He's probably referring to other products, but it seems he overlooked the SBC, which come in all forms and sizes and, generally, being low power devices, can't really do much besides hosting some not so demanding apps (looking at containers).
or the myriad of SBC-based computers like this. The only thing you should do is to install an OS, which is something everybody tends to do and succeed nowadays.
I fully agree with you, but it's still a PC, although not envisioned for the power users, but for the people who just need to do the most basic of stuff (e-mails, maybe a few documents, etc.), that's what a ~6W CPU can deliver.
Our school computer labs had BBC Micros, the predecessor to the Acorn Archimedes and all ARM processors.
It’s funny to think back at how quickly that computer was out of date, we only had it a year or so before getting a Windows 95 PC. At about the same time we came back from a school break and all the BBCs were gone, Windows PCs in their place. Now we are all carrying around an ARM in our pockets.
Being from the Cambridge UK area it’s amazing the heritage Acorn produced. About 10 years ago I worked on a project that was being run by the original lead developer of RISC OS. The connections are everywhere.
It's a game of definitons, but for a proper desktop PC, wouldn't you need a socketable CPU? Mainboard, CPU and probably GPU as separate units doesn't seem to be something the ARM world is aiming at. (Neither does Apple, but they've been blurring the lines between laptops and desktops for a long time)
Running Windows 11 ARM in Parallels on my M1 MBP is a surprisingly smooth experience. Most x86 applications work well (both in terms of reliability and performance) using Microsoft's inbuilt emulation. I have not run any games or larger applications in the VM, but I suspect most Windows users could run on ARM and not notice the difference.
That web page is insane - normally I sympathise with the 'scrolljacking' bemoaners, but that's just so funny (maybe it would annoy me if I was actually shopping), not at all what I expected from the page, or Apple at all.
It's at least as annoying as your initial impulse said it was...more so from the people who dictate UI down to a gnat's ass in every app we submit to their store, but somehow exempt themselves from even common courtesy when it comes to designing a page. I'm typing this on a mac, but for god's sake don't make excuses for them.
I'm not making excuses for it, I'm typing this from a ThinkPad (work, and personally have a Framework & a desktop I built, Android phone), it just amused me.
It's just an input right? Forget your expectation that an input causes a 'page' to move vertically; the screen is just a canvas that can change over time or in response to inputs.
(But, because it breaks your expectation is why people don't like it, call it 'scrolljacking' i.e. hijacking scroll behaviour, etc. It's worst IMO when it's done to make the page 'sticky' at certain banner points, so it takes unintuitively more or less scroll movement at certain points as you move down the page.)
MacOS never supported Blu-ray, and Steve Jobs described the licensing of it as "a bag of hurt." This was still ~2008 when the licensing of H.264, VC1, MPEG-2, Blu-ray Patents, AACS, BD+, Cinavia, and all the stuff that makes it work was still in the air. Arguably it never came down[1].
However, I would still beg Apple to reconsider this, even a decade later, at least for desktop professional use. Blu-ray discs come in 25GB, 50GB, 100GB, and even (rare) 128GB disc sizes. It's still a really cheap, portable, distributable, long-term (as long as you check every ~5 years) storage medium.
[1]Blu-ray discs for movies were pretty shamefully designed. Do we seriously need 3 different codecs? Or 3 different copy-protection mechanisms?
I’d suggest that an SSD makes a far better option. For the price of a burner and blank media, you can buy a whole bunch of cheap 128gb ssds. They’re re-writeable, durable and easy to connect with a twenty dollar usb dongle.
The catch with SSDs is they are not long-term reliable when left without power. If left as little as 12 months without energy, data degradation can occur. SSDs need power regularly to automatically refresh their memory cells. No power, no refresh, slow loss. They are almost like RAM in that regard, just with a really slow discharge.
Compared to good quality writable media which is designed for storage without power. This, and the write-once versions (no ransomware), are great for backups of faster modern media.
I have an old MacBook with a cd/dvd-r drive and SD card reader in it. I literally keep this machine just to use these features. I never understood the incentive for getting rid of them. The cd/dvd drive was something I used every day when that was my primary computer.
I just don't think many people have any DVD media anymore, let alone any CDs (!) outside of enthusiasts.
> I never understood the incentive for getting rid of them.
Can't be that much of a mystery? Less weight, less space, less cost, less clunky mechanical stuff - given almost nobody would ever use them it's a no-brainer.
> I just don't think many people have any DVD media anymore
Don't you think that's sort of a self-fulfilling prophecy? DVDs are absolutely not obsolete, and are still manufactured. Lots of content exists on DVD but not streaming. We get movies and shows at the library all the time that are not on any streaming platform.
Sure, everybody has a computer built only to stream, so streaming is the default for anybody who doesn't want to carry an external drive in their laptop bag. Design decisions influence how your product is used. Surprise, people who have phones without headphones jacks are less likely to use wired headphones.
But I already have an external blu-ray drive that connects via USB, and I don't ever need audio beyond the headphone jack. Those who do can use the USB-C ports for that.
Maximum flexibility via USB, without a bunch of ports and parts that will be unused by most people, please.
> why ARM-based complete desktop PCs are not available. What are the blockers?
Apple has shown what is possible with an ARM-based design, and everyone is now clamoring for M1/M2-level CPUs everywhere.
Somehow everyone forgot that it took Apple 15 or more years to get to M1/M2.
Their first custom CPU, the A4, appeared in 2010, and it takes anywhere from 4 to 6 years to design a new CPU. So it's quite likely Apple started designing their own CPUs... around the time they transitioned from PowerPC to Intel, since before the launch of the very first iPhone.
There. That is the answer to "why ARM-based complete desktop PCs are not available". You need to spend 15 years or more actually working on one, and the only people who have done it are Apple.
Or perhaps other players don't have the incentive to do it? In addition to the development time and cost, you also need to make customers willing to buy it. For Apple, they have tight control over the ecosystem so they can make the transition really smooth, but for others this may not be that easy. The way I see this is that, customers already using Apple product will not give up using it if the switch is not too painful, while Windows users will not switch to using some new architecture unless there is some significant gain for them.
Very true. Apple could envision a path of "start with mobile CPUs and see if we can make them better and better, and then eventually laptops, and then maybe desktops". And they have their own OSes on those devices.
Everybody else has only parts of the whole picture. Samsung is probably the only one that has all the hardware, and CPU expertise, but they control none of the OSes.
I run several “windows only” apps on it as well as native Mac apps. Parallels + Windows-for-arm + Microsoft’s x86 emulation isn’t as smooth as Rosetta 2, but it’s still very very usable.
Running Altium[1] as I type, with a reasonably complex PCB in design (8 layers, DDR3, FPGA, HDMI) and it’s indistinguishable from running on a high-end PC. Said PC has been powered down for several months now, I just haven’t felt the need to use it.
Running similar configuration Altium on Macbook Air M1 (16GB). Not smooth, but can get things done. Well, except some keys not working properly (Shift; and having to remap Delete)
I don't think there are any technical blockers but there isn't much in the way of product market fit at this point. Desktop users tend to be less concerned with e.g perf/watt and are more concerned with perf/$ which Intel and AMD compete ferociously in.
I think there's also a lack of incentive to come up with new desktop solutions specifically that might be a better fit as well because desktop PC is seen as a shrinking market. Why do a bunch of R&D to compete in a shrinking market along with the difficulty of breaking into the established market, along with all the software compatibility issues etc.?
Technically a laptop and not a desktop PC, but the strengths of ARM are much more apparent in power and heat constrained mobile settings. As long as ARM hasn't even gained a solid foothold in laptops (outside Apple), what's the point of putting it in a form factor where it will be even less attractive.
Searched for how performant these chips (Qualcomm Snapdragon 8cx Gen 3 SC8280XP) were and summary is that they're okay: similar to Intel's i5-1130G7, but no match for Apple's M1.[1]
The blocker is Windows backward-compatibility. Macs are only sold with ARMs these days. Many (most?) Chromebooks are also using ARM. But one of the main reasons for Windows is backward-compatibility, and you won't get it with ARM.
No it is not. I have Windows 11 on a test machine Samsung Galaxy Book with Snapdragon 7c and can run x86/x64 apps just fine. It is a fine device, but honestly it does not offer any killer feature compared to x86 notebooks.
I have a Huawei Kunpeng 920 based ARM desktop/workstation running Ubuntu 20.04 LTS. It has 8 cores running at 2.6GHz. Picked the processor + motherboard combo up online for $300 USD, added a WD SN640 7.68T SSD and 32GB DDR4 ECC RAM, connected to my home network with a Mellanox-4 40G NIC. 7z benchmark shows it has comparable multicore performance with a intel 6770 4 cores processor.
There are 24 cores Kunpeng 920 boards available at roughly the same price. The problem is those boards only have two PCI-E slots rather 3 slots on mine.
Really happy with the one I got, as it gives me an inexpensive platform for building & testing ARM64 software, it also give me the ability to install SSDs/NICs that won't be supported by a Mac.
I’d say the blocker is Windows and the nonstandard SoC nature of the ARM ecosystem.
An OEM can't just take an ARM CPU, stick it omn a board together with a chipset with UFI on it, bundle a Windows CD and call it a day. The ARM ecosystem is full of weird hacks like the raspberry pi being bootstrapped by the GPU.
Add onto that the rumored Microsoft-Qualcomm exclusivity deal and it just makes the barrier to entry insurmountable. For what? You don't even care about battery life.
It's not that there's one or two blockers that just need to get cleared out of the way. It's that ARM desktop PCs don't have solid momentum and commitment.
When Apple announced they were moving to ARM, it was "this is happening and everyone needs to get on board or they're going to be left behind." Microsoft has ARM support in Windows including an x86 translation layer, but there's certainly no "get on board or you'll be left behind" ultimatum. Microsoft intends to keep supporting x86 and in fact x86 will continue to be the premier platform for Windows. That leaves desktop ARM without momentum.
Some things in the world are about momentum and the commitments people are willing to make given the commitments of others. Every Mac developer knew that time invested in porting to ARM would pay off. Apple was committed to ARM. They'd spent a decade making their own chips. They weren't just going to abandon ARM a few years later. Developers knew that competing developers would port to Mac-ARM so they had to make that commitment as well. Developers knew that Mac buyers were going to be clamoring for the Mac-ARM machines.
Does Microsoft care about Windows on ARM? Do Windows Developers? Do PC buyers? The answer to all three seems to be "no" (or at best "eh, a bit"). Given that no one is committing to Windows-ARM, everyone just takes a wait-and-see approach. Without the commitment and direction, others don't want to make those commitments. Would you buy an ARM PC knowing that everything will likely need to run through translation layers? Knowing that Microsoft's support for ARM doesn't even extend to flagship products like Visual Studio? Knowing that so many third-party devs won't see enough ARM users to justify the port? And if you're not going to commit to ARM, why should anyone else in the ecosystem.
Apple solved that indecision/commitment issue by 1) saying unreservedly that it was moving to ARM and would be leaving Intel entirely; 2) offering chips that didn't just match Intel performance, but solidly beat it offering users a huge reason to want those new machines fast.
Microsoft doesn't have either of those and Intel is making huge strides. They've released big-little architecture chips and started improving performance a lot. They're likely to start closing the gap on chip process either by renting TSMC or improving in-house fabs. Plus, "desktop" PC users (rather than laptop) often don't care about heat and form factor like Apple does - they aren't looking for a Mac mini or Mac Studio. They're happy with a tower that they can stick a giant graphics card in and that can accommodate a 100W CPU.
Beyond the issue of commitment, as others have pointed out, Apple is just farther ahead on ARM performance than anyone else. Qualcomm is trying to make their own custom chips to improve performance, but right now Intel has better performance and Intel seems recommitted to maintaining their performance advantage (against non-Apple chips and at least wants to bluster about beating Apple).
Given this environment, it's hard to be the one willing to invest the money to break the logjam. Maybe Qualcomm will if it can leverage its investment in mobile CPUs for desktops. However, it's hard for a chip startup to want to take on Intel when users, developers, and Microsoft aren't enthusiastic about Windows-ARM. It's hard for a user to want an ARM PC when performance isn't dramatically better, developers don't care, etc. It's hard for developers to care without users caring. But I don't think even Qualcomm can break the logjam. I think they like their margins too high to want to price their chips at a level that might see people clamoring for their processors and I think Intel's 12th-gen (and successive generation) chips will make it hard to beat them on performance.
Apple just controlled enough of the ecosystem to offer the amoun...
I think you've hit on an important point. What does PC mean? Obviously not "something in the lineage of the IBM PC," since we're talking about an ARM desktop PC. And we're not talking about off the shelf computers that have ARM chips, because we have the Mac and Chromebooks.
I'm guessing it probably means something like - why isn't there the equivalent of the x86 hardware world, but using ARM? Why can't I buy a CPU and motherboard with modern standard interconnects that I can use for a video card, SSD, etc; or maybe get a beige box computer, but it's ARM-based?
The holdup has been that Qualcomm and Microsoft had a deal that ARM Windows would only run on Qualcomm hardware. That ended in March, but Microsoft still seems to be holding to it. Evidence: they will still not let me buy a license for ARM Windows for a Mac Studio VM. I am using a “Developer’s Preview” that is identical to x86 Windows 11
So the holdup is that until Qualcomm bought Nuvia, their own designs were not good. The Surface Pro X sucked. The Nuvia-Snapdragons should be the turning point, but also ARM is suing Qualcomm over that so who knows.
95 comments
[ 3.3 ms ] story [ 48.0 ms ] thread[0] https://www.youtube.com/watch?v=qfv6Ah_MVJU
If your definition of PC is "runs Windows" there's a business reason Apple silicon doesn't.
It's rumoured to be because Microsoft made a deal with Qualcomm: https://www.macrumors.com/2021/11/22/microsoft-qualcomm-arm-...
So if you want to buy (or build...) an arm PC with Windows then as of today it had to come with a Qualcomm SOC.
So I think the TL;DR answer for your question is really "because Qualcomm aren't super interested in the desktop market".
I think there are now some "mini PCs" but they are not internally expandable etc.
There are no major technical blockers to slapping a Qualcomm dev kit in a desktop case but the experience is not gonna be that great.
SBC is not a complete PC. I was talking about one which an end user who is not tech-savvy might use.
or the myriad of SBC-based computers like this. The only thing you should do is to install an OS, which is something everybody tends to do and succeed nowadays.
It feels like eeePCs of yesteryear. Underpowered and slow
https://en.wikipedia.org/wiki/Phoebe_(computer)
Our school computer labs had BBC Micros, the predecessor to the Acorn Archimedes and all ARM processors.
It’s funny to think back at how quickly that computer was out of date, we only had it a year or so before getting a Windows 95 PC. At about the same time we came back from a school break and all the BBCs were gone, Windows PCs in their place. Now we are all carrying around an ARM in our pockets.
Being from the Cambridge UK area it’s amazing the heritage Acorn produced. About 10 years ago I worked on a project that was being run by the original lead developer of RISC OS. The connections are everywhere.
With linux you might be partially ok, but with Windows you are not ok.
It's not the fault of the OS.
I always thought, by definition, it had to be intel 8088 compatible to be a PC.
The rest of the industry seems fine with what AMD and Intel has at the moment, or incapable of making strong ARM-based competitors.
(But, because it breaks your expectation is why people don't like it, call it 'scrolljacking' i.e. hijacking scroll behaviour, etc. It's worst IMO when it's done to make the page 'sticky' at certain banner points, so it takes unintuitively more or less scroll movement at certain points as you move down the page.)
... and 2000 $
For this price a Bluray drive and some audio I/O would have been nice.
However, I would still beg Apple to reconsider this, even a decade later, at least for desktop professional use. Blu-ray discs come in 25GB, 50GB, 100GB, and even (rare) 128GB disc sizes. It's still a really cheap, portable, distributable, long-term (as long as you check every ~5 years) storage medium.
[1]Blu-ray discs for movies were pretty shamefully designed. Do we seriously need 3 different codecs? Or 3 different copy-protection mechanisms?
Consider the source.
Compared to good quality writable media which is designed for storage without power. This, and the write-once versions (no ransomware), are great for backups of faster modern media.
0. http://www.pricegrabber.com/blu-ray-external/products/?sort=...
And you can send audio over USB. You can't possibly have more than 10 Gbps of audio?
> I never understood the incentive for getting rid of them.
Can't be that much of a mystery? Less weight, less space, less cost, less clunky mechanical stuff - given almost nobody would ever use them it's a no-brainer.
Don't you think that's sort of a self-fulfilling prophecy? DVDs are absolutely not obsolete, and are still manufactured. Lots of content exists on DVD but not streaming. We get movies and shows at the library all the time that are not on any streaming platform.
Sure, everybody has a computer built only to stream, so streaming is the default for anybody who doesn't want to carry an external drive in their laptop bag. Design decisions influence how your product is used. Surprise, people who have phones without headphones jacks are less likely to use wired headphones.
But I already have an external blu-ray drive that connects via USB, and I don't ever need audio beyond the headphone jack. Those who do can use the USB-C ports for that.
Maximum flexibility via USB, without a bunch of ports and parts that will be unused by most people, please.
Apple has shown what is possible with an ARM-based design, and everyone is now clamoring for M1/M2-level CPUs everywhere.
Somehow everyone forgot that it took Apple 15 or more years to get to M1/M2.
Their first custom CPU, the A4, appeared in 2010, and it takes anywhere from 4 to 6 years to design a new CPU. So it's quite likely Apple started designing their own CPUs... around the time they transitioned from PowerPC to Intel, since before the launch of the very first iPhone.
There. That is the answer to "why ARM-based complete desktop PCs are not available". You need to spend 15 years or more actually working on one, and the only people who have done it are Apple.
Everybody else has only parts of the whole picture. Samsung is probably the only one that has all the hardware, and CPU expertise, but they control none of the OSes.
I'll repost my own comment in a previous thread:
ARM outside of Macs are not beefy. Excluding apples cpus, x86_64 is ahead in perf/per watt. Remember that until 3 months ago, Apple was one CPU node ahead: https://en.wikipedia.org/wiki/5_nm_process#Commercialization
It contains an ATX motherboard [1] with a StrongARM SA110 CPU, NVIDIA NV1 graphics card, PCI network card, ATA hard disk. Mine runs NetBSD/cats.
[1] http://www.simtec.co.uk/products/boards.html
https://git.kernel.org/pub/scm/linux/kernel/git/soc/soc.git/...
I get that the FYI directed it to you, but given the HN guidelines I read that more as PSA.
I run several “windows only” apps on it as well as native Mac apps. Parallels + Windows-for-arm + Microsoft’s x86 emulation isn’t as smooth as Rosetta 2, but it’s still very very usable.
Running Altium[1] as I type, with a reasonably complex PCB in design (8 layers, DDR3, FPGA, HDMI) and it’s indistinguishable from running on a high-end PC. Said PC has been powered down for several months now, I just haven’t felt the need to use it.
[1] https://www.altium.com/altium-designer/
I think there's also a lack of incentive to come up with new desktop solutions specifically that might be a better fit as well because desktop PC is seen as a shrinking market. Why do a bunch of R&D to compete in a shrinking market along with the difficulty of breaking into the established market, along with all the software compatibility issues etc.?
Technically a laptop and not a desktop PC, but the strengths of ARM are much more apparent in power and heat constrained mobile settings. As long as ARM hasn't even gained a solid foothold in laptops (outside Apple), what's the point of putting it in a form factor where it will be even less attractive.
https://www.lenovo.com/us/en/p/laptops/thinkpad/thinkpadx/th...
[1] https://www.androidauthority.com/snapdragon-8cx-gen-3-intel-...
https://blogs.windows.com/windows-insider/2020/12/10/introdu...
There are 24 cores Kunpeng 920 boards available at roughly the same price. The problem is those boards only have two PCI-E slots rather 3 slots on mine.
Really happy with the one I got, as it gives me an inexpensive platform for building & testing ARM64 software, it also give me the ability to install SSDs/NICs that won't be supported by a Mac.
you need to install that Chinese only app on your phone to use it.
Plus any apple machine and the thinkpad x13s
And the raspberry pi
An OEM can't just take an ARM CPU, stick it omn a board together with a chipset with UFI on it, bundle a Windows CD and call it a day. The ARM ecosystem is full of weird hacks like the raspberry pi being bootstrapped by the GPU.
Add onto that the rumored Microsoft-Qualcomm exclusivity deal and it just makes the barrier to entry insurmountable. For what? You don't even care about battery life.
When Apple announced they were moving to ARM, it was "this is happening and everyone needs to get on board or they're going to be left behind." Microsoft has ARM support in Windows including an x86 translation layer, but there's certainly no "get on board or you'll be left behind" ultimatum. Microsoft intends to keep supporting x86 and in fact x86 will continue to be the premier platform for Windows. That leaves desktop ARM without momentum.
Some things in the world are about momentum and the commitments people are willing to make given the commitments of others. Every Mac developer knew that time invested in porting to ARM would pay off. Apple was committed to ARM. They'd spent a decade making their own chips. They weren't just going to abandon ARM a few years later. Developers knew that competing developers would port to Mac-ARM so they had to make that commitment as well. Developers knew that Mac buyers were going to be clamoring for the Mac-ARM machines.
Does Microsoft care about Windows on ARM? Do Windows Developers? Do PC buyers? The answer to all three seems to be "no" (or at best "eh, a bit"). Given that no one is committing to Windows-ARM, everyone just takes a wait-and-see approach. Without the commitment and direction, others don't want to make those commitments. Would you buy an ARM PC knowing that everything will likely need to run through translation layers? Knowing that Microsoft's support for ARM doesn't even extend to flagship products like Visual Studio? Knowing that so many third-party devs won't see enough ARM users to justify the port? And if you're not going to commit to ARM, why should anyone else in the ecosystem.
Apple solved that indecision/commitment issue by 1) saying unreservedly that it was moving to ARM and would be leaving Intel entirely; 2) offering chips that didn't just match Intel performance, but solidly beat it offering users a huge reason to want those new machines fast.
Microsoft doesn't have either of those and Intel is making huge strides. They've released big-little architecture chips and started improving performance a lot. They're likely to start closing the gap on chip process either by renting TSMC or improving in-house fabs. Plus, "desktop" PC users (rather than laptop) often don't care about heat and form factor like Apple does - they aren't looking for a Mac mini or Mac Studio. They're happy with a tower that they can stick a giant graphics card in and that can accommodate a 100W CPU.
Beyond the issue of commitment, as others have pointed out, Apple is just farther ahead on ARM performance than anyone else. Qualcomm is trying to make their own custom chips to improve performance, but right now Intel has better performance and Intel seems recommitted to maintaining their performance advantage (against non-Apple chips and at least wants to bluster about beating Apple).
Given this environment, it's hard to be the one willing to invest the money to break the logjam. Maybe Qualcomm will if it can leverage its investment in mobile CPUs for desktops. However, it's hard for a chip startup to want to take on Intel when users, developers, and Microsoft aren't enthusiastic about Windows-ARM. It's hard for a user to want an ARM PC when performance isn't dramatically better, developers don't care, etc. It's hard for developers to care without users caring. But I don't think even Qualcomm can break the logjam. I think they like their margins too high to want to price their chips at a level that might see people clamoring for their processors and I think Intel's 12th-gen (and successive generation) chips will make it hard to beat them on performance.
Apple just controlled enough of the ecosystem to offer the amoun...
I'm guessing it probably means something like - why isn't there the equivalent of the x86 hardware world, but using ARM? Why can't I buy a CPU and motherboard with modern standard interconnects that I can use for a video card, SSD, etc; or maybe get a beige box computer, but it's ARM-based?
The argument that ARM based computers aren’t PC’s because they aren’t like x86 machines is self-fulfilling.
OTOH, it's the only RISC Unix workstation you can easily get.
- Microsoft for making an inefficient OS
- Intel because of politics
- Qualcomm because of mediocrity
Apple annihilates the competition, i went with a mac mini personally, that thing is a beast
https://secure.raptorcs.com/
Not as pretty as the Macs, but a 2-socket 44-core, 176-thread deskside that's 100% Windows-proof is still quite impressive.
So the holdup is that until Qualcomm bought Nuvia, their own designs were not good. The Surface Pro X sucked. The Nuvia-Snapdragons should be the turning point, but also ARM is suing Qualcomm over that so who knows.
More here on ARM v QCOM: https://seekingalpha.com/article/4541671-getting-chippy-arm-...