Qualcomm will try to have its Apple Silicon moment in PCs with "Snapdragon X"
arstechnica.com
arstechnica.com
A: Easy to port apps
B: Transparent to translate them.
Selling an M1 Mac was easy: It has some new tech, gets better battery life, all while running your old apps faster than your old Mac. Windows on the other hand fails aggressively here. To sell a Windows on ARM device implies the need to explain a whole lot of mumbo jumbo to the customer about 32Bit and 64Bit apps, about stuff that won't emulate at all, about the massive performance hit.
I honestly think that a Snapdragon X Chromebook has much better odds. Google is in a far better space to have a singular, cohesive, understandable product. Chrome OS doesn't care at all what CPU is in there. It all "just works" for the average user.
apt install binfmt-support qemu-user
update-binfmts --enable qemu-x86_64
With latter, like with any full-system emulation, for top graphics performance, you probably want a separate video card and [pci pass-through] it to the guest VM.
[qemu-user] https://www.qemu.org/docs/master/user/main.html
[qemu-system] https://www.qemu.org/docs/master/system/index.html
[pci pass-through] https://askubuntu.com/questions/1406888/
For simpler needs (ie: mostly 2D, very minimal 3D) it's great. I use this mostly for isolated browsing environments and am very pleased.
I believe this works with both Windows and Linux guests, though drivers are more annoying where you'd expect.
I'm a little surprised the maintainers haven't done so - but I haven't actually used the workflow. The implementation appears transparent; it registers a format with the kernel.
Maybe there are downsides. You hint at a very likely one - performance.
QEMU can do remarkably well with virtualization... but this, I'm not sure. I'm hopeful!
Anyway, with all that in mind... I think the point against requiring such extreme 'vertical integration' remains
If the argument against these two commands is it, then it's extremely trivial to resolve.
I think with more effort they will get there. I think the bigger problem is the one you outlined: M2 is so far ahead. Hard to see Qualcomm catching up and being price competitive
Like getting an M1 laptop and virtualizing Windows 11 has better performance than most actual Windows 11 ARM laptops
Qualcomm stuff won't sell well, not because it will be terrible but because unless it is priced extremely competitively nobody will care; because "battery life" is not a good enough argument that makes worth the hassle of dealing with a new architecture. There are so many cheap ways to extend the battery life of existing notebooks that it is a non-issue and complete Apple reality distortion field.
Also poor support on GitHub. So poor that I’m suspecting they do it on purpose.
They failed greatly at motivating the Windows developer community to migrate away from Win32 APIs, into WinRT/UAP/UWP, and now although WinUI 3.0/WinAPPSDK tries to bring that tooling on top of regular Win32, the developer experience is really bad versus what is still possible in UWP, in features and VS tooling.
Mostly due to how underresourced the team happens to be, and most app business units focusing on Web widgets, or SaaS products running on top of Azure.
Likewise there have been endless attempts of Windows on ARM, having all major compiler toolchains able to target it isn't enough, because without Apple's approach, most ISVs targeting Windows won't bother with yet another ISA on their development workflows if there isn't any money to be made up for it, only additional development costs.
https://www.theguardian.com/technology/2017/may/03/windows-1...
forcing people to "drink applications through a straw" from the slim pickings on the Windows store. Why? I think platforms always want to have what some other platform has. Microsoft is envious of Apple's App Store, YouTube wants to be TikTok, Instagram wants to be Twitter, pursuing this vainglorious dream at best dies with a whimper, at worst it is just plain death.
(Now it it a business school case study: Why does Microsoft fail when third parties succeed? How is it you can download games from Steam with the greatest of ease whereas the Windows store is a struggle? How is it Dropbox works and OneDrive doesn't?)
A distant second motive is better battery life. A third seems to be "acceleration for acceleration's sake". I remember reading some articles six months ago about an acceleration framework Microsoft was pushing that would give some effects for video conferencing... But I can't find it now because it vanished without a trace. Nobody cares.
I'm an Xbox Pass user on PC and I didn't notice any outstanding issues trying to install games from the store, or more often from the Xbox app. My bigger issue is that consistently games from the MS store feel subjectively slower than ones from GOG for example, like longer load times. But that might be related to DRM or maybe UWP. Whenever I have a choice I'll end up playing the GOG version over the MS or any other store.
The difference in experience could be related to my individual setup or personal preference.
One other big question for me is how many tech giants want to be the market leader is so many fields. Eventually they will tip the scale into regulation that mandates such companies to be broken up. For MS the store is probably an afterthought so leading in that segment is irrelevant.
At least for the time being, as the UWP sandbox model is in the process of being extended to Win32, in upcoming Windows 11 versions, and there is already some support on Windows 11 latest.
The other factor is basically money, Windows is no longer a profit center, Azure and XBox have replaced it alongside Office, thus the store is the only way to still make some money out of Windows in a world where everyone doesn't want to pay for Windows updates.
(Obviously you can add any repos you want, but the ones that come enabled by default on your distro are likely to be very heavily policed.)
The process by which Ubuntu or Red Hat decide what packages are in there is not one where anybody can walk up and ask to have their package included and to be treated "fairly", which gives them a big advantage.
From a technical standpoint, OneDrive and the Windows Store work fine for me. The worst problem for the Windows Store that there are few compelling reasons to use it, both from the perspective of the end user and the perspective of the software developer/publisher.
And of course regarding games the case study really has to be titled "Why does every single Steam competitor fail". Multiple big name publishers tried to compete with Steam, all of them failed miserably (well, Epic Store is still around, but I think it's fair to already call it a failure).
When a big feature update rolled out I'd find the configuration database would be rebuilt and it would work again. I had a struggle with Microsoft support over this, I wanted some procedure to rebuild that database and fix it, they wanted me to create a whole new account on my Windows machine and I'm like "no way I'm going to spend hours reconfiguring all the applications that I use every day just so I can download one crapplet"
Which is weird, because as I understand it the App Store isn't a _particularly_ big deal on MacOS; most software is either distributed outside it, or both in it and outside it. I've very little App Store-installed software on my Mac (though a lot of the software I have is also available on the App Store).
Microsoft and Intel seem so consumed for jealousy for phones that at times I’d worry that they will throw their existing platforms into the bonfire. It seems that The Innovator’s Dilemma has been thoroughly embraced by platform owners today.
My RSS reader YOShInOn is written in Python and hypothetically it could use multiprocessing, celery, and other stuff that would work better with WSL2, but I have it working just fine in ordinary windows and when I need to do something that doesn't fit will with the asyncio paradigm (long running CPU-bound tasks) I just open a pipe to a new process.
The development experience is a bit different, because you can run both arm64 and x64 toolchains, and of course everything has to match there. But that's not a big issue.
maybe most new apps are browser based, but in terms of number of users per app, definitely not.
If I look at the PC of any of my relatives or friends, the only stuff that isnt native is chat apps and the occasional weird poweruser tool. Most everyday work apps are native.
> development is done in WSL2 so not much of that matters and
thats also mostly a bubble phenomenon. a lot of development is done in visual studio, or some JetBrains product, vscode, and that's pretty much it. a lot of Linux devs run Linux, instead, too.
if MSFT admitted this they would be telling everyone that Chromebooks are a solution
https://www.tomshardware.com/news/microsoft-developing-chrom...
Don't answer the question, since Win 11 tries to show adds, it is pretty clear MS is going...
Performance was not good, and the CPU was a bit hot (but fanless). Linux support was imperfect, but for a first attempt it is not too bad. A M2-like machine running Linux, with good support, would be fantastic as an ultra-mobile laptop.
However, AMD is catching up in terms of performance per Watt, and has already caught up in terms of raw performance. Will a Snapdragon X have a sufficiently wide market niche?
Anecdotally, the operating system also determines, considerably, the system efficiency. Case in point - I installed chromeos (brunch) on my HP ultrabook (originally a windows 10 machine, 8th gen i7 "U" processor) and the battery lasts 12-18 hours vs 6-8 with windows doing essentially the same light workload.
CPU backend side x86 and ARM is almost identical. It's mainly just the front end where the things are different. Like instruction decoding.
This is often repeated, but very thinly substantiated.
ARM originally wasn't engineered for low power usage, just "normal power usage" in PC-like computers. It found its niche in low power devices mainly because the larger chips got dominated by other ISAs and weren't much interested in low margin embedded space.
It's very unclear if ARM as an ISA has any intrinsically meaningful advantage in power consumption. Modern chip design kind of converged and most of them work on similar principles like decoding the ISA instructions into microcode which is then executed. x86 decoder might be more complex than an ARM one, but the difference on a chip is small and likely meaningless in terms of power consumption.
A near-perfect example is Qualcomm's switch of its Snapdragon SOCs to 64-bit to catch up to Apple with its SD 808 and 810 lineup. But those were notorious for overheating and it wasn't until the later SD 820 and 835 - but by then Apple's A-series performance lead was significantly greater.
Also others noted, hardware performance is only 1/2 of the equation - the software OS, drivers, and apps all also need to be optimized. Microsoft seems to be perennially struggling in its Windows-to-ARM transition so certainly doesn't help matters.
The dark horse is the ongoing ARM-Qualcomm lawsuit to unwind its Nuvia acquisition and prevent the IP from being used in future products. If that comes to fruition, then guaranteed there'll be huge delays.
They do need to push devs a little harder to get ARM versions out there. It's kinda silly that there seems to be more ARM native apps available through WSL2 than Windows itself. They really need to have a team that at least goes around doing things like adding WoA support to GCC to make it easier to build open source apps for Windows on ARM.
https://blogs.windows.com/windowsdeveloper/2023/10/16/window...
Now lenovo, asus, Dell and HP and many of the meaningful windows/pc manufacturers can take advantage and offer compelling arm solutions that aren't watered down (looking at you, entire chromebook range).
Honestly I think this will be a much more significant push for ARM outside of mobile than everything we've seen up to now.
> All subsequent versions...have been tweaked versions of off-the-shelf Arm Cortex processors rather than fully custom designs.
I thought ARM was licensed? What does this mean, there are off-the-shelf processors and there are also fully custom designs?
Some OEMs take ARM's design as they are, others design them further, Samsung's Exynos is another example.
https://patents.justia.com/assignee/arm-limited
and of course trademarks, such as "ARM" itself.
So it is vaporware. It it sees the light of day, will be nice to see how it compares to Ryzen. (and no, laptops are not really PCs).
It's just not really worth risking that much money on a unknown, when a known good product is cheaper.
At least direct people to the previous discussion over here: https://news.ycombinator.com/item?id=37899676
They have an uphill battle without all the heavy lifting Apple has done to invest in the platform and force developers to make it work.