Microsoft today released a prerecorded video timed for Build 2017 that details Windows 10 on ARM and how its x86 emulation software works.
You can watch the video for yourself, of course, and it’s only 13 minutes long. But here are the highlights.
Sign up for our new free newsletter to get three time-saving tips each Friday — and get free copies of Paul Thurrott's Windows 11 and Windows 10 Field Guides (normally $9.99) as a special welcome gift!
"*" indicates required fields
Why Windows 10 on ARM? As Terry Myerson explained last December when this initiative was first announced, Windows 10 on ARM seeks to address key customer needs: Always-on connectivity through the addition of integrated LTE capabilities and great battery life.
It’s not Windows RT. Windows 10 on ARM will offer the full desktop experience, including iconic Windows 10 features like Cortana, Edge, and Windows Hello. It will include the same apps and work with all Store apps, plus x86 Win32 apps in emulation.
It is Windows 10 Pro. The demo machine (see below) is running a 64-bit version of Windows 10 Pro on an 8-core Qualcomm Snapdragon 835 processor with 4 GB of RAM. It’s “real” Windows 10 Pro, too, meaning that windows 10 Pro-specific features like domain join work as expected.
It’s not even close to ready yet. As the Microsofties in the video demonstrate, they’re using an early engineering sample provided by Qualcomm to run Windows 10 on ARM. It’s not a laptop or a hybrid device, but just a black box. The final devices will be “2-in-1s and laptops” from a variety of PC makers, they say.
It will have “great” device support. In addition to running Windows apps, Windows 10 on ARM supports a wide range of modern (often USB-based) devices using in-box (and ARM-based) class drivers. (This is going to be an issue with Windows 10 S, by the way, since you can’t run x86 Setup utilities on that system.)
The x86 emulation layer works. Native apps like Edge, of course, work normally, but this system’s ability to run x86 apps is, of course, of more interest. Based on the handful of x86 apps they run in the video, performance looks solid.
How does it work? According to Microsoft, running x86 apps (normal desktop apps) is transparent to the user (and to the developer of those apps); they just work normally as on any other PC. Emulation works as it always has in Windows: There is a WOW layer (Windows On Windows) that abstracts the underlying kernel, drivers, and other system components and makes the PC/OS look and work like the real thing. 64-bit Intel versions of Windows 10 do the same thing when running 32-bit code too. But the CPU isn’t emulated in software; that happens in hardware. On ARM, CPU emulation happens in software.
UWP apps run better. x86 run, but you will of course have the best experience with native UWP apps. These apps won’t task the processor, RAM, or battery as hard as emulated x86 code. They’re native to ARM on this platform.
skane2600
<blockquote><a href="#115956"><em>In reply to Jorge Garcia:</em></a></blockquote><p>I just don't get what advantages a desktop-ish version of iOS would have over the existing MacOS. IMO it would end up being less appropriate for mobile than iOS is today and less appropriate for desktop than MacOS is today. Combining devices that have different needs and use cases rarely results in an optimal solution.</p>
skane2600
<blockquote><a href="#116018"><em>In reply to Jorge Garcia:</em></a></blockquote><p>You're right, it is a crazy analogy, or at least not a very valid one. CUVs, sedans, and Vans all are non-commercial vehicles that are, at their core, the same. They all can carry people and cargo and they all share a common "user interface" (although different makes and models have slight differences, they aren't based on vehicle class). Despite the marketing pitch CUVs aren't really a combination of any two vehicle types, they are just tall vehicles with some extra space. I don't like these kind of analogies, but a better one would be that Windows and Macs are cars and iOS and Android are motorcycles. The closest combination would probably be a motorized tricycle and those aren't very popular.</p><p><br></p><p>But we don't really need to consider vehicles when we can actually look directly at the differences between iOS and MacOS.</p>
skane2600
<blockquote><a href="#116090"><em>In reply to Ugur:</em></a></blockquote><p>Yes. Although sometimes there is unnecessary complexity, often an increase in complexity is inevitably tied to increased capability. The old fashioned "toaster" interface was simple because it gave you very limited control over toasting.</p><p><br></p><p>Jobs was against a mini version of an iPad and I honestly think he made the original iPad as small as he could while still retaining the possibility of replacing a PC. He certainly didn't see smartphones as PC replacements. Yet the marketplace embraced the iPad mini (at least for awhile) and smartphones continued to be popular. I agree that for most people these devices are primarily about consumption and social communication.</p>
skane2600
<blockquote><a href="#116267"><em>In reply to Jorge Garcia:</em></a></blockquote><p>It depends on what kind of "computing" you are talking about. I can take an old TV without any digital components and pick up an analog TV broadcast from Mexico or I can watch Netflix on my Windows Phone that obviously involves a lot of computing, but fundamentally it's the same activity – I'm watching video. Likewise my modern dishwasher has a microprocessor in it but serves the same fundamental purpose as the old mechanical dishwasher my mom used in the 60s. These kind of activities aren't really "computing" in the same sense as using a computer as a tool for intellectual work. The fact that many people own smartphones doesn't mean that a lot of them are doing traditional "computing" on them. Other than consuming content most of what people do on smartphones is a natural extension of a phone's fundamental purpose – to communicate. </p>
skane2600
<p>I suspect that the emulation is the reason for the delay. Demos should always be taken with a grain of salt. It doesn't seem that these devices require substantially different hardware than what already exists on many different devices. The original demo was 6 months ago, plenty of time to design the devices.</p>
jbuccola
<blockquote><a href="#116191"><em>In reply to lvthunder:</em></a></blockquote><p>The smart phone market has distortions in it, like carrier subsidies and monthly payment programs, that tweak the purchase price in ways that don't normally affect tablets.</p><p><br></p><p>One that has always raised my eyebrows is the iPad, which is now about 1/2 the cost of a similarly speced iPhone. Nearly identical device, larger screen — cheaper. </p>
jbuccola
<blockquote><a href="#116072"><em>In reply to wright_is:</em></a></blockquote><p>When the LTE is engaged and connected on Intel devices, it severely impacts battery life.</p><p><br></p><p>With ARM-based tablets and phones, that's just not the case.</p>
skane2600
<p>I can see one argument for MS making a version of Windows on ARM because ARM has become the "toast of the town", but I think it remains to be seen whether they'll be a compelling reason to choose an ARM device over an Intel one. The only place that Windows on ARM would be potentially useful is on a 2-in-1 or on a laptop. Those devices are the least sensitive to power consumption of all mobile devices. Small tablets and smartphones have the wrong ergonomics for running legacy programs.</p>
skane2600
<blockquote><a href="#116246"><em>In reply to trevor_chdwck:</em></a></blockquote><p>It doesn't matter if a battery cost comes from the alleged "inefficiency" of X86 or because of an emulation error, it's still a cost. Since the primary purpose of a Windows on ARM device is to run X96 applications, the fact that "modern" apps are allegedly less power hungry is not particularly significant.</p><p><br></p><p>But there are reasons to doubt the level of compatibility, the power consumption advantages, and the comparative cost. We'll have to see what happens when a real product emerges and it is tested against a myriad of legacy apps to determine its relative value.</p>
skane2600
<blockquote><a href="#116312"><em>In reply to trevor_chdwck:</em></a></blockquote><p>It remains to be seen how many Win32 applications will use centennial to bring them to the Windows Store, but those applications will still be fundamentally Win32 apps with a different method for installation than they previously used. They aren't designed to be suspended in the same way UWP apps are. </p>
skane2600
<blockquote><a href="#116517"><em>In reply to Waethorn:</em></a></blockquote><p>I don't think we are in disagreement. As I said, Centennial apps are still fundamentally Win32 and thus you are correct to say they aren't compiled for ARM. But the other optimization limitation of Centennial apps is that they aren't designed to follow the UWP life cycle which purports to be a memory and power saver, Centennial apps don't know they are supposed to save state when they are suspended in case they are later abruptly terminated by the OS. I don't know how this is handled but I believe the only two options are for the OS to keep a Centennial app running when it's not in focus (thus not optimizing power or memory) or potentially losing its state. Hopefully it's not the latter which would be quite unexpected behavior.</p>
skane2600
<blockquote><a href="#116199"><em>In reply to SvenJ:</em></a></blockquote><p>The proposed ban is only for laptops in the cabin, not in checked baggage. They could ban smartphones in the cabin just as easily. </p>
skane2600
<p>Do they really mean 32-bit only, or are they using x86 to refer to desktop applications generically? If they mean the former, WoA is DOA. Nobody wants partial Windows compatibility. </p>