M$ announces ARM emulator soon

[ejolson]

The funny thing is that M$ had a Windows NT running on Digital Alpha architecture long ago.

We've had Windows X11 on Unix since 1987

Smokey Wallace of DEC WSL and Jim Gettys proposed that DEC WSL build X11 and make it freely available under the same terms as X9 and X10. This process started in May 1986, with the protocol finalized in August. Alpha testing of the software started in February 1987, beta-testing in May; the release of X11 finally occurred on 15 September 1987.

and the Pi has been running X11 since it was first shipped. What is all this fuss about Windows 10? Is it something to do with how trademark laws work in certain countries?

Creating efficient closed-source emulators for x86 instructions on ARM is made easy because the tiny code generator used in QEMU which does the just in time translation of x86 to ARM is BSD licensed. It would be interesting to know how much (if any) code from TCG Microsoft used and how the speeds compare.

[Heater]

DEC Alpha was great.

However there was no way your average consumer was going to be buying an Alpha machine.

That was the end of that.

[alphanumeric]

I don't see the common interface at this point in time on Windows devices? My XBOX ONE is about as different from my PC GUI wise, as they can be while still both running Windows 10. I don't own a Windows phone but my guess is its closer to my XBOX than my PC, Interface wise. Or even different again from those two? And Windows 10 IoT is different again. I don't know how they can even call it Windows 10. There is no GUI to speak of really. Even so, I'd still like to see something along the lines of 10 Mobile that will run on a Pi. Or like what's on my XBOX?

It's marketing. MS has the brand--"Windows 10"--so every time that is mentioned for a given device, no matter how different it may be from the x86 desktop system, either internally or externally, many people *think* it's really the same thing. Look at how many threads we've gotten here from people who have heard that "Windows 10" runs on the Pi and they want to do that, and then we have to explain that "Win 10 IoT" (which does run on a Pi3) is decidedly NOT the Win 10 they're looking for.

Every time I mention on a forum that I would like to see Windows 10 Mobile run on a Pi. I get the "Windows 10 already runs on a Pi" reply. The when I ask if they have ever run 10 IoT I get a "well no, but its Windows 10 right? Then as you mentioned, I have to explain the difference between it and any other version of Windows. face palm.

[HawaiianPi]

[sigh...] Not this again.

Even if MS does make an ARM version of Winbloat-10, and they include x86 emulation, it's not going to be some magic that will give you a $35 Windows 10 PC. Emulation requires a significantly more powerful processor than the one you are emulating to run at decent speed. In the case of the Raspberry Pi (even the Pi3) we are dealing with a significantly LESS powerful processor than what is available in most modern Windows machines. This means your x86 applications will run like crap. Sure, you can run DOSbox and ancient applications from decades ago, but trying to run any modern application is going to be disappointing.

And that's BEFORE we factor in the limited memory on the Pi3. Add that in and now you're trying to run Windows-10, and an x86 emulation layer, and the x86 application you want to run, all in only 1GB of RAM.

Seriously people, give it up already. If you want a Windows 10 PC, go and buy one. The Raspberry Pi is NOT a Windows PC!

[Heater]

alphanumeric ,

Every time I mention on a forum that I would like to see Windows 10 Mobile run on a Pi. I get the "Windows 10 already runs on a Pi" reply.

That there is why the Pi exists at all. To teach people some minimal basics of computing such that they don't think that what they buy off the shelf is all there can ever be, like some cargo cult.

[hippy]

I don't want everyone to move off of Windows. I want eveyone to have a choice--and to know that choice exists--to move off (or onto) Windows as they wish.

I would guess that's Microsoft's view as well. If people are using ARM hardware then Microsoft won't get a look-in unless they can make their OS and software run on an ARM platform, and it's not unreasonable for them to go after that market, offer an alternative to what else does run on ARM.

I will note that I think the whole user interface issue is a red herring at this point. Never mind that I can easily move back and forth between Windows and Linux, my 8-year-old grandson does it routinely. The computers at school are running Windows while his own computer at home is a Pi running Raspbian. I'm not sure he even notices the differences.

I doubt he does, kids are like that. Some adults are too but not all find such things so easy, and many aren't always open minded or flexible in their outlook. People get used to the way things are, stuck in their ways, get irritable and awkward when things are just a little different to how they have been or how they think they should be. They are the people who will ask "does it run Windows?" and walk away if it doesn't.

In time the kids who are more flexible and accommodating will supplant the older generations we have but that's not going to happen overnight.

[broe23]

Sounds more like they are trying to reinvent the Surface RT again. It will most likely not be a public available copy of the 2017 release of Redstone 3. Microsoft is dying in the hardware sector that is branded with their name. They continue to see losses and also that there has been a very high rate of failure of the Surface 3 and 4. Especially the power adapter.

Big difference this time around though. RT couldn't run 32-bit Win32/x86 desktop applications.

This is full Windows 10 for PCs, not some stripped down version. It’s Windows 10 Home and Pro, on ARM. And Windows 10 Enterprise, with all the functionality that businesses expect, including domain join. This is Windows RT done right.
Even better, Windows 10 on ARM will supply a long-rumored feature: The ability to run 32-bit Win32/x86 desktop applications—Apple iTunes, Adobe Photoshop, Google Chrome, whatever—directly on the system, unchanged.

The sad thing is that it was the consumers fault on that one, when Microsoft started to have people complain. It is just like back when XP was released, that Microsoft had a tablet and Gates complained that it did not have Outlook on it, even though it was not asked by him to have the developers and coders to find a way to place Outlook on it, let alone make a cloud version. Microsoft has been their own worse enemy for decades.

[broe23]

But heck, when it comes to it I'm going to get in there one way or another

Don't bet on it: https://www.ifixit.com/Teardown/Microso ... down/51568

Microsoft Surface Pro 4 Repairability Score: 2 out of 10 (10 is easiest to repair).

That unit, just like the newer units for the Macbook line are being made to not be user service capable, along with the fact that they are doing away with the larger motherboards, so that they can put in more battery packs to extend the time that you do not have to be tethered to a outlet.

The company that is producing the Surface line and other OEM's are finally taking a queue from Apple in how to make a lighter weight and longer battery life.

[peterlite]

Windows 10 on a Pi Zero with an 8 GB card? Where does Microsoft fit the 60 GB of advertising for Microsoft Office, etc, and the 20 GB of Microsoft Office, etc, time limited trials?

[tweak42]

Windows 10 on a Pi Zero with an 8 GB card? Where does Microsoft fit the 60 GB of advertising for Microsoft Office, etc, and the 20 GB of Microsoft Office, etc, time limited trials?

If Microsoft can get it running on something as small and inexpensive as the Zero, it would probably be cloud based like the current Chromebooks. So minimal local storage needed, and freemium subscription service required to use.

[W. H. Heydt]

[sigh...]
And that's BEFORE we factor in the limited memory on the Pi3. Add that in and now you're trying to run Windows-10, and an x86 emulation layer, and the x86 application you want to run, all in only 1GB of RAM.

I'm going to be a bit of language purist here for a minute... This strkes to the heart of the difference between a simulator and an emulator. Formally, a simulator is just software while and emulator includes some hardware to support a simulation process.

One could, at least in principle, include hardware that would take the x86 instructions and treat each one as a microcode macro and translate it into a series of actual machine instructions. The x86 does this internally, and it is was the basic idea behind the Transputer. IF you do that, you could run x86 instructions on a non-x86 SoC within spitting distance of the native speed of an actual x86 chip.

If you do it strictly in software (a simulator) then, yes, it's going to something like an order of magnitude slower than the architecture being simulated, though even there, a lot will depend in just how dissimilar the two architectures actually are.

[lumsdot]

the LattePanda http://www.lattepanda.com
is a good way to run windows 10 on a small form factor board.

[fruitoftheloom]

the LattePanda http://www.lattepanda.com
is a good way to run windows 10 on a small form factor board.

Other x86 SBC's are available:

http://up-shop.org/up-boards/2-up-board ... emory.html

http://www.jaguarboard.org/index.php/pr ... based.html

[HawaiianPi]

I'm going to be a bit of language purist here for a minute... This strikes to the heart of the difference between a simulator and an emulator. Formally, a simulator is just software while and emulator includes some hardware to support a simulation process...

We can agree to disagree on that point. There are software emulators, and have been for a long time. While hardware can certainly be used to augment the process, it is not necessary by definition.

Emulation is the process of mimicking the outwardly observable behavior to match an existing target. The internal state of the emulation mechanism does not have to accurately reflect the internal state of the target which it is emulating. In other words, I want to run x86 code on an ARM SOC, but I don't care how it's done as long as it works.

Simulation, on the other hand, involves modeling the underlying state of the target. In this case I want to see exactly how an x86 processor works when running code. The end result of a good simulation is that the model will emulate the target, but the process is much more complex due to matching the internal behavior of the target as well as the functional result.

[ejolson]

One could, at least in principle, include hardware that would take the x86 instructions and treat each one as a microcode macro and translate it into a series of actual machine instructions. The x86 does this internally, and it is was the basic idea behind the Transputer. IF you do that, you could run x86 instructions on a non-x86 SoC within spitting distance of the native speed of an actual x86 chip.

If you do it strictly in software (a simulator) then, yes, it's going to something like an order of magnitude slower than the architecture being simulated, though even there, a lot will depend in just how dissimilar the two architectures actually are.

Modern simulators like QEMU use just in time compilers to run x86 code on ARM. Just in time compilation is also used by Java virtual machines and in modern web browsers for JavaScript. The result is slower, but not orders of magnitude slower than properly compiled code.

At the same time, even properly compiled code runs orders of magnitude slower on the Pi. This is simply because the Pi hardware itself is orders of magnitude slower. The effect is noticeable, for example, when running the Chromium web browser in Raspbian.