Arm Targets 50% of Windows PC Market Share in Five Years, CEO Says (reuters.com) 106
British chip designer Arm expects to capture more than half of the Windows PC market within the next five years, CEO Rene Haas said in an interview. The company's optimism comes as Microsoft and its hardware partners gear up to introduce a new generation of AI-powered PCs running on Arm-designed chips, potentially reshaping the Intel-dominated industry. Haas attributed Microsoft's commitment to supporting Arm's technology through enhanced developer tools as a key factor in the anticipated market shift.
Heard this before.. (Score:3, Informative)
When Windows 8 released back in 2012, they made a lot of noise about the ARM build, at the time to fight back against the perceived existential threat of Android tablets (around the same time you also had Intel emphasizing their x86 Android tablet partners).
Of course it failed, because it couldn't run x86 applications, nor would they let it run third party 'traditional' apps even if compiled for arm.
So this time, *maybe*, but unlikely. They've learned their lesson about letting traditional apps exist, and they may include some sort of x86 compatibility, but it still won't be as good as native for those applications.
Apple managed to migrate their userbase multiple times based on their ability to unilaterally say " is dead, get used to it, you won't have options so you have to live with porting your applications and rosetta performance impact in the mean time". For an ecosystem where every customer and every partner is able to hedge their bets, it's going to be damn near impossible to manifest such a dramatic shift when the outcome is unlikely to be that much better even if embraced.
Re: (Score:1)
" is dead, get used to it, you won't have options so you have to live with porting your applications and rosetta performance impact in the mean time".
Spoken like someone who has absolutely no experience living with "rosetta performance impact". If you want to pretend to be an expert, don't tell everyone how ignorant you are.
Re:Heard this before.. (Score:4, Insightful)
Spoken like someone who has absolutely no experience living with "rosetta performance impact". If you want to pretend to be an expert, don't tell everyone how ignorant you are.
He does not give an opinion as to where the performance is good or not. He is merely stating that is the strategy Apple has decided multiple times. Rosetta was used to translate PowerPC to Intel back in 2006. Rosetta 2 is the new translation for Intel to ARM.
Re: Heard this before.. (Score:1)
Re: Heard this before.. (Score:1)
Shit man that's quite the grudge.
Re: Heard this before.. (Score:2)
Damn grandpa
Re: (Score:2)
I've used 6800, 68020, and 68030-based Macs, and owned PowerPC (601, G3, and G4), Intel (Core 2 Duo, and circa 2013), and now M-series (M1 Max and M3). The older generation software ran on the newer machines, but slower than it could have run after being recompiled. Sometimes even large parts of the OS was run through the emulator; newer point-releases of the software would recompile more of the OS, so updates would usually make your machine run faster, and of course upgrading the apps to native was always
Re:Heard this before.. (Score:4, Informative)
Apple also does a pretty amazing job of providing backwards compatibility layers for old applications during the transition. Ive been through 3 huge transitions in apple hardware and every time apps pretty much just work and it doesn't take long for developers to support the new architecture. If there is an issue it's that the last generation of the old machines are still around when new versions of software are no longer being developed.
They provide excellent tools to manage the transition. Sure they say "it's dead, get used to it" but they also don't leave their users or the devs dead in the water during the transition. Something I have never seen from Microsoft. Of course Microsoft doesn't really care what hardware you are buying so that has something to do with it.
Re: (Score:2)
>> Ive been through 3 huge transitions in apple hardware and every time apps pretty much just work and it doesn't take long for developers to support the new architecture.
That's a lot easier to do when the number of third party developers that support your OS can be counted on one hand.
It's a lot easier to do when you provide development tools that make the transition easy.
Re: (Score:2)
You must have some pretty funny looking hands. AI generated, perhaps?
Re: Heard this before.. (Score:1)
Here I'll get started:
Adobe.
Umm.... Yea who else?
Re: (Score:3)
This is good and bad. Sometimes an app vendor just decided to not bother with Apple after a change, even though the "change" might just be a simple recompile to support ARM/x86 binaries.
Apple and the PC world have different meta-philosophies. One can install Windows 1.0 on a PC and upgrade it to Windows 11. On the other hand, macOS has been through so many twists and turns that there is absolutely no way System 1 or System 2 can run on a M3 iMac, barring emulation.
Is the Mac way good? Sometimes it throw
Re: Heard this before.. (Score:2)
"One can install Windows 1.0 on a PC and upgrade it to Windows 11."
You're going to need multiple PCs for that. There's not one generation of PC that will successfully run all of those operating systems. I'm willing to believe you could upgrade a Windows install that way, but not on one machine. I might be inclined to try it in a VM though, which could be reconfigured as you went through the process...
Re: (Score:2)
" One can install Windows 1.0 on a PC and upgrade it to Windows 11"
Yes, there were some Windows 10 PCs (sold new with Windows 10) that were unable to run Windows 11 due to the TPM being version 1.2 instead of 2.0).
So yeah, no.
The Windows 1.0 memory model supports at most 16 MB (or maybe less), while (for example) Window 7 needs at least 128 MB to start but 256 MB to install (my memory might be unclear on exact numbers and versions though).
Re: (Score:2)
One can install Windows 1.0 on a PC and upgrade it to Windows 11.
You physically cannot.
With some hackery, you can find the right combination of machine that supports PS2/AT/Serial ports, BIOS, and TPM- but it'll have been a while ago.
Re:Heard this before.. (Score:5, Insightful)
Yea, except Microsoft has made massive strides towards the phone-ification (enshitification) of "your" computer. Own nothing, tricks to try to force a cloud logins, the soon to be Recall malware, browser hijacking, forced updates you have 0 control over.
They want ALL the rent money. Now and forever. Pay for the OS you can't do shit with, while having forced upgrades or else, while paying a premium for them taking a shit in your mouth over and over again.
Even if you don't use their shit you're still paying the tax, but they will have even more control over an ARM based platform. They've been selling Windows Store App Locked Windows S trash already, this is another step in that direction - they're absolutely rock hard at the though of the kind of lock-in Apple has.
Most people are totally unaware and don't care at all. They don't know they're about to lose Wordpad either and won't be able to open docx without an ad popping up. Enjoy Clippy 2.0 button on your keyboard.
Re: (Score:2)
I don't see signs that their new ARM initiative is any more locked down than an x86 variant.
I grant they have pushed for lock-in and recurring revenue, but I don't think that correlates at all between x86 and ARM. The 'S' attempt and subsequent 'S mode' was pretty much all an x86 move. Windows RT was a more locked down vision that happened to be ARM based, but that was more about 'no legacy to worry about anyway' rather than 'ARM is better for locking down'.
So whatever MS ambitions may be about screwing ov
Re: Heard this before.. (Score:1)
Nope
Re: (Score:2)
I don't see signs that their new ARM initiative is any more locked down than an x86 variant.
Other than MS controls ARM software through their store? With x86 a consumer can buy and install from a number of sources including optical discs if the PC still has a disc drive. With his new initiative is every software only available from the MS Store?
Re: Heard this before.. (Score:2)
From what i understand, what you describe was their first attempt with Windows RT. That didn't work, so they tried to do the same thing, but with x86, and call it S. That didn't work and now it looks like this new run at ARM is not requiring app store as the only way in. In fact they brag on how you can run unmodified installers and it'll run (with emulation), and suggests that maybe applications will be treated similarly.
Re: (Score:2)
In fact they brag on how you can run unmodified installers and it'll run (with emulation), and suggests that maybe applications will be treated similarly.
But that seems to me that only applies to x86 software ie a consumer can run existing x86 software on their new ARM PCs. ARM software will be controlled by MS.
Re: (Score:2)
According to MS:
I want to use Windows programs that aren’t in the Microsoft Store. Can I run them on my Windows 11 Arm-based PC?
Yes, you can install Windows apps that aren’t available in the Microsoft Store in Windows. Peripherals and devices only work if the drivers they depend on are built into Windows 11, or if the hardware developer has released Arm64 drivers for the device. It's a good idea to check whether the hardware developer has published a version of the driver that runs on a Windows 11 Arm-based PC.
I suppose that could be theoretically limited to "x86" but I feel like I've read that isn't a restriction they planned for this run at it. I can't really find resources definitively speaking one way or the other on a cursory search.
Again, I know they would *love* a store locked down story, but they would also be ecstatic to do it to x86, and they have tried. There's nothing about ARM v. x86 that makes one more amenable to locking into a store than the other, it's just what Microsoft thinks
Re: (Score:2)
I don't see signs that their new ARM initiative is any more locked down than an x86 variant.
Windows RT was a more locked down vision that happened to be ARM based, but that was more about 'no legacy to worry about anyway' rather than 'ARM is better for locking down'.
Arm was better for locking down in the instance of Windows RT.
It supported CPU-validated signed bootloaders.
It's significantly more difficult (see: impossible) to lock someone out of their x86 without going through quite a lot of work, and being responsible for the UEFI for every device shipped (which then won't be UEFI compliant, since UEFI compliance requires user key management)
I very much hope the new devices don't utilize QC's ability to lock its bootloaders. If the machine is UEFI, I'll buy one ju
Re: (Score:2)
Arm was better for locking down in the instance of Windows RT.
It supported CPU-validated signed bootloaders.
So does x86.
It's significantly more difficult (see: impossible) to lock someone out of their x86 without going through quite a lot of work, and being responsible for the UEFI for every device shipped (which then won't be UEFI compliant, since UEFI compliance requires user key management)
That's by agreement rather than inherent to architecture. Also, that implies that ARM is locked down, but there are plenty of SBC ARM designs without SecureBoot. ARM also is largely moving to UEFI, with similar capacity for what keys are/are not included in a platform and whether the user is allowed to disable secureboot or not.
Besides, the SecureBoot locking out third party OS is generally the least of the concerns with Windows, the real concern is once you have Windows booted, will it let you
Re: (Score:1)
So does x86.
Nope.
x86 CPUs have a standardized bootstrap process. They jump blindly to motherboard manufacturer mask ROM.
Signature protection after that is part of UEFI, not the CPU itself.
That's by agreement rather than inherent to architecture.
Huh?
If your UEFI does not support user key management, then you are flatly not UEFI compliant.
Also, that implies that ARM is locked down, but there are plenty of SBC ARM designs without SecureBoot.
All QC parts require a QC-signed first-stage bootloader. It's non-negotiable.
SecureBoot is architecture-independent.
You're mixing up different layers.
ARM also is largely moving to UEFI, with similar capacity for what keys are/are not included in a platform and whether the user is allowed to disable secureboot or not.
Arm isn't moving anywhere.
There is no standardized boot process for Arm.
You are free
Re: (Score:2)
Microsoft's own Windows ARM devices, like the Surface Pro 9 with 5G, has full UEFI implementation including the ability to disable TPM and Secure Boot.
Obviously it remains to be seen whether QC do the same with their Copilot+ devices, but given that the Surface is a very Microsoft-centric device and they allowed it, I don't see the downside for QC to do the same.
Like Stephanie Jim Sterling Says (Score:2)
Re: (Score:2)
Pay for the OS you can't do shit with, while having forced upgrades or else, while paying a premium for them taking a shit in your mouth over and over again.
Wow. Please, show us on the doll exactly where Bill Gates touched you.
Re:Heard this before.. (Score:5, Interesting)
I have an M1 mac, and for the most part it's a lovely machine, and runs all the things I'd like it to run. EXCEPT that every once in a while you have to use x86. Back in the day when Macs were PowerPC you'd have been out in the cold - which is in part why Macs never gained the market share they probably should have done.
For me (as a techie) to be truly productive, I have to keep an x86 machine around somewhere for the occasional compile or run or whatever that's needed. Otherwise, yes, 90% of my work is fine on my mac. Shame they can't/won't give me an x86 co-processor - then I could use Arm 90% of the time and the coprocessor for the odd job here and there, saving me the hassle of the other machine "just in case".
However, back to MS... Windows eats battery and uses the fan on x86 not just because x86 is inherently so bad, but because Windows is so bad. I'll bet that out of the box things will be pretty good, but a service pack or two later, plus any sort of corporate management, anti-malware and other protections and that Arm chip will be sweating away as badly as it did on x86 - but this time, MS won't have put in the hardware to do it properly, so it'll be utterly hopeless.
My personal barometer here... if your x86 laptop needs to use the fan for you to type an email or browse the Internet, then your system is terrible. I couldn't tell you if it's the OS per-se, or the stuff you've layered on top of it, but it's working out to be terrible. Arm isn't going to fix it, and yet, this is the state that too many Windows laptops seem to be in (or get in).
Re: (Score:2)
I'll confess that I don't do MacOS, but I would have assumed Rosetta gets you there for most applications?
You mentioned an occasional compile run, I would have assumed cross-compile would be an option? Some languages/implementations are easier than others, but I do a fair amount of ARM builds from x86, and in some of those cases even set up qemu invocation of binaries to let it run arm binaries on x86. This is more convoluted than native compilation, but I wager that a coprocessor would still be pretty con
Apple has no real responsibilities (Score:2)
Apple managed to migrate their userbase multiple times based on their ability to unilaterally say " is dead, get used to it, you won't have options so you have to live with porting your applications and rosetta performance impact in the mean time". For an ecosystem where every customer and every partner is able to hedge their bets, it's going to be damn near impossible to manifest such a dramatic shift when the outcome is unlikely to be that much better even if embraced.
Apple got by with this by being a luxury brand for playtime. There are too many mission-critical applications written for Windows. Apple is not a serious "work" platform for most people and the few, like myself, who can use Apple for work can only do so because they work on cross-platform languages and do all of their work in the cloud. LOTS of industrial devices rely on Windows x86, so MS will have a LOT of angry users...not to mention gamers or server applications.
You could argue "fuck them"...they
Re: (Score:2)
Every G4 Mac could run OSX, and anyone running OSX could pop open a terminal window and have the full power of BSD Unix at their fingertips. So saying you couldn't do real work on G4 but you could on Linux is just BSD bashing.
Re: (Score:2)
A first question would be if full MS Office will be available on those desktops. If they have Office and printer drivers, almost all business desktops could run Arm. On gaming desktops, usually found at home, the situation is the opposite, a switch is unlikely. Other than those, there are niche cases, where it might be possible or it might be not: home users who don't play games, business desktop with special needs and such.
Re: (Score:2)
If they have Office and printer drivers, almost all business desktops could run Arm.
I think the printer drivers don't even matter for most office tasks. However, the challenge is 'almost', where the alternative option is 'all'. For a business, why risk things on 'almost' unless you are under vendor pressure to migrate away from an 'officially' legacy platform (hell, even then, I've seen companies banking on software that's been defunct over a decade). The eternal problem with Microsoft on ARM is that you have the x86 platform, which can run *everything* and is a strategic viable option
Re: (Score:2)
When Windows 8 released back in 2012, they made a lot of noise about the ARM build, at the time to fight back against the perceived existential threat of Android tablets (around the same time you also had Intel emphasizing their x86 Android tablet partners).
Of course it failed, because it couldn't run x86 applications, nor would they let it run third party 'traditional' apps even if compiled for arm.
So this time, *maybe*, but unlikely. They've learned their lesson about letting traditional apps exist, and they may include some sort of x86 compatibility, but it still won't be as good as native for those applications.
Apple managed to migrate their userbase multiple times based on their ability to unilaterally say " is dead, get used to it, you won't have options so you have to live with porting your applications and rosetta performance impact in the mean time". For an ecosystem where every customer and every partner is able to hedge their bets, it's going to be damn near impossible to manifest such a dramatic shift when the outcome is unlikely to be that much better even if embraced.
This... As I've had to explain to Mac fanboys before... No-one buys a computer for the operating system, we buy it to run programs on. When it comes to Windows, I'm certain I'm not the only one with a 30+ year back catalogue of software that I still use on occasion (OK, mostly games) and that already has enough issues with changes to operating systems on x86. I reinstalled and modded Fallout 3 the other day, works great except if you alt-tab out you cant alt-tab back in... and that was a game from 2007, me
Good luck! (Score:4, Insightful)
Unless ARM somehow convinces every Windows game and application developer on the planet to port their code to ARM native binaries in the next three years, they're going to continue to have performance issues due to their code running in an emulator.
Games are going to be the likely pain point here, because the integrated graphics built into ARM SoC's that the laptop manufacturers are using is well behind what a dedicated GPU can do. Adding x86-64 emulation on top of that is just going to make things worse.
Yeah... I know that people aren't buying $500 laptops to play games on them, but at least having the option do to so is nice to have. When the reviewers try doing this (and trust me, they will), these products are going to get roasted.
Re: (Score:3)
Microsoft isn't trying to fully replace ARM for mainstream use...yet. But there's a market segment that loves locked down PCs - businesses. Incompatibility with games is a bonus. Long battery life so you can keep workers on the job during public transit commutes.
Re: (Score:3)
As an IT guy, though, I'm always going to have to worry about that handful of "legacy" business applications that hasn't been properly updated in 10 years. Odds are, they probably already have some of them running in some awful Windows XP compatibility mode already to get it working right on Windows 10. If it doesn't work well on an ARM processor, that's going to be deal-breaker for them.
Re: (Score:2)
It's too bad Microsoft overcharges for and overcomplicates application hosting with something like Terminal Services. Would be great to just leave that other software installed on another personal desktop and be able to stream it over.
Really, that would be a great thing to have for any heavy lifting apps being run while on the go, even if they run on ARM.
Or maybe it's already possible [github.com]. Can't look into it now.
Re: (Score:2)
Re: (Score:2)
The WinXP compatibility layer is a full VM and given the advances in hardware of the past ten years, the emulator won't affect performance.
Most likely will- a *lot*.
Binary translation can offer pretty good performance for userspace processes, particularly ones that spend a lot of time executing native kernel code- but with a full VM, that's not a luxury you get. You're emulating it all. And the performance is generally pretty fucking terrible.
I've had to do some x86-64 linux guests on my Arm MBP, and it is truly fucking terrible.
Re: (Score:2)
Re: (Score:2)
Windows XP is 23 years old.
Full-virtualization of an x86 machine on an aarch64 host is a fraction of the speed a native machine would have been back then.
As explained elsewhere, emulating a userspace binary is easy. Syscalls are still entirely native.
When you have to emulate the kernel, you have to fully emulate the actual CPU, not just a subset of its ISA.
That gets very expensive.
Re: (Score:2)
As an IT guy, though, I'm always going to have to worry about that handful of "legacy" business applications that hasn't been properly updated in 10 years. Odds are, they probably already have some of them running in some awful Windows XP compatibility mode already to get it working right on Windows 10. If it doesn't work well on an ARM processor, that's going to be deal-breaker for them.
Currently if you are running applications on Windows 11 that require the WindowsXP VM layer, you're already in a full VM. There's no architecture emulation and you get some hardware support for virtualization. The ARM version of the WindowsXP VM might use more CPU cycles, but the apps will still run faster than they did natively under Windows/XP.
Yes, emulating the kernel is expensive but we already to this with things like DosBox and current hardware i
Re: (Score:2)
The comment to which I was originally responding was
I know. I was referring to the idea that hardware has become an order of magnitude faster in that last 10 years. It has not ;)
Currently if you are running applications on Windows 11 that require the WindowsXP VM layer, you're already in a full VM. There's no architecture emulation and you get some hardware support for virtualization.
Yes, you are running in a full VM, and incorrect- there absolutely is architecture emulation.
The Windows XP utility virt is paravirtualized.
The ARM version of the WindowsXP VM might use more CPU cycles, but the apps will still run faster than they did natively under Windows/XP.
No, they will not.
Not even close.
Because the Windows XP virt will require full hardware emulation of the CPU to work.
A userspace program that is binary-translated is currently "fast-ish" because the syscalls are all run natively, so that not
Re: (Score:2)
The minimum system requirements for Windows 11 are pretty much exactly 10x the minimum system requirements for Windows/XP.
Since you want to be very exact here, Windows XP required a 233Mhz processor and 64MB of RAM. Windows/11 requires a 1 GhZ processer and 4GB of RAM.
Re: (Score:2)
I struggle with what exactly you are trying to prove here. The original assertion that I made was that Windows64/ARM devices should run Windows/XP in an emulation mode just fine or more specifically it should run the Windows/XP 32-bit WoW subsystem just fine.
I'm proving that is simply not the case, and cannot be the case.
I don't think any of us know how Microsoft will deliver the Windows/XP compatibility mode for ARM or even if they will include it at all.
I think they won't, just due to the cost of full virtualizing a non-native architecture, which you are *vastly* underestimating.
However, the number of people running Intel Docker images on ARM Macs using QEMU is significant and those images run tolerably. Those are 64-bit images running in an emulator. More specifically the emulator runs an entire 64-bit Ubuntu image in an emulator and then starts Docker containers within it.
I have an M1 Max MacBook Pro.
Docker on Mac can run in 2 modes: Either with an x86-64 VM (which runs at about 20% that of the native CPU speed), or with an aarch64 VM using Rosetta for Linux userspace binary translation.
The latter works well, for reasons I've explain to you already.
Nobody does the former anymore if t
Re: (Score:2)
Most kernels fit into L3. If you're trying to imply that the kernel, your running application, and all of its shared libraries fit into L3- in 99% of cases- no. That's laughably off the mark.
The applications in question here are those that were build for Windows/XP-32-bit and won't run unmodified on Windows 7 and are still in use. Microsoft used to keep a list of these but I can't find it. The 99% number came from the OP who asserted that 1% of people in the world are still using such an application. That seems high to me. But he may have been using the number more for dramatic effect than a decimal precision statement. Given that there are plenty of processors with 64MB L3 cache and that
Re: (Score:1)
The applications in question here are those that were build for Windows/XP-32-bit and won't run unmodified on Windows 7 and are still in use. Microsoft used to keep a list of these but I can't find it. The 99% number came from the OP who asserted that 1% of people in the world are still using such an application. That seems high to me. But he may have been using the number more for dramatic effect than a decimal precision statement. Given that there are plenty of processors with 64MB L3 cache and that the minimum RAM requirements for windows XP was 64MB, 100% of programs that would run on Windows-XP minimum system requirements will fit in a 64MB L3 cache by definition.
I've used several applications that required Windows XP compatibility mode. Games, mostly, I'll admit.
If you're now asserting that you were just trying to say that some weird defined subset of applications will fit entirely in L3... well, then still no. Because it's the host's L3, and not yours. And you get to share it with the host OS and the hypervisor too.
This is a weird assertion to make, I wouldn't waste too much effort trying to shore it up.
This is *exactly* what I'm saying and now you are arguing with it. That's the goal here. To run Windows/XP-32 bit applications that ran just fine on a machine from 20 years ago on par with a machine from 20 years ago. So you spent like seven posts and insane amount of vitriol to finally conclude the exact same thing. I'm still shaking my head. You just argued with me repeatedly and then repeated my original statement pretty much exactly as if it were your own conclusion!
If that is what you think happened, you aren't reading.
It
Re: (Score:1)
I thought Apple announced that Rosetta 2 would be available for Linux guests to use some while ago. Did that not work out?
It's available as a userspace interpreter- but not to run the actual OS code itself.
Anyhow, there's no reason in principle why Arm PCs couldn't use binary translation of x86 in VMs. Code that runs in VMs is ultimately just code.
Of course there isn't- that's how x86 VMs work right now.
The performance is just a lot worse than you would get from something like Rosetta, because emulation of the entire CPU (including MMUs, etc) is a lot more complicated than the pretty simple userspace environment.
Re: (Score:2)
It's more of a problem in the manufacturing sector, I guess, where you have to interface some old CNC machine who's software hasn't been updated since Obama was president.
Although, I've heard that hospitals have had similar problems with their equipment as well.
Re: (Score:2)
It's not like the software hasn't been updated. But when it costs 10s of thousands to get that update companies will do a lot of work to get around that
businesses do not want app store locked down syste (Score:2)
businesses do not want app store locked down systems?
They want the control you get now with domains / group policy. But the apple level lock down? No and not at the 30% cut.
Re: (Score:1)
As on every other OS, high end computing will still be in the realm of x86-64 which is already slowly transforming to remove its 8, 16 and 32-bit era dependencies. I think once you get to "pure x86_64" (aka Itanium 3.0) that Intel's focus on performance over power consumption will still give it a significant edge, hopefully they've learned that cooperation is better than the exclusivity they tried with Itanium 1 & 2.. Ampere is really currently the only contender in that space and they simply don't have
Re: (Score:2)
Last year, Intel made a paper on the x86S [intel.com] architecture. Pretty much it removes a lot of old stuff, moving that to be emulated or virtualized. Not sure if this will help things, but it will at least streamline the ISA.
I do wish SPARC and POWER were still in the running. Those used to be ahead of x86/amd64, but just over the years, have lost their lead.
Where I'd probably see a threat to Intel is in the RISC-V market. Because this isn't as patent encumbered as ARM, Chinese companies are making leaps and bo
Re: (Score:1)
Sure, if there is 1 RISC-V builder that can coalesce the market around itself. I don't think these kinds of billion-dollar investments for the current node sizes allow for a broad number of customized vendors (at least not yet). SPARC was great, PowerPC was great but they all went the way of the dodo because of exclusivity contracts and lock-in. They technically still exists but is locked into Oracle/IBM (zOS etc) and thus doesn't see much improvement/invention. SPARC was where Ampere is now with very large
Re: (Score:2)
If RISC-V could make a common bootloader that isn't patent encumbered and is open source, similar to SeaBIOS, that would greatly help their market, and expand trust in the architecture. Especially if no proprietary kernel blobs or drivers are needed to take advantage of all the SoC stuff. However, this is less of the actual RISC-V CPU, than getting board and component makers to agree on something.
Re: (Score:2)
Very much this. Might be an opportunity for a Linux hardware company like Pine64 to create a boot platform that other like-minded linux hardware companies can use and follow. Anything to get away from the devicetree rubbish that ARM (and current RISC-V SoCs) devices are hobbled with. Need something that implement ACPI. SeaBIOS is rather x86-specific and 32-bit, emulating the old BIOS system. But CoreBoot is a completely viable firmware for ARM and RISC-V systems.
Re: (Score:2)
AWS already offers their "graviton" compute instances. They're ARM-based machines, at a discount compared to AMD-64 images. It's a choice, and if you're running Java code or pure Python on linux containers, why not? I'm not familiar enough with the way Python binds in native code, and whether popular (read: AI) libraries are available for both AMD-64 and ARM to say whether transitioning to the less-expensive instances is always worthwhile, but AWS thinks enough of ARM to offer it as a service.
Re: (Score:2)
What percentage of PCs play games?
In my household there are 2 work laptops and two personally owned computers. Only 1 of those 4 plays games.
50% is ambitious, but I think achievable.
Re: (Score:2)
And a lot of people play browser/mobile games, or games on consoles. Gaming on a general purpose computer is quite a niche.
Re: (Score:2)
'for the most part' (famous last words) porting from x86/64 to arm64 is mostly trivial these days. Essentially no company is building CPU architicture specific pipelines outside of mmx/sse/avx and there are arm equivalents that are generally (again, famous last words) handle by the compiler.
Game companies should basically be able to include arm64 binaries pretty easily since microsoft has 'ported' directx etc over.
I think the problem that arm boxes will face is that intel/amd still have the high-power adva
Re: (Score:2)
There are server-class ARM Altra platforms right now that have DDR4, multiple 16 lane PCIe 4 slots and m.2 disk interfaces on board. I'm pretty curious if we'll see Snapdragon Elite systems with either storage or RAM upgradeable, but at least the full-fat server platforms have it regardless.
Re: (Score:2)
Unless ARM somehow convinces every Windows game and application developer on the planet to port their code to ARM native binaries in the next three years, they're going to continue to have performance issues due to their code running in an emulator.
The overwhelming majority of Windows machines are not used for gaming, and we've got plenty of experience to show that these days code translation is not a major performance setback for other applications - yeah it makes things inefficient, but the vast majority of PC applications are not CPU bound in performance.
Re: (Score:2)
> Unless ARM somehow convinces every Windows game and application developer on the planet to port their code to ARM native binaries in the next three years...
Apple went through this cycle, successfully. I think MS can do the same, if they follow a similar path. And porting an app is much much easier than it used to be in the past.
Also, I would hazard a guess that most computers sold are not gaming PCs. Many work apps these days run within a web browser.
Re: (Score:2)
The big difference between Apple and Microsoft is that Apple owns both the hardware and software experience. So, they can force their users to transition to any new platform, kicking and screaming if needed. I'd imagine that we'll see a lot of that, too, as Apple starts killing off support for older Intel based systems. It's not like you can go somewhere else if you want a supported Mac OS system.
Microsoft, on the other hand, needs to get the other hardware manufacturers like Dell/Lenovo/HP/Acer/Others to p
Re: (Score:2)
They're going to have performance issues regardless, because when it comes to raw performance at high power, ARM is awful in comparison to x64.
ARM's superiority is in its low power performance. Gaming is the opposite of that.
needs ram slots and at least the same pci-e lanes (Score:5, Insightful)
needs ram slots and at least the same pci-e lanes as what you get with an AMD or Intel cpu.
No boot loader lock in.
No app store lock in. (at least the MS store has non sand boxed win 32 apps in it)
Don't do the apple of non nvme storage with the big mark up and need to have 2th system to reload / change storage cards.
Don't lock out 3rd party video cards.
Re: (Score:2)
While I sympathize with your perspective, realistically, most laptop consumers are getting soldered memory and not having access to the pcie lanes.
I don't know if they plan on mirroring the secureboot situation from x86, but again, most users don't care.
The app store lockin would be a problem, and it did pretty much kill the Surface RT back in the day. I suspect this will not be an issue.
I would however bet the fact that "Intel" and "AMD" most comfortably support everything the fastest, and for now the ARM
Re:needs ram slots and at least the same pci-e lan (Score:4, Interesting)
While I like the idea, I think that we're a LONG way away from being to build your own custom gaming PC with a ARM processor.
For one thing, the motherboard makers will have to make "gaming" motherboards for the platform with the typical things that gamers want, like:
A processor socket with plenty of headroom for a custom cooler :)
Multiple free RAM slots
2 NVME drive slots
Multiple x16 PCI-E slots for a dedicated GPU (Which means that we also need good ARM video card drivers from AMD and NVidia, that might be stretch)
2.5 GB Ethernet ports and the latest Wi-Fi standard built in.
Lots of USB-A and USB-C ports, and a RGB header. Because it's not a "gaming" PC without an obnoxious amount of RGB
The other thing is that you're going to need to get the game developers to port their games to ARM native binaries, because "lag spikes" caused by emulation will be unacceptable.
Re: (Score:2)
I think we're not even going to see gaming PCs with ARM processors ever. Because there's no need to.
ARM isn't targeting the high end CPU market. Intel and AMD have got that area covered pretty well.
ARM is instead targeting the mid-range market - the average PC user who can do everything they need using mid range PC components - they aren't buying the Ryzen 7s or 9s, or Core i7s or i9s
Re: (Score:2)
needs ram slots and at least the same pci-e lanes as what you get with an AMD or Intel cpu.
It does not. They are targeting 50% of PCs not 100% of PCs. The overwhelming majority of PCs haven't been performance bound for a decade now. If it has at 8GB of RAM it may as well have a hamster doing sums in its head while running on his wheel for a CPU as far as modern applications are concerned.
Leave the PCI-e lanes to high performance enthusiasts who were never even going to consider buying this anyway.
Underachiever! (Score:4, Insightful)
Gee while you're shooting for the moon, why not claim 100% of the Windows market, and make it THIS year! Cause that's just about as likely to happen. The only possible path I see to even a small percentage, and it's really a longshot, would be if MS decides to ditch the "legacy" architecture and pull an Apple with emulators, etc. But MS is not Apple, and doesn't control the hardware platforms, and the Intel/AMD architecture isn't going away anytime soon, especially if Intel and AMD have any say in it.
Re: (Score:2)
Re: (Score:2)
Gee while you're shooting for the moon, why not claim 100% of the Windows market
Because that target is technically unachievable. To do that you need to also capture the high performance market and the gaming market, neither of which a foreign architecture has any chance of doing. The reality is right now, there's no reason 50% of computers can't be ARM, heck even previous generation ARM. We just don't really use our computers as much as people think.
Might be a good thing, create standardization (Score:2)
Maybe with Microsoft getting serious about ARM, we'll see some desktops and laptops with a standard set of hardware, perhaps including UEFI, and maybe that will enable better Linux support on ARM computers. That is if MS doesn't screw up and require a locked boot loader like they did before.
We have a number of standards for ARM that relatively few vendors support unfortunately. It's all quite confusing to a person used to using standardized PCs with chipsets from AMD and Intel. See https://uefi.org/sites [uefi.org]
Re: (Score:3)
I agree with your post though I think you aren't seeing it from Microsoft's perspective. Vendor lock-in is a feature. Hence, a locked boot-loader is a feature.
I also doubt you'll see any kind of unified hardware platform such as Apple or any of the phone's. PC hardware has many many choices. This is great if you are a hardware vendor but not so great as a software developer. More hardware choices means more support issues.
If you always stick to a single hardware vendor, aka Dell, HP, etc, then you get close
Re: (Score:2)
The CPU architecture is already a standard, what's missing is a standard method of booting and baseline support for peripherals (eg things like generic vga - just enough to bootstrap so you can load drivers for the actual hardware).
ARM Windows is less capable than ARM Linux (Score:5, Insightful)
Re: (Score:2)
Re: (Score:2)
You can recompile *if* you have the source code.
On Linux the vast majority of software comes with source code so you can do this quite easily. On windows the vast majority of software is supplied without source code.
Re: (Score:2)
Re: (Score:2)
With Windows, you're stuck with proprietary software that runs in slow emulation.
I see you've never used ARM emulation before. Or at least maybe you have once over a decade ago. I had a Surface Pro X for a while and it performed perfectly adequately for virtually anything most people use their devices for. No you're not running complex simulations or using them in a render farm or gaming on them, but as far as an office suite, video watching, or browsing (that covers the use cases for most PCs) is concerned even ARM from several generations back using old emulators worked perfectly fine
Re: (Score:2)
Does it really matter? Linux's backward is pretty impressive; just a few years ago, I installed (the native Linux version of) UT99 on my Linux machine and it worked fine.
I care about the total package (Score:1)
Every single app I run has an ARM version these days, or is made by Microsoft and will be recompiled.
I wanted a 4k+, ~13" HDR touchscreen in a thin, light laptop. Thus I ended up with a Dell XPS 13 Plus. If I can get 50% more battery life and the same or better performance in an ARM chip while keeping those features, I'll take it in my next laptop.
Why don't I have a Macbook Air? No touch. No 4k. I don't much like MacOS, but if it was touch and 4k, I'd have made the move.
This will be bad, really bad (Score:2)
People who use simple, popular stuff will be fine
Unfortunately, there is a LOT of old x86 code running specialized applications
Much of it will never be rewritten foe ARM, because the company has lost interest or has gone out of business
A lot of custom programs were made by consultants or programmers who are long gone
Some stuff may be rewritten at exorbitant cost
Computers are used for a LOT more than email and web browsing
Re: (Score:2)
Re: (Score:2)
And when it inevitably doesn't work properly, the correct answer to those being confused why their windows doesn't run things it always ran fine is "fuck you".
And then to publicly moan why windows RT v2 wasn't successful.
Re: (Score:2)
Re: (Score:2)
You confuse "depend on" with "will run into random intermittent problems".
Former may be a very low number. And true that it's also pretty irrelevant in the grand scheme of things.
Latter? Large and exceedingly relevant, to the point where both microsoft and those that develop for windows spend billions if not trillions to keep their software up to date.
My Fear (Score:2)
What I'm wondering is... (Score:2)
ARM's CEO has a bridge he wants us to buy (Score:2)
Where are your new factories? (Score:2)
The current generation of MS Surface is ARM only (Score:2)
The current generation of the Microsoft Surface Pro and Surface Laptop are ARM only.