type support after Windows NT is partially why the x86/Windows platform kinda stagnated for decades. So the old stuff wont run on ARM, ok fine use x86. Over time more and more software will be updated to support ARM which is exactly what we should be cheering. Heck, i wish MS would support ARM and RISC-V while they are off re-compiling their OS, drivers, and office suite of applications.
Re: (Score:2)
It's not like ARM is revolutionary. Apple is just able to buy out the best node, blow money on massive die area and silicon interposers ... architecture wise Snapdragon X Elite is actually more impressive, but even then the impact of ISA is minimal.
With good system design, x86 is good enough.
Didn't Microsoft try ... (Score:3)
A Year Late (Score:2)
So their 2024 processor will beat the most baseline 2023 processor Apple offers. Got it.
Was just on the verge of selling my MSFT stock.... (Score:2)
Now, I might hold off for a bit to see where this goes.
Who cares about metrics? (Score:3)
Within a certain margin of error, other qualitative factors dominate.
If Windows is your platform, the Apple ARM systems aren't going to be worth a hypothetical bump. Conversely, you aren't going to get macOS users to convert just because your laptop can outperform on a benchmark.
Seen this all too much in the vendor space, myopic focus on technical wins/losses at benchmark numbers while overlooking qualitative drivers of a decision.
Battery Life, Sleep, Stability (Score:4, Interesting)
These are what Windows needs. Not better performance.
If they want to compete with Apple that is.
My Mac will run for like 15+ hours on a single charge. I don't even know because it lasts so long I don't even think about it anymore. It will go to sleep and wake up instantly, and everything just works. My windows laptop takes forever to wake up and about 20% of the time it ends up blue-screening because it thinks some device was removed while it was asleep.
Re: (Score:2)
Agreed on the "wake from sleep" issue. My corporate Windows laptop, sometimes it does, sometimes it doesn't for no reason that I can see. I have always gotten it going again using various tricks.
My personal MacBook just works.
Re: (Score:1)
Cant believe Windows laptops STILL have this issue. Ive told people for years to stick with MS brand laptops in hopes the support and reliability was better for windows, otherwise tell them get a Macbook laptop.
Re: (Score:2)
My Mac will run for like 15+ hours on a single charge.
Let me guess, you don't drive an EV because you can't drive 1000miles on it without a recharge right? 15+ hours is the realm of battery anxiety, not a legitimate user requirement. The real acid test is can you get through the work day, and can you get from one airport to another. For most devices in the 8hour range that is already more than sufficient. My laptop doesn't do 15+ hours, and I can't say I've ever thought about it either.
The waking issue however is serious. While I do have one device which sleep
Re: (Score:2)
15+ hours is the realm of battery anxiety, not a legitimate user requirement. The real acid test is can you get through the work day, and can you get from one airport to another.
Not quite. The real acid test is whether, once you get to your destination, you can do a full workday without waiting six hours to recharge.
Conversely, my Mac's ~14-hour battery life means I've never left the house with the (bulky) power adapter unless I'm traveling overnight.
No thanks (Score:3)
Huh? What is there to beat? Is this just disguised Apple advertising?
Nothing can make WinRT fast (Score:2)
Windows is an operating system to run x86/win32 applications ... and Microsoft manages to make it do that worse every year.
It's still Windows (Score:3)
So what if it beats anything, it's still Windows. Yuck.
Oh really? (Score:2)
Microsoft is confident that a round of new Arm-powered Windows laptops will beat Apple's M3-powered MacBook Air both in CPU performance and AI-accelerated tasks.
"A product that only we can see beats something our competitor released weeks / months ago! We promise!"
Unless the product is quotes in a press release, don't waste our time until you've actually released the product and the claims can be independently verified. I'm tired of "we're gonna be faster / better than ${COMPETITOR} stuff REAL SOON NOW(tm)" and then when these claims don't actually intersect with easily observable reality they expect us to not call them Chicken Little and lap up the mediocrity wh
Qualcomm still a bit slower (Score:1)
The Snapdragon X is ~30% slower at single-core tasks than Apple's M3. (all M3 CPU types have similar single-core performance)
The Snapdragon X is ~15% faster in multi-core tasks than the base M3.
The Snapdragon is ~30% slower than the high-end M3-Max at multi-core tasks.
Snapdragon's performance is competitive with the entry-level Apple Macbooks and outclassed by higher-end Macbooks. Real-world energy usage on M3 Macbooks is RIDICULOUSLY
Microshit. (Score:1)
Versus M3? (Score:2)
By the time Microsoft's come out, or shortly afterwards, Apple will have M4 Macs.
It's the UX, stupid (Score:2)
Windows has a lot of fundamental flaws where it decides that which it wants to do is far more important than what you want to do. Even Linux has a better UX than Windows does.
Is CPU speed really the issue (Score:3)
Beat? (Score:1)
As of March 2024, Windows has something like 72% desktop/laptop OS market share; Mac something like 15%.
Microsoft seems to be going after a problem it doesn't have.
How big a diff with same architecture and fab? (Score:2)
Both products will be running the same instruction sets. Seems like Microsoft should be able to have a similar speed
The M2 and M3 CPU speed gains were mostly due to geometry reductions. That will run out of gas given the current fab limitations.
Apple put the graphics on-die. This has advantages and disadvantages. It will be interesting to see how far MS goes with integrated graphics. and what that means for AMD and NVIDIA (outside the AI space).