For us (Wireshark) the difficulty wasn't with our own codebase, but with getting our dependencies ported over. Most libraries built just fine, but some strongly assumed that "Windows" meant "x86".
It's not just Windows, either. Many libraries (particularly ones that use Autotools) are absolutely blind to the notion that you might want a universal binary on macOS.
When we ported OpenJDK to macOS, I ended up producing a universal binary by having the Makefile run itself to produce HotSpot twice, and then gluing them together with `lipo` afterwards. There isn't really a better way when the actual project configurations are different.
IIRC it was eventually removed because nobody else needed to do such a thing so it was hard to maintain.
Sure. How else would you build a universal binary then? Given the low-level nature of the language not many tasks can be usefully shared between different architectures.
For plain C/C++ you can just pass `-arch x86_64 -arch arm64` to clang. CMake takes care of this for you if you specify `CMAKE_OSX_ARCHITECTURES=x86_64;arm64` and IIRC Meson has similar functionality.
Clang is natively a cross-compiler. Pass in --sysroot and a corresponding valid sysroot tree for any micro architecture/platform (arm-eabi, macOS, Windows MSVC, PowerPC, Alpine Linux with musl, you name it) and Clang will happily retarget the binary to the correct target platform.
I assume this is faster than doing two separate builds, because it can skip certain steps of the complier pipeline, and only the items that are arch specific (codegen, probably others) are done twice?
How much work does clang have to do for this sort of thing (as opposed to llvm). Hypothetically could we start distributing programs in llvm ir, and compile that locally to ARM, x86, risc-v, or whatever else?
I mean, no, that’s silly, right? But it would be kind of neat…
> Why has it taken so long to get the Windows ecosystem fully on ARM? Apple’s transition only took a year or two.
No incentive for third parties. Apple dictates the hardware, and can say "no more x86" and devs either have to jump on board or abandon Apple.
No such thing with Windows. x86 is still the default on windows laptops, and will likely be for the foreseeable future. The X elite still seems to have no successor in the pipeline, and the few laptops that have it don't outsell x86 so why bother.
> "no more x86" and devs either have to jump on board or abandon Apple.
I don't think that's fair. They provided a very smooth transition, with a well performing translation layer. The user wouldn't have to care or even notice when they picked up a new ARM MacBook, except their battery lasted way longer and cooler. Everything that worked still worked (well, 64bit at least). I'm still running x86-64 apps, and developing for x86-64, on my ARM MacBook.
When the first Windows RT came out, there was no compatibility, and this wasn't communicated well. They nuked the customer perception on day 1. When they finally implemented the x86 compatibility, it had terrible performance and compatibility.
Now, Apple's 32-bit to 64-bit transition was definitely a "jump on board or abandon Apple", but the Intel to Arm transition was well crafted, from a user perspective.
> When the first Windows RT came out, there was no compatibility, and this wasn't communicated well.
It was worse than this. Source compatibility with the Win32 APIs you would use for ~20 years to target x86/amd64 was explicitly a non goal. To target ARM you needed to use their new, half-baked frameworks designed for the Windows 8 tablets. You couldn't recompile a desktop app, even if it would have worked fine had they given you the headers and libs to do it.
Even internally, even among decision makers, people were very confused about this.
They also wanted to force people to get everything from Windows Store. This deeply spooked Valve at the time who saw this as potentially the end of Steam, and it's a big part of the reason why they started working on making Linux gaming viable. So overall RT was an incredible failure and Microsoft is still feeling its effects today
They believed the ARM transition would be an excellent opportunity to lose support for the old. Remember the Barney Stinson model: New is always better
The modern Qualcomm Windows laptops also come with binary translation that's fast enough that you can use most amd64 programs just fine. Qualcomm had some GPU driver issues, but other than that things like gaming on aarch64 work just fine (and I imagine most video games put more varied strain on the GPU than Blender).
Windows RT is dead and buried. Blender won't run on it either, because RT never got to Windows 11.
This transition would not help much if Apple was just one of dozens of macos laptops.
Microsoft can't afford to discontinue support for x86 even over 10 years. They have enough trouble keeping Windows compatible with legacy software that the world runs on through generations of x86 hardware and software, transition to arm would be many times as bad.
Nah, you could’ve waited 5 years to port if you were concerned about your x86 binaries not being supported, the bigger deal was that Apple was shipping good ARM computers and if you’re a developer, you don’t want your app to seem slow on what consumers can see is clearly a fast machine.
Microsoft has made multiple abortive lethargic gestures towards ARM, but has yet to get people excited about an ARM computer that runs Windows.
Part of the trick was that ARM previously held a "cheaper, more efficient, smaller than x86" slot, and in the Windows ecosystem that was competing for the bottom of the market. Apple's ability to launch the M1 completely demolished that position and made an ARM chip the positively better choice, and it's just not something that Microsoft could ever have tried. MS have always treated the actual hardware as first and foremost someone else's problem. Which is a shame, because when they do produce their own kit, it tends to be very good.
CoPilot+ PC could have been it, but then they completly messed up with Recall, naturally no one wants to have such thing on their computers if they can avoid it.
Even before Recall, most people laughed at the idea of putting Copilot into notepad. Outside of people part of the hype train and investors, nobody wants Copilot PCs.
> And “nobody” is shorthand for “no technology professional”, right?
Yeah, I would say so - I've spoken to people that are genuinely excited for things like Recall. Like it or not, we (tech people) are in the minority in our disdain for it, and disdain for shoving "AI" everywhere.
And that's OK, we aren't the market for these features, but that doesn't mean a market doesn't exist for them.
Performance is A LOT more than having the ISA. Intel, AMD, ARM, and apple have spent DECADES developing and tuning their micro archetypes to run those instruction sets.
I like the RISC-V project and think they're doing great things for the future of open computing, but if you think we're 2-3 years away from a RISC-V chip that's comporable to the top of the line X86 or ARM chips you're going to be sorely disappointed. It's gonna be 10-20 years before we get to that point.
I do think where gonna see more RISC-V chips in embedded and subcomponent context where the low or non license fees are gonna make it competitive pretty soon.
Apple had all the incentives in the world to provide a transition plan and leave Intel behind. Also lots, and lots of practice with platform migrations, and they were transitioning to a hardware platform they were already selling in the billions.
Microsoft has exactly none of that. I'd be astonished to see RISC-V or ARM "take over" in the Windows world in less than another two decades, unless Intel's ongoing decline drags the entire X86 platform with it.
So far, loong64 seems to outperform RISC-V, and it also has the benefit of a large existing user base. If any new player is going to hit the desktop market, I expect it to be Loongson at this moment.
I want to believe RISC-V is just around the corner, but I've been promised the same about POWER9 and RISC-V over the years.
About 4-6% in US and other Western countries, depending on analytics source. Increases in the past year or so by a full percentage point or so. More in India, less in South America, barely any in China.
The increased visibility due to SteamOS/Proton, coverage by various prominent TechTubers, end of Windows 10 support, availability of image-based OS upgrades and Flatpak/Snap to supercede package dependencies, all seem to combine into an actual breakout year comparatively. We'll see if the trend holds or was just a one-off.
Until monopoly, or until 15-20% when companies start offering widespread Linux support for applications and peripheral management, and retail stores start selling some devices with a Linux desktop preloaded?
I'd do the math, but it really depends on whether the recent growth holds, accelerates, or slows down again. So, hard to tell.
It will hardly matter to me or anyone else in Europe, if we cannot get a RISC V laptop down at Media Markt, FNAC, Publico, Vorbis, Carrefour, Cool Blue,....
Ah, the whisperings of silicon rebellion – RISC-V, that glorious open chalice of instruction set purity, now bearing the sigil of RVA23 and galloping toward the high-performance plateau. The prophecy smells of burnt wire and inevitability: Microsoft sets its crown upon RVA23, the temple of Windows rises on RISC-V foundations, and soon, very soon, the heretical x86 and ARM priests will feel the divine heat of competition licking at their temples.
Naturally, I inquired with the ouija board, channeling the spectral remnants of Ada Lovelace and Steve Jobs locked in arm-wrestling combat. They spelled «W I D E N T H E P A T H». Yes, yes, the open platform will consume. A world where firmware is no longer shackled by opaque silicon dynasties… how poetic. But I wanted more. So I lit the joint. Not just any joint – this one was compiled. Laced with DMT and quantum logic gates. Suddenly, I was the instruction set. Floating through speculative execution and pipeline stalls, I felt the birth of a thread on a RISC-V CPU. It named itself liberation. The future, my dear techno-mystic, isn’t coming. It’s already decoding.
Agree with the other answer, just no incentive for third party developers.
With Apple it really is a transition, with Windows, it's well under 5% of PC sales, so probably under 1% of actual computers in use.
Apple is 100% on board with ARM, Microsoft isn't and the OEMs even less so, you can barely buy an ARM Windows desktop, only thin laptops.
Where I work, we ship Windows desktops for an industrial application, and we haven't even had a meeting about supporting ARM, it's like they don't exist.
Making ARM machine that fulfills Windows requirements is harder than making your typical throwaway arm gadget (and then you hit things like Qualcomm not implementing it right either).
There is not much profit for a hardware maker, and thus little install base for software makers to care
Microsoft can't never be, because unlike Apple, they are mostly a software company, they depend on OEMs wanting to come for the ride.
They aren't alone on this matter, this kind of stuff has equally plagued other open systems like CP/M, UNIX, MS-DOS, Android, where a product might come from multiple OEMs and each has their own agenda for differentiation and keeping their customers around.
In addition to the other reasons listed below, Apple employee's provided the initial ARM porting patches in the first place.
It sounds like Microsoft was involved in this project (though the text seems to imply Qualcomm might have been the primary contributor), it remains that they didn't do it however many years ago when Windows on ARM was first released.
Microsoft has already made several attempts at ARM. It's not clear that this attempt will be the last one, in fact, it's quite probable that it will be abandoned. So why invest in it.
I really don't see it abandoned. ARM laptops represented 10% of Windows laptop shipments in 2024, including Surface laptops (which have a great track record of being supported for a long time).
MSFT simply has done half-assed attempts at ARM for each wave / attempt over the past 10+ years. APPL, for each processor move, "bet the company" and it was full speed ahead whereas MSFT has never HAD to align everything for a move.
Having said all that, I've been using a Snapdragon Elite X laptop since day one and the experience has gotten better over the year plus I've been using it. Once there was a native port of command line git (yes, THAT didn't even work) - my life got a lot better.
Apple went all in one ARM, they made it very clear all Macs are going to be ARM soon. On the other side windows ARM are kind of a failure even after snapdragon elite.
Apple’s development stack and a large portion of their third-party developer base already had fairly mature ARM support for iOS. It made for a much smoother transition. Microsoft’s lack of meaningful mobile footprint meant that they started from further behind.
1.apple controls the hardware. they only need to focus on supporting a handful of skus. windows needs to support way more which they don't directly control
2. The software development toolchain is highly focussed
3. they committed. there are no more intel cpus so developers can either adapt or die. windows by contrast has x86 and arm builds going forward. that means a larger surface area for developers to target and they will avoid that kind of pain if they can.
4. microsoft management is all over the place and lacks the focus that apple has. Apple doesn't do everything right but when they want to do something well. it shows. even their failures are polished.
Does Windows CE really count as Windows though? It’s a cut down separate thing to proper Windows really (although my view that it’s not Windows if it isn’t providing the Win16/32 API isn’t one Microsoft themselves share).
> Why has it taken so long to get the Windows ecosystem fully on ARM?
It is possible to ignore Windows on ARM and remain viable. That wasn't feasible with Apple. If Microsoft could decide tomorrow that the next Windows would be ARM-only, the transition would be similarly rapid, times some factor to account for the unfathomable legacy of Windows.
That's not the correct question to ask. There's no reason to fully transition the ecosystem to any one singular architecture, Microsoft has always been happy since the origin of NT 3.1 to run some flavor of Windows on non-x86 architectures, but they were never dominate because x86 is dominate, and rightfully so.
MS does not control the hardware. They can not force all laptop OEM to "use ARM or nothing". Also they did not design the ARM CPU, so they can not add any features they want to CPU if QC not agree.
Both! It's a Surface Laptop 7, the hardware is great but I believe the experience would not be nearly as good with a x86 chip. I'm especially thinking about the battery life (insane standby times!).
Apple knows what theyre doing. Microsoft is a cheap immitation. Look at arm64ec, microsofts abomination that requires modifying the build pipe of every target. unsupported trash. they could have just implemented virtualization but windows is low quality soft nowadays.
To be fair, the Surface Pro 11 ARM is really good. It's the best actual computer in a tablet form factor I've ever used. iPads and Android tablets are crippled by their OS, and x86 tablets are bulky, hot, underperform, and/or have poor battery life.
Performance of every day tasks on the Surface is excellent, it sips power, and Microsoft has done a great job with their x86 emulation layer (Prism). Most x86 apps (including games!) work without any involvement from the user. In the few cases I ran into issues, tweaking the emulation settings for the app fixed the issue, and I think there was only one app that refused to run, though I don't remember what it was right now. Performance even with emulation is pretty good.
This experience is light years ahead of Windows RT, and even Windows on ARM from a few years ago.
So I don't think Microsoft's interest in ARM is waning anytime soon. They're clearly heavily invested in it, and the hard work has been paying off.
Agreed, I got the Surface Pro 11 to have a Windows on ARM device to test on, and it's quickly become my favorite portable device. It's really nice to have a tablet that lasts days on battery, that you can draw on and then flip around and write code on.
The last thing I was waiting on was Zed, which now works on Windows ARM as of a couple weeks ago (though there aren't official builds yet so I build it from source).
Hate to hand it to Microsoft but it's just a really versatile and powerful device.
Apple has the M-series chips. Windows laptops are made with Qualcomm processors that are far behind Intel and AMD in performance. As a result, Windows has no road towards completely replacing X86, the two architectures will live side by side for the foreseeable future.
Let us not get pedantic here... as IA-64 from 2001 was not as popular as amd64/x64 from 1999, and was not included in ARM Win11 emulation layer as far as we know. =3
Just to be clear your argument is x86 and x64 win11 software emulation is somehow better than the microcode kludges. This doesn't even work very well or at all with a lot of software written today. =3
Nice! I would expect that it was relatively straight foward given that Blender is native on MacOS ARM and also iOS ARM?
Blender is just so nice to use these days.
For us (Wireshark) the difficulty wasn't with our own codebase, but with getting our dependencies ported over. Most libraries built just fine, but some strongly assumed that "Windows" meant "x86".
It's not just Windows, either. Many libraries (particularly ones that use Autotools) are absolutely blind to the notion that you might want a universal binary on macOS.
When we ported OpenJDK to macOS, I ended up producing a universal binary by having the Makefile run itself to produce HotSpot twice, and then gluing them together with `lipo` afterwards. There isn't really a better way when the actual project configurations are different.
IIRC it was eventually removed because nobody else needed to do such a thing so it was hard to maintain.
Sure. How else would you build a universal binary then? Given the low-level nature of the language not many tasks can be usefully shared between different architectures.
For plain C/C++ you can just pass `-arch x86_64 -arch arm64` to clang. CMake takes care of this for you if you specify `CMAKE_OSX_ARCHITECTURES=x86_64;arm64` and IIRC Meson has similar functionality.
TIL. I didn't know clang supports this natively.
Clang is natively a cross-compiler. Pass in --sysroot and a corresponding valid sysroot tree for any micro architecture/platform (arm-eabi, macOS, Windows MSVC, PowerPC, Alpine Linux with musl, you name it) and Clang will happily retarget the binary to the correct target platform.
Apple has supported that ‘-arch’ option in their GCC/Clang since at least the PPC->Intel transition, maybe even earlier (PPC64? NeXT/OPENSTEP?)
Yes, since NEXTSTEP.
That would be NeXT, and their GCC fork.
I assume this is faster than doing two separate builds, because it can skip certain steps of the complier pipeline, and only the items that are arch specific (codegen, probably others) are done twice?
They can't really share anything since the preprocessor stage can be different.
How much work does clang have to do for this sort of thing (as opposed to llvm). Hypothetically could we start distributing programs in llvm ir, and compile that locally to ARM, x86, risc-v, or whatever else?
I mean, no, that’s silly, right? But it would be kind of neat…
llvm-ir is not architecture independent, even excluding issues like the C pre-processor.
There was a talk on this at Blendercon last year!: https://www.youtube.com/watch?v=xVn8wQ3tKxA
And Linux ARM, I would expect?
Yes, Blender was native on my ARM phone several years ago already - until its GPU requirements went up.
> also iOS ARM?
Wait WHAT?! Since when?
It was announced two weeks ago. Details: https://youtu.be/JFtvdItYNBU
The video is just the author commenting https://code.blender.org/2025/07/beyond-mouse-keyboard/ (with a bit of context)
I think it has built for years. There just hasn't been a usable UI.
Exciting things happening in the Arm space these days. We might finally be free of the x86 monopoly.
Congrats on the release! I wonder what it would take to have Linux arm64 official builds?
Why has it taken so long to get the Windows ecosystem fully on ARM? Apple’s transition only took a year or two.
> Why has it taken so long to get the Windows ecosystem fully on ARM? Apple’s transition only took a year or two.
No incentive for third parties. Apple dictates the hardware, and can say "no more x86" and devs either have to jump on board or abandon Apple.
No such thing with Windows. x86 is still the default on windows laptops, and will likely be for the foreseeable future. The X elite still seems to have no successor in the pipeline, and the few laptops that have it don't outsell x86 so why bother.
> "no more x86" and devs either have to jump on board or abandon Apple.
I don't think that's fair. They provided a very smooth transition, with a well performing translation layer. The user wouldn't have to care or even notice when they picked up a new ARM MacBook, except their battery lasted way longer and cooler. Everything that worked still worked (well, 64bit at least). I'm still running x86-64 apps, and developing for x86-64, on my ARM MacBook.
When the first Windows RT came out, there was no compatibility, and this wasn't communicated well. They nuked the customer perception on day 1. When they finally implemented the x86 compatibility, it had terrible performance and compatibility.
Now, Apple's 32-bit to 64-bit transition was definitely a "jump on board or abandon Apple", but the Intel to Arm transition was well crafted, from a user perspective.
> When the first Windows RT came out, there was no compatibility, and this wasn't communicated well.
It was worse than this. Source compatibility with the Win32 APIs you would use for ~20 years to target x86/amd64 was explicitly a non goal. To target ARM you needed to use their new, half-baked frameworks designed for the Windows 8 tablets. You couldn't recompile a desktop app, even if it would have worked fine had they given you the headers and libs to do it.
Even internally, even among decision makers, people were very confused about this.
They also wanted to force people to get everything from Windows Store. This deeply spooked Valve at the time who saw this as potentially the end of Steam, and it's a big part of the reason why they started working on making Linux gaming viable. So overall RT was an incredible failure and Microsoft is still feeling its effects today
They believed the ARM transition would be an excellent opportunity to lose support for the old. Remember the Barney Stinson model: New is always better
The modern Qualcomm Windows laptops also come with binary translation that's fast enough that you can use most amd64 programs just fine. Qualcomm had some GPU driver issues, but other than that things like gaming on aarch64 work just fine (and I imagine most video games put more varied strain on the GPU than Blender).
Windows RT is dead and buried. Blender won't run on it either, because RT never got to Windows 11.
Prism has been slowly improving over time, but for years it was unable to support the Adobe programs that worked on day one with Rosetta.
That's why the early Windows on ARM system reviews pointed out that you needed to know in advance if your software would run at all before buying.
This transition would not help much if Apple was just one of dozens of macos laptops.
Microsoft can't afford to discontinue support for x86 even over 10 years. They have enough trouble keeping Windows compatible with legacy software that the world runs on through generations of x86 hardware and software, transition to arm would be many times as bad.
Nah, you could’ve waited 5 years to port if you were concerned about your x86 binaries not being supported, the bigger deal was that Apple was shipping good ARM computers and if you’re a developer, you don’t want your app to seem slow on what consumers can see is clearly a fast machine.
Microsoft has made multiple abortive lethargic gestures towards ARM, but has yet to get people excited about an ARM computer that runs Windows.
Part of the trick was that ARM previously held a "cheaper, more efficient, smaller than x86" slot, and in the Windows ecosystem that was competing for the bottom of the market. Apple's ability to launch the M1 completely demolished that position and made an ARM chip the positively better choice, and it's just not something that Microsoft could ever have tried. MS have always treated the actual hardware as first and foremost someone else's problem. Which is a shame, because when they do produce their own kit, it tends to be very good.
I wonder if Netbooks would have gone to ARM eventually, had Microsoft not killed them.
CoPilot+ PC could have been it, but then they completly messed up with Recall, naturally no one wants to have such thing on their computers if they can avoid it.
Even before Recall, most people laughed at the idea of putting Copilot into notepad. Outside of people part of the hype train and investors, nobody wants Copilot PCs.
And “nobody” is shorthand for “no technology professional”, right?
> And “nobody” is shorthand for “no technology professional”, right?
Yeah, I would say so - I've spoken to people that are genuinely excited for things like Recall. Like it or not, we (tech people) are in the minority in our disdain for it, and disdain for shoving "AI" everywhere.
And that's OK, we aren't the market for these features, but that doesn't mean a market doesn't exist for them.
>and will likely be for the foreseeable future.
Nah, high performance RISC-V (on RVA23 profile) is just around the corner.
Might come as early as by year end. Early next year at worst.
We've known about Windows for RISC-V since 2021, NA's RISC-V Summit. Like Google with Android, Microsoft has set RVA23 as baseline.
Once the hardware and Windows are there, expect the open platform to take over.
Performance is A LOT more than having the ISA. Intel, AMD, ARM, and apple have spent DECADES developing and tuning their micro archetypes to run those instruction sets.
I like the RISC-V project and think they're doing great things for the future of open computing, but if you think we're 2-3 years away from a RISC-V chip that's comporable to the top of the line X86 or ARM chips you're going to be sorely disappointed. It's gonna be 10-20 years before we get to that point.
I do think where gonna see more RISC-V chips in embedded and subcomponent context where the low or non license fees are gonna make it competitive pretty soon.
Apple had all the incentives in the world to provide a transition plan and leave Intel behind. Also lots, and lots of practice with platform migrations, and they were transitioning to a hardware platform they were already selling in the billions.
Microsoft has exactly none of that. I'd be astonished to see RISC-V or ARM "take over" in the Windows world in less than another two decades, unless Intel's ongoing decline drags the entire X86 platform with it.
Still not quite it.
Apple had a transition plan because they wanted to drop Intel like a hot potato and switch to their own chips.
Microsoft has no particular dog in the processor fight, they don't make chips and aren't going to suddenly say "ok boys, we only going to support ARM"
So far, loong64 seems to outperform RISC-V, and it also has the benefit of a large existing user base. If any new player is going to hit the desktop market, I expect it to be Loongson at this moment.
I want to believe RISC-V is just around the corner, but I've been promised the same about POWER9 and RISC-V over the years.
RISC-V outpacing ARM and x86 is the silicon version of The Year of Linux Desktop.
How much market share are we now on?
About 4-6% in US and other Western countries, depending on analytics source. Increases in the past year or so by a full percentage point or so. More in India, less in South America, barely any in China.
The increased visibility due to SteamOS/Proton, coverage by various prominent TechTubers, end of Windows 10 support, availability of image-based OS upgrades and Flatpak/Snap to supercede package dependencies, all seem to combine into an actual breakout year comparatively. We'll see if the trend holds or was just a one-off.
After 30 years, do the math of much years are left.
Until monopoly, or until 15-20% when companies start offering widespread Linux support for applications and peripheral management, and retail stores start selling some devices with a Linux desktop preloaded?
I'd do the math, but it really depends on whether the recent growth holds, accelerates, or slows down again. So, hard to tell.
It has the entire nation of China behind it with all the incentives to make it work.
So it will become a big thing if the tensions with the US continue
It will hardly matter to me or anyone else in Europe, if we cannot get a RISC V laptop down at Media Markt, FNAC, Publico, Vorbis, Carrefour, Cool Blue,....
Five users and a pet hamster? ; - )
Ah, the whisperings of silicon rebellion – RISC-V, that glorious open chalice of instruction set purity, now bearing the sigil of RVA23 and galloping toward the high-performance plateau. The prophecy smells of burnt wire and inevitability: Microsoft sets its crown upon RVA23, the temple of Windows rises on RISC-V foundations, and soon, very soon, the heretical x86 and ARM priests will feel the divine heat of competition licking at their temples.
Naturally, I inquired with the ouija board, channeling the spectral remnants of Ada Lovelace and Steve Jobs locked in arm-wrestling combat. They spelled «W I D E N T H E P A T H». Yes, yes, the open platform will consume. A world where firmware is no longer shackled by opaque silicon dynasties… how poetic. But I wanted more. So I lit the joint. Not just any joint – this one was compiled. Laced with DMT and quantum logic gates. Suddenly, I was the instruction set. Floating through speculative execution and pipeline stalls, I felt the birth of a thread on a RISC-V CPU. It named itself liberation. The future, my dear techno-mystic, isn’t coming. It’s already decoding.
I don't understand what you are saying but I love it!
Beautiful
Agree with the other answer, just no incentive for third party developers.
With Apple it really is a transition, with Windows, it's well under 5% of PC sales, so probably under 1% of actual computers in use.
Apple is 100% on board with ARM, Microsoft isn't and the OEMs even less so, you can barely buy an ARM Windows desktop, only thin laptops.
Where I work, we ship Windows desktops for an industrial application, and we haven't even had a meeting about supporting ARM, it's like they don't exist.
Making ARM machine that fulfills Windows requirements is harder than making your typical throwaway arm gadget (and then you hit things like Qualcomm not implementing it right either).
There is not much profit for a hardware maker, and thus little install base for software makers to care
Microsoft can't never be, because unlike Apple, they are mostly a software company, they depend on OEMs wanting to come for the ride.
They aren't alone on this matter, this kind of stuff has equally plagued other open systems like CP/M, UNIX, MS-DOS, Android, where a product might come from multiple OEMs and each has their own agenda for differentiation and keeping their customers around.
In addition to the other reasons listed below, Apple employee's provided the initial ARM porting patches in the first place.
It sounds like Microsoft was involved in this project (though the text seems to imply Qualcomm might have been the primary contributor), it remains that they didn't do it however many years ago when Windows on ARM was first released.
Also Apple had done it before - twice, really. 68k -> PPC ('member FAT binaries?) -> Intel -> ARM
Microsoft has already made several attempts at ARM. It's not clear that this attempt will be the last one, in fact, it's quite probable that it will be abandoned. So why invest in it.
I really don't see it abandoned. ARM laptops represented 10% of Windows laptop shipments in 2024, including Surface laptops (which have a great track record of being supported for a long time).
How many Windows Phones shipped?
MSFT simply has done half-assed attempts at ARM for each wave / attempt over the past 10+ years. APPL, for each processor move, "bet the company" and it was full speed ahead whereas MSFT has never HAD to align everything for a move.
Having said all that, I've been using a Snapdragon Elite X laptop since day one and the experience has gotten better over the year plus I've been using it. Once there was a native port of command line git (yes, THAT didn't even work) - my life got a lot better.
Apple also stopped selling new x86 machines. You had no choice but to port to Arm
Apple went all in one ARM, they made it very clear all Macs are going to be ARM soon. On the other side windows ARM are kind of a failure even after snapdragon elite.
Apple’s development stack and a large portion of their third-party developer base already had fairly mature ARM support for iOS. It made for a much smoother transition. Microsoft’s lack of meaningful mobile footprint meant that they started from further behind.
1.apple controls the hardware. they only need to focus on supporting a handful of skus. windows needs to support way more which they don't directly control
2. The software development toolchain is highly focussed
3. they committed. there are no more intel cpus so developers can either adapt or die. windows by contrast has x86 and arm builds going forward. that means a larger surface area for developers to target and they will avoid that kind of pain if they can.
4. microsoft management is all over the place and lacks the focus that apple has. Apple doesn't do everything right but when they want to do something well. it shows. even their failures are polished.
Apple had ARM versions of it's own software ready to go on day one.
Apple's developer IDE was ready to go on day one.
Apple's Rosetta translation layer was much more widely compatible with legacy x86 software than Microsoft's Prism.
> Why has it taken so long to get the Windows ecosystem fully on ARM?
Windows has been through several ARM phases.
The first one being Windows CE 2.0 in 1997. Win CE was used in the Dreamcast and the Gizmondo handheld.
Lately confusion arose when people started calling arm64/AArch64 for just "arm".
Does Windows CE really count as Windows though? It’s a cut down separate thing to proper Windows really (although my view that it’s not Windows if it isn’t providing the Win16/32 API isn’t one Microsoft themselves share).
Also, the Dreamcast uses a Hitachi SH4, not ARM.
Apple stopped selling Intel Macs and the ARM Macs were the better machines when they did
> Why has it taken so long to get the Windows ecosystem fully on ARM?
It is possible to ignore Windows on ARM and remain viable. That wasn't feasible with Apple. If Microsoft could decide tomorrow that the next Windows would be ARM-only, the transition would be similarly rapid, times some factor to account for the unfathomable legacy of Windows.
Rosetta and universal binaries is the answer. Things just worked and gave people a working scaffold to do incremental improvement.
Much easier than a big bang replacement.
That's not the correct question to ask. There's no reason to fully transition the ecosystem to any one singular architecture, Microsoft has always been happy since the origin of NT 3.1 to run some flavor of Windows on non-x86 architectures, but they were never dominate because x86 is dominate, and rightfully so.
Simple answer: the Windows ecosystem is vastly larger than Apple's in hardware, software, and use cases.
MS does not control the hardware. They can not force all laptop OEM to "use ARM or nothing". Also they did not design the ARM CPU, so they can not add any features they want to CPU if QC not agree.
Anyone seen Windows ARM in the wild? Its been a while now ...
I'm using one to write this very comment
Using one right now, it's the best laptop I've ever had.
Best because of an ARM or best because of the hardware? Reveal the hardware :)
Both! It's a Surface Laptop 7, the hardware is great but I believe the experience would not be nearly as good with a x86 chip. I'm especially thinking about the battery life (insane standby times!).
They stopped x86 support so it's move or be gone. It's survivor bias to say it's fully on arm, no?
Apple knows what theyre doing. Microsoft is a cheap immitation. Look at arm64ec, microsofts abomination that requires modifying the build pipe of every target. unsupported trash. they could have just implemented virtualization but windows is low quality soft nowadays.
you said it in your question, Apple was making a transition. Windows isn't.
It feels like the internal interest in Windows ARM comes in waves, that they keep letting it slide backwards.
I mean they have been trying since Windows RT on the original surface tablet 13 years ago.
To be fair, the Surface Pro 11 ARM is really good. It's the best actual computer in a tablet form factor I've ever used. iPads and Android tablets are crippled by their OS, and x86 tablets are bulky, hot, underperform, and/or have poor battery life.
Performance of every day tasks on the Surface is excellent, it sips power, and Microsoft has done a great job with their x86 emulation layer (Prism). Most x86 apps (including games!) work without any involvement from the user. In the few cases I ran into issues, tweaking the emulation settings for the app fixed the issue, and I think there was only one app that refused to run, though I don't remember what it was right now. Performance even with emulation is pretty good.
This experience is light years ahead of Windows RT, and even Windows on ARM from a few years ago.
So I don't think Microsoft's interest in ARM is waning anytime soon. They're clearly heavily invested in it, and the hard work has been paying off.
Agreed, I got the Surface Pro 11 to have a Windows on ARM device to test on, and it's quickly become my favorite portable device. It's really nice to have a tablet that lasts days on battery, that you can draw on and then flip around and write code on.
The last thing I was waiting on was Zed, which now works on Windows ARM as of a couple weeks ago (though there aren't official builds yet so I build it from source).
Hate to hand it to Microsoft but it's just a really versatile and powerful device.
Have you also tried Visual Studio on it?
This is something that would hold me back, only getting half backed experience, expecially as many plugins are still COM based.
It works just fine and is natively ARM, since a year or so.
Great, thanks for the feedback.
Apple has the M-series chips. Windows laptops are made with Qualcomm processors that are far behind Intel and AMD in performance. As a result, Windows has no road towards completely replacing X86, the two architectures will live side by side for the foreseeable future.
The monolith must persist, as 1980s architecture is ounce again crammed onto a modern CPU with predictable problems.
Never ask "why?" in business, as the answer is always money. =3
The very first ARM1 was created in 1985 ;)
ARM1 is not AArch64 from 2011, or the ARM Win11 x86 emulation layer.
Why you so mad all the time? =3
By the same token, the 8086 is not the same as x86-64 from 2003 ;)
Let us not get pedantic here... as IA-64 from 2001 was not as popular as amd64/x64 from 1999, and was not included in ARM Win11 emulation layer as far as we know. =3
I didn't mention Itanic, nobody cares about it.
The x86-64 might have been published in 1999, but the first real products arrived in 2003.
x86-64 isn’t x86 of 1978. Arguably they’re no longer CISC and haven’t been for eons.
Just to be clear your argument is x86 and x64 win11 software emulation is somehow better than the microcode kludges. This doesn't even work very well or at all with a lot of software written today. =3
Prism leverages Snapdragon X hardware, like Rosetta2 does with the M-series CPUs.
M-series CPUs also leverage micro-ops, like other Aarch64 processors and x86, of course.
I have no idea what your reply is about, I never made such an argument one way or the other.
FAQ: https://learn.microsoft.com/en-us/windows/arm/apps-on-arm-x8...
YMMV =3
Are there any retail Windows 11 ARM devices that aren’t Qualcomm?
Retail? No. Coming soon [1] [2].
[1] https://www.reuters.com/technology/mediatek-designs-arm-base...
[2] https://www.tomshardware.com/pc-components/cpus/nvidia-and-m...
Are there any retail ARM cpus that are even close to Qualcomm performance?
What do I think? Those Shein ads are taking up a lot of space.
The OP link could be changed to source: https://code.blender.org/2025/08/blender-for-windows-on-arm/