“Project Volterra” review: Microsoft’s $600 Arm PC that almost doesn’t suck

There's a reason it isn't a Surface PC, but it's good for its intended purpose.

See full article...

 

[olgrandad]

Might make a great little Linux box, though I suspect the installation won't be straightforward.

 

[Rowboat]

Is there an end goal for windows on ARM or target users?

 

[cramleir]

I think the end goal is less to transition the ecosystem to ARM en masse a la Apple, but to allow ARM vendors to compete with Intel and AMD and drive innovation. However it all shakes out in the end is fine by Microsoft,

 

[tenocticatl]

I feel like the end of x86 is eventually inevitable, but I shudder to think what it will do for backwards compatibility. There's a lot of old abandonware that I'd really like to be able to keep using on any future devices, mainly games.

 

[StikyPad]

I feel like the end of x86 is eventually inevitable, but I shudder to think what it will do for backwards compatibility. There's a lot of old abandonware that I'd really like to be able to keep using on any future devices, mainly games.

I'm not sure x86 will ever really die, but if it does, it will be very slow indeed. At that point, any vestigial apps (especially what is already "old abandonware" today) will run just fine through emulation.

 

[crazy_vag]

I feel like the end of x86 is eventually inevitable, but I shudder to think what it will do for backwards compatibility. There's a lot of old abandonware that I'd really like to be able to keep using on any future devices, mainly games.

I agree. It really depends on how well you can implement x86 emulation. Apple seems to have done it pretty well for most apps.

I bet you could support 80% of apps pretty easily in a year or so.
Then spend another year to get that up to 90% of apps supporting the ones that need special APIs.
Another year to get the ones requiring low-level CPU access to reach 95%.
By then many of the remaining apps will have been recompiled for native ARM - this will be made easier by Apple.

 

[ScifiGeek]

This feels like a distinction without a difference.

Anyone who spends $600 on such an underdone product has rocks in their head, and suggesting anyone put up with this nonsense is doing them a disservice

This isn't aimed at general consumers, but for use as developer test box. The continued improvements in both the WinARM software and HW will eventually deliver a consumer product.

 

[malcolm_]

I wonder who is it really for. "Dev Kit" suggests developers - but with hardware this weak, the device can merely serve as a terminal for a much more powerful remote machine. Perhaps that's the intended purpose.

The intended use is for developers to test the functionality and performance of their ARM port on real hardware at a reasonable price. I doubt even Microsoft really expect it to be a "daily driver".

 

[RickRoyLeonPrisZhoraRacha]

The Dev Kit is a lot cheaper than a Mac mini once you take specs into account—both devices start at $600, but in the Dev Kit, that gets you 32GB of RAM and 512GB of storage. The mini doesn't offer that much memory, but upgrading to 16GB of RAM and 512GB of storage adds another $400 to the price tag.

See. Right here. Its about cost. Apple's Mini is way overpriced per features. $1400 for a 16GB/1TB is just crazy when you can have TWO simple 32GB/512GB ARM boxes for $1200. But is there a dev Mac mini price?

 

[ScifiGeek]

Is there an end goal for windows on ARM or target users?

Making WinARM viable for general consumers. It's not there yet but could be in a couple of years.

I wouldn't be surprised if within 5 years, most general purpose Windows Laptops sold were ARM based.

I don't think Microsoft will force the change, they are just trying to make WinARM viable, and then let the market decide where it goes after that.

 

[gijames1225]

The degree to which I want this (a lot) is completely inverse to the need I have for it. Little things that compute are just so collectible though!

 

[dwrd]

If you're writing windows apps, seems easy enough to plunk one of these on your desk to debug your ARM builds. I don't see a lot of utility for it beyond that niche.

 

[gijames1225]

I wonder who is it really for. "Dev Kit" suggests developers - but with hardware this weak, the device can merely serve as a terminal for a much more powerful remote machine. Perhaps that's the intended purpose.

If you are compiling Win Forms or .NET MAUI apps the this would be perfectly functional for everything but the craziest projects. Sure, it could always be faster, but it's certainly better to do this on the actual architecture you're aiming for rather than not.

Not noted here, but very relevant for the developers and timing around this is a native ARM version of Visual Studio just launched and .NET 7 as well saw it's full release last week. Basically, there's not much excuse going forward in the .NET space to not have an ARM binary on tap.

 

[thomsirveaux]

The intended use is for developers to test the functionality and performance of their ARM port on real hardware at a reasonable price. I doubt even Microsoft really expect it to be a "daily driver".

I think that's true of this particular box, though note the similarity with the Surface Pro 9 5G hardware - Microsoft does intend very similar hardware to be capable of daily driving, despite the same shortcomings (and an even higher price.)

 

[fknuckles]

This feels like a distinction without a difference.

Anyone who spends $600 on such an underdone product has rocks in their head, and suggesting anyone put up with this nonsense is doing them a disservice

Out of the gate the Apple M1 worked impressively well, even in the earliest of days when a lot of your software was running through Rosetta. Today? I don’t think any of the apps I run aren’t native.

Clearly this Qualcomm chip and WinARM are not ready for prime time, and if they really wanted to simply get it out there as a dev kit to spur adoption and encourage ARM software development for Windows it would be half the price.

IIRC, Apple also launched a dev-focused device ahead of the proper launch of the M1 suite.

 

[Unsheept]

Is it weird that I want one, even though I haven't come up with a "need" for it (yet)?

 

[skazz]

Well if you compare pricing of this device to building a NUC or the Mac Mini, you actually get quite a lot of hardware+software for what is a reasonable price. And the styling is certainly better than many mini PCs.

So as long as your needs are met despite the limitations of ARM with Windows, it's actually quite an interesting choice.

 

[bryanlarsen]

ARM is an insurance policy for Windows. If Intel & AMD both #@% up, then Microsoft Windows is %#@$ed. This exists to prevent Intel & AMD from doing a rug pull to Windows. So don't expect a Microsoft transition to ARM, this just exists to make that transition easier if it ever becomes necessary.

This is similar to Steam on Linux. Valve is quite happy selling Steam on Windows, but knows they are vulnerable to a rug pull. They put enough effort into Steam on Linux to ensure that they can transition if they have to, but still hope that they never have to. Its existence makes it less likely that the rug does get pulled.

 

[whiteknave]

The bad​

• Qualcomm's best hardware still falls short of the 2-year-old M1, to say nothing of the M2 or more powerful M1 variants

I think that we are still in a chicken/egg situation with Desktop ARM hardware and Windows (similar to how Android Wearables stagnated for years). Once Microsoft gets Windows on ARM in a better shape, I think that we'll see other hardware folks introduce some new desktop ARM chip designs, most likely Samsung. Maybe even AMD (who has an ARM license and was supposedly working on an ARM-based chip years ago parallel with the original Zen design) or Google. However, Qualcomm and any other ARM desktop chip maker will probably always trail Apple in the leading edge performance.

It will be also be interesting to see if Intel can convince Microsoft to make a Windows on RISC-V version.

 

[whiteknave]

ARM is an insurance policy for Windows. If Intel & AMD both #@% up, then Microsoft Windows is %#@$ed. This exists to prevent Intel & AMD from doing a rug pull to Windows. So don't expect a Microsoft transition to ARM, this just exists to make that transition easier if it ever becomes necessary.

I think that ARM on the server will be much more commonplace than ARM on desktop for a while. To my knowledge, ARM servers are all linux-based. Maybe Microsoft is iterating Windows on ARM to get to a server version.

Long term, I think that x86/CISC-based platforms will eventually die with maybe some niche use cases keeping it alive. Eventually, ARM and RISC-V will overtake and replace x86. However, that will be many years from now. Unless Intel/AMD can figure out how to make x86 as power efficient as ARM designs.

 

[flunk]

"the fastest Arm chip that can run Windows."

This assertion is not strictly true now that you can install Windows 11 on a VM on M1 and M2 Macs. Fastest supported Arm chip that can run Windows, yes.

 

[flunk]

I think that ARM on the server will be much more commonplace than ARM on desktop for a while. To my knowledge, ARM servers are all linux-based. Maybe Microsoft is iterating Windows on ARM to get to a server version.

Long term, I think that x86/CISC-based platforms will eventually die with maybe some niche use cases keeping it alive. Eventually, ARM and RISC-V will overtake and replace x86. However, that will be many years from now. Unless Intel/AMD can figure out how to make x86 as power efficient as ARM designs.

For web servers/application servers. As long as your platform supports the OS it generally doesn't matter which one you use. It's different on the desktop. I've been running .NET on Linux servers for years, .NET now supports ARM on Linux so I can switch whenever it makes sense to me. Most web server platforms support ARM on Linux or BSD and some also support Windows. In general it doesn't matter as long as it'll run your code.

 

[sfmitch]

"Apple's Rosetta technology has some of these problems, too; scrolling in a Rosetta app, for instance, is noticeably jerkier than scrolling in an Apple Silicon-native app."

Not that I've ever seen. Rosetta 2 apps have been indistinguishable from M1 apps in my usage.

 

[Qwertilot]

Making WinARM viable for general consumers. It's not there yet but could be in a couple of years.

Is it going to get there so fast? People have been selling ARM based things running Windows for quite a few years now, the progress on software really does seem to have been rather glacial to date.

Almost depressingly so compared to Apple, say, although obviously very different ecosystems.

 

[PhilipStorry]

I'm not sure x86 will ever really die, but if it does, it will be very slow indeed. At that point, any vestigial apps (especially what is already "old abandonware" today) will run just fine through emulation.

There's lots of industrial and medical devices still being driven by 32-bit Windows XP machines and even old DOS hardware. It's a niche, and one which seems rather stubborn.
When an organisation is spending hundreds of thousands or even millions on a medical imaging device or a forge, the ~$1000 PC that controls it is not really a consideration. They get air-gapped (hopefully!) and just plod along, because replacing them would mean also replacing whatever they control!
Eventually, systems administrators/technicians start going to eBay and hoarding spare parts, because losing a simple ATA disk can cost millions...
So yeah, x86 is going to be in certain niches for a few decades or so.

 

[ScifiGeek]

Is it going to get there so fast? People have been selling ARM based things running Windows for quite a few years now, the progress on software really does seem to have been rather glacial to date.

Almost depressingly so compared to Apple, say, although obviously very different ecosystems.

A big piece fell into place with 64bit emulation, a couple more years of refinement, making critical pieces native, and getting a real Apple Mx challenger from the Nuvia acquisition could push it over the line, into reasonable alternative.

 

[MightyPez]

There's lots of industrial and medical devices still being driven by 32-bit Windows XP machines and even old DOS hardware. It's a niche, and one which seems rather stubborn.
When an organisation is spending hundreds of thousands or even millions on a medical imaging device or a forge, the ~$1000 PC that controls it is not really a consideration. They get air-gapped (hopefully!) and just plod along, because replacing them would mean also replacing whatever they control!
Eventually, systems administrators/technicians start going to eBay and hoarding spare parts, because losing a simple ATA disk can cost millions...
So yeah, x86 is going to be in certain niches for a few decades or so.

This is depressingly true, particularly in medical research. An organization will buy or be donated a $500,000piece of equipment with a $300 Dell controlling it from a serial port. The vendor calls it good and will never ever provide any support or path to update that Dell without replacing the entire instrument.

Part of it is poor vendor/purchase management from the organization. CEO/Director/Doctor/Scientitician/whatever who gets what they want in every instance was wowed by a slick talker at a conference and will bypass any RFP to get it. The other part of it is these vendors can get away with it.

If you ever want to make bank as a consultant, learn the ins and outs of research equipment and find a way to shoe-horn in off the shelf parts like a cheap PC or a Raspberry Pi to control them.

 

[Unsheept]

Just a heads up: Microsoft has a strict no-returns policy with this Dev Kit.

Microsoft is well aware of Windows on Arm causing buyers remorse and is unwilling to lose a single penny to disappointed buyers.

Not very confident. Not very friendly.

//

The article lays out great evidence to show how closely this is just a Surface PCB.

No headphone jack makes sense: most people will remote in anyways.

Great callout, thanks! I wasn't aware.

I disagree on the not confident / not friendly comment - this isn't intended to be a consumer PC, it's a Dev kit. It's meant for software engineers who are going to use it as a tool for a very specific purpose for which it's either suited or it's not. Their site is very clear about this, although maybe they should limit or "gate" purchases, for example through the Microsoft developer program or similar.

 

[Fast Turtle]

I wonder who is it really for. "Dev Kit" suggests developers - but with hardware this weak, the device can merely serve as a terminal for a much more powerful remote machine. Perhaps that's the intended purpose.

Too many devs think they need engineering workstations (Threadripper/Xeon) to write code on that they forget the purpose. No wonder Chrome requires a mid to high end CPU to even run worth a damn now isn't it?

Keep in mind that Windows 10 still runs on Mips, Itanium, Arm and Arm64 along with a host of other CPU types. MS has the core OS ready to run should Intel go out of business or a major change in Computer Paradigns happen and unlike Apple their primary business apps already run natively on the various cpu types.

From a corporate standpoint, all of MS various office software already runs natively on Arm64 so this is to get developers onboard with the new VS/DotNet IDE. Too many times, Devs think they need a high end engineering workstation and forget about all the cube drones using lowly dual core Haswell Celerons and yes, there are lots of them out there.

I can also see MS offering an Azure Server for business, similar to the Google Search Boxes they'd offered for a while. Yep, it would be leased instead of owned but then would only need a single update for all of these Arm64 systems to be usable.

 

[Maury Markowitz]

But x86 apps running on the Dev Kit regularly stutter or hesitate when rendering, waiting on the translation to happen before responding to input.

I'm having trouble understand this bit. Wouldn't whatever code you need to display... whatever, already be translated? Why is it doing code translation while you scroll?

Is this not more likely the issue of having to pass the data through multiple layers when a native version only has one?

 

[ubercurmudgeon]

I probably won't buy one myself, but the upgradable storage at least makes it a possibility, unlike the Mac Mini.