2024 could be the year the PC finally dumps x86 for Arm, all thanks to Windows 12 and Qualcomm's new chip

L4sBot@lemmy.worldmod to Technology@lemmy.world – 160 points – 2024 could be the year the PC finally dumps x86 for Arm, all thanks to Windows 12 and Qualcomm's new chip
pcgamer.com

2024 could be the year the PC finally dumps x86 for Arm, all thanks to Windows 12 and Qualcomm's new chip::We've already reported on Qualcomm's new 12-core Arm uberchip, the Snapdragon X Elite, and its claims of x86-beating performance and efficiency. But it takes two to tango when it comes a maj

83

Having used an ARM Mac, and the pains of countless utilities and apps that are x86/x64 only, as well as the pains of virtualising x86/x64 operating systems, I’m not a fan. I can virtualise ARM just fine on x64 but not the other way around.

(Edit: I’m not referring to OS utilities and apps - Apple have done a fine job with porting the OS to ARM, but the same can’t be said for the wider ecosystem - especially FOSS and niche developer toolchains).

People probably said the same thing when Apple dropped PowerPC for x86, there's going to be an awkward transition period but when it becomes a standard you'll feel differently.

yeah, but were not talkin some niche audience like apple powerpc products. lets not pretend apple had actual marketshare.

this is messing with legacy windows products that are deeply ingrained the world over. it will be far messier than that apple crap

I had a Mac G4 just before the transition from PPC and while that was painful (since x86 emulation sucked) this is a whole different kettle of fish.

These days I’m running all sorts of VMs for research and UTM or QEMU on macOS ARM just doesn’t cut the cheese. On a laptop, sure, ARM is fine. Heck, even in a data centre it’s fine, but on workstations, ARM is too sluggish for virtualisation or anything except ARM. Not to mention the shocking state of Windows 11 on ARM and how loads of Windows components don’t actually function properly or even run. Defenders GUI doesn’t even open!

cut the cheese

This doesn’t mean what you think it means, haha

Haha thank you for pointing that out! I’m European and clearly awful at American slang. I won’t edit my post as that’s too funny a mistake to remove.

Apple yes, windows? Not so sure, in windows there's alot of x86 games and everything, people just won't drop that you know? And with Linux gaining traction in gaming community x86 going to live at least another decade, ONLY way people going to drop x86 if you can launch x86 apps on arm without terrible drop off performance, while apple have that, others don't, so until then except mobile devices only apple and niche laptops gonna be on arm, because gaming and other legacy software people not gonna drop until you can launch it on arm without terrible drops of performance

Apple is a pretty closed system made for graphics designers and people who don't like choices. It isn't as simple to make it happen for the pc market.

I also doubt Microsoft will go arm on the next console. They've been enjoying all the easy ports to PC for their games and having to port over games based around arm to run on x86 would probably cost a lot more money and time. Game makers wouldn't like it.

I’m confused, my M1 MBP had like 1-2 things max that were x86 still that I needed and those ran fine on Rosetta.

I know docker is a bit more annoying but it’s not that bad IMHO.

Came in to say the same, and I run all sorts of weird shit. Rosetta is so seamless the only way I know it’s an x86 thing is that it takes a while to launch the first time.

That's because macs don't have games. They've had 3 iterations of ARM processors and I still can't download steam natively. If I could, most of my steam library wouldn't run natively.

Steam runs absolutely fine on my m1. I haven’t checked if it’s running Rosetta or native arm code, but I can’t tell at all so it doesn’t matter. All my Mac games run fine on steam, unless they are old and 32 bit. But macs dropped 32 bit support a while ago even on intel chips. The games run great too.

I’m guessing the ARM version will be for regular people that just go to the main websites. Maybe the x64 version will become enterprise only.

They're going to have to get the emulation working better for x86/x64 software. And they're going to have to get the driver situation sorted -- which likely means requiring ARM drivers alongside x86/x64 drivers in order to meet certification for having a Windows sticker or WHQL certification to gradually build up the list of supported hardware.

The CPU and processing power benefits would be great, but if I'm going to lose software support then I'm only going to do it for RISC V.

Yaaaah, came here to something something RISC-V ^.^ One of these days I'll have a RISC-V system. I'll have no actual use for it but I'll love it stubbornly just because :D

Anyway I'm gonna be over here daydreaming about RISC-V taking over the world instead of ARM. Bwehehehehe.

(Edited to fix my ^.^-face)

Fundamentally, I'm not sure Qualcomm is the brand I'd trust to lead the world off of x86.

I understand nobody actually likes Qualcomm products in the cellular space, but they're stuck with them due to patent minefields. That's not really a great vibe to bring in when trying to compete against known-quantity x86 vendors.

I figured we'd see homogenous CPUs-- either in the same socket or as an addon module, so you can cast off some stuff to ARM or RISC-V but keep big x86 for games and heavy closed-source software, then flip to RISC-V main with x86 addon cards, and finally emulation.

Sort of thinking about a Pinetab-V, but even the flaky, doesn't suspend right 20% of the time, wigi was weird on every OS except OpenBSD, Ryzen 2700U it would replace demolishes it. The Lichee Console looked neat with the EEE PC sizing and Trackpoint, but it's way pricier.

Where would those benefits be? Let's start with gaming on the M3 Mac - it's CPU bound in many games even though apple's compatibility later is actually good. And the GPU is a joke, even compared to the Intel dGPU offerings. Let's not start on encoding (besides iMovie), packing or compiling things. Or even actually rendering stuff...

Compatibility layers are comprehensive, but they're generally not performant. For me personally, I use a real computer that runs my daily workload, servers and games all at once on different virtual desktops, so a faster CPU will definitely be impactful.

It's not just about avoiding 100% CPU either. CPUs not being the bottleneck for performance sounds like a great problem to have

Not a chance. We have several more years of x86 dominance.

I don’t know what the author was smoking, but nobody that knows what x86 and ARM are would reasonably say x86 is anywhere near its end. I want it to be, fuck I want it to be, but I’m also not stupid enough to think it’s happening even remotely soon.

I'm rather hoping RISC-V comes up and eats their lunch before it happens.

My reasoning for this is that I've lost too many hours trying to kludge finnicky ARM boards into supporting proper mainline video acceleration. It's awful. It's horrible. It's a waste of time.

The silly x86 SBC I got worked out of the box with OneAPI with no complaints at all.

The ARM boards ran the gamut from gibberish/garbage rendering, dropped frames, washed out images because of cheap tricks to up performance.

I know this is more down to the weak (and proprietary) video cores included on these boards.. but after spending a significant amount of time playing with them, I'm going to say "No, thank you."

I don’t care as long as desktops remain modular.

Ampère makes some ARM CPUs that go with modular boards, where you can socket your own CPU, DDR5 RAM, NVMe drives and PCIe accessories.

Nope, I'll never run windows on anything other than x86 (for my desktop).

I'm very happy with my ARM MBP for work, but I occasionally pull up software written decades ago (either music production plugins or games typically) on windows and it still runs, some of the companies that wrote that software no longer exist, so no first party patches will be coming.

I'll never run Windows.

I'll never run

I'll never

I'll

ll'I

reven ll'I

nur reven ll'I

.swodniW nur reven ll'I

.(potksed ym rof) 68x naht rehto gnihtyna no swodniw nur reven ll'I ,epoN

.gnimoc eb lliw sehctap ytrap tsrif on os ,tsixe regnol on erawtfos taht etorw taht seinapmoc eht fo emos ,snur llits ti dna swodniw no (yllacipyt semag ro snigulp noitcudorp cisum rehtie) oga sedaced nettirw erawtfos pu llup yllanoisacco I tub ,krow rof PBM MRA ym htiw yppah yrev m'I

1 more...
1 more...
1 more...

Fair play, if you don't need the compatibility with Windows drivers and software, there's no real reason to choose it in 2023.

1 more...

Windows 11 on ARM will probably that software without modification.

Assuming it runs on Windows 11 at all...

1 more...

For all the informed technical analysis and debates about this, the vast majority of consumers don't care about any of this stuff, and they're the ones who will decide this "year of the whatever." The worse option technically speaking has won out many times in the past.

Consumers will choose more battery life, all else being equal.

Tell me you never used Arm based system for daily drive without telling me you never used Arm based system. General software compatibility is not there and PC is not only on Windows or Mac. Sure on Mac they have enterprise support for their user. By having more power (bruteforcing) to run the emulation simply does not mean the software run flawless.

Maybe I'm a bit bias since I'm comparing it with SBCs (but thats what is affordable). As someone who have Raspberry Pi 4 and Orange Pi 5, the situation is a bit different. Raspberry Pi have a well supported system by communities and the devs, meanwhile on Orange Pi 5 some drivers are not released by the Orange Pi/left to the dust if there are no maintainer, not to mention if you want specific build of binary which not covered by repo/ppa, you have to build yourself from source, and the GPU driver situation for OPi5 which not yet have Vulkan support and sub par performance on Linux meanwhile on RPi5 they have Vulkan support 2 weeks after release.

All the top posts here are people saying it won’t happen.

I was at the store over the weekend and saw a full display of chromebooks. Someone purchased one right in front of me.

I’m sure there’s a market for both technologies to exist at the same time.

They're purchasing the OS, not the CPU/SOC architecture.

Microsoft doesn't have the same loyalty that Apple does. They can't afford to release an ARM OS that isn't already supported by all major software applications, and majority support for normal x86 apps, with assistance and roadmap to completely bridge the gap.

When the transition is seamless, or 90% seamless, the architecture won't matter, and customers won't even realize they've switched.

If they release ARM hardware that doubles battery life and performance, but doesn't offer a seamless transition, it'll flop. Just like their last attempt did.

Microsoft doesn't have loyalty? They have practical market dominance. I say this as a Linux user but ain't no way Microsoft can do anything to drive away their user base. If their users buy a laptop and find half their software doesn't run on it or runs sh*ttily due to emulation, I'm pretty sure they'll blame the laptop manufacturers before Microsoft or demand the laptops have a x86 variant and even that's a long way before moving to another os.

Exactly.

This quote reveals the lack of understanding about how MS works:

Thus far, there's little doubt that Microsoft's efforts with Windows on Arm have been half hearted.

Half-hearted? I bet MS research developed a plan for ARM before the public knew ARM existed. They have a massive research department.

They don't need to support ARM until it's well-established at a performance point that can supplant x86 even under emulation. Their major clients are business, and frankly laptop battery life is more than good enough for business users today (I can run almost all day on battery, and I do a lot of file management with a 3 year old, midrange business laptop).

Now what's compelling for big business is power consumption in data centers or even office buildings. But those systems aren't running Windows directly on iron - it's all virtualized. So even there I'm not sure ARM competes yet. Maybe for desktops in the office, call centers, etc. But those already use Mini PCs.

Gonna be interesting to see how it all works out. Will we really see dramatically better battery life on a Windows ARM laptop? Will this also be the return of Windows Phone/Tablet (does this change the tablet definition if it can run Windows/Linux?)

Edit: Forgot that MS already has Win10 IoT Enterprise, and Win10 IoT Core - so Enterprise for managing IoT devices, and Core to run on IoT devices. Core is much like Embedded Windows in that it's stripped down and only runs one app. I'm sure MS considered ARM devices when building it, and wouldn't be surprised they've already done extensive testing with an ARM version already.

There are tons of x86 Chromebooks still tho.

Yeah I bought a $300 amd Chromebook and run Linux on it. Had to flash a new efi firmware to make it fully usable though. Worst part of it is the soldered 8gb of RAM but it works for my usage.

There's a lot of focus on Windows for these types of chips, but Chromebooks are probably the best use case for them right now. ChromeOS runs great on ARM and there's no legacy software to worry about, but they feel kind of slow because the ARM chips they've used have been slow. I'd love an ARM Chromebook that actually rips.

Could we have a future where we have an arm main CPU, gaming GPU, and also an x86 card?

I've used an ARM Mac for the past 3 years on both macOS and Linux. My trusty M1 Air has been an absolute joy to use. I would like more options for a fast, battery efficient and most importantly fanless laptop, so I'm looking forward to this.

Limping along with a wonky hinge on my 5 year old laptop waiting for these to come out. Haven't run windows for years now so I don't think I'll be missing intel much at all. Might have to do some cross compiling for deploying software to intel cloud nodes, but arm VMs for android development will speedy.

Depending on which problem you're experiencing, I've used jb weld to fix it.

JB Weld is magic (as is it's competitor PC7, seems to be the same stuff). I just used it to fix an iPad keyboard - the part holding the iPad cracked near the hinge. Just filled the interior with JB and reassembled (it squeezed some JB out). Crack is still there, but the hinge area is even more rigid now. Of course it'll never come apart again.

I'd like to promote Goop. It's like a super thick rubber cement. There's very little it won't stick to. Very strong. It's pretty much all I use anymore. Oh, and it peels right off when you need to remove it (though it takes some effort).

I've glued phone mounts to the dash with it. Power strips to the underside of the desk. Glued laptops back together when the screw mounts break. Use it for the metal disks on back of phones for magnetic mounts. It's holding L-brackets to the back of the TV so I could put the soundbar on top. Also a bunch of industrial Velcro (the thick plastic stuff) on the back of the TV to hold a wifi router, Bluetooth adapter, USB hub, power supplies, and cable management.

Oh I'll have to look for that. It took a week for your message to appear for me. Not sure why

The main advantage of ARM right now is that there are low power cores available. The actual instruction set is unrelated to this advantage. If Intel or AMD put more serious effort into power efficiency most of the advantages go out the window.

As for instruction set changes impacting what software you can run I think that is still a big issue. Yes porting to ARM is straitforward in more modern programming environments but most software actively developed at the moment has a lot of old cruft that won't easily port if the engineers can even be convinced to touch it. Most businesses are dependent on old software not all of which is still maintained. Most gamers are even more tied to old software that is not going to get ported and often has annoying anti-virtualization checks (see games breaking on systems with enabled intel e-cores).

I am not sure how large the modern non gaming personal pc market is (tablets, phones, works computers, and chromebooks probably took a chunk out of it) but that could be in play.

Steam. Almost all games would be impacted. On Linux we already use translation layer (Windows -> Linux), but I am not sure if it's a good idea to emulate X86_64 on top of translation layer.

Getting anti-cheat that technically already works enabled on Linux has been a lot of work and Epic still won't enable it. Piracy protection systems will also be an issue. Most EA games inspect your CPU to see if they like it on startup (I think this is using vmprotect and some non-OS x86 calls but don't quote me on that). These kinds on anti virtualization checks are really common (not just in games ProctorU and lock down browser do them too). I don't think valve running an open virtualization layer will be well received by companies and they will probably ban it from running games. MMOs (due to botting) and anything with anticheat will look particularly askance at this. I also suspect Valve won't want to try hiding the VM signatures as it borders on violating DMCA.

Newer games will probably get ported if a large part of the market buys into ARM. Unity stuff might get re-released as it is .net if the publishers can be bothered. Minecraft java edition will also always love you (the launcher might not though).

I think this is likely. After dealing with how bad W11 is, MS realized they don't need working software or backwards compatibility to sell units.

Don't care what Windows does. What's going on on Linux here.

Linux has been kinda great with ARM due to devices like the Rasberry Pi. Ubuntu, Fedora, and Manjaro all offer ARM variants

Great to know, I have only run Raspberry Pi OS so far. But I think we also need parity in applications since a lot of packages are architecture specific. I'm sure most core packages and application packages in Linux must be supporting ARM.

I'm all for it, just don't leave out the DIY market. I would love a socketable ARM platform. Risc-V would be even better.

Wake me up when they make faster ARM than my 5900x

These new are highend. So expensive, powerhungry and not hugly better than x86 CPUs on their tasks. Must have something that makes it a nobrainer so that it is much better on every aspect.

Apple just had 100% control of all Apples products so they basically forced them over when their new product was just better in most cases.

A Galaxy Book 360 that runs cold and has great battery life would be perfect. I need photoshop and illustrator on a Windows machine that runs as well as a Macbook, I don't care about gaming with an ultrabook.