Windows 7 will be ARMless

Even if one gives MS the benefit of the doubt that windows is portable enough to be compiled for ARM without excessive effort (and I think it actually is portable), then the problem is that no existing windows app will run on it.

The linux world is used to portability, the source code is almost always available, and ready to be compiled for a gazillion different architectures.

In the windows world, compiled binaries rule, and those will be utterly useless on an ARM device, and that would cause a lot of confusion for the end user.

I can understand MS doesn't want to do this. Basically they cant. At least not without a huge effort to re-educate its developers and its customers.

OK so Intel's biggest advantage over ARM in the netbook / notebook space is that desktop Windows does not run on ARM. Linux-based notebooks and desktops have categorically failed in the market.

Based on Intel's normal behaviour, do we not suspect that Intel may somehow be involved with stopping Microsoft from supporting their key competitors and keeping a monopoly of x86 on Windows?

Intel are certainly applying pressure on Microsoft by investing hard in mobile Linux.

aren't you contradicting yourself here?

if Intel are investing hard in Mobile linux then they're supporting the Linux ecosystem and thereby aiding the ARM movement...

Nick, do you remember the '90s? How about the word 'WinTel'?

It does not so much create an opening for Linux derivatives as it makes ARM unattractive to manufacturers because no Windows available.

Who benefits from that? What does that suggest to you? Is there an applicable lesson in recent history?

No OS for ARM??? I'm still wondering when M$ will ever release an OS for Intel! Perhaps they are trying to get a heart for Gates or brain for Ballmer first.

Seriously. What flavor Windows OS besides CE/WM runs on ARM? Did MS state they were going to support Windows on non-x86 platforms before today?

Real news would be if they stated they were going to drop ARM from CE, but that would make alotta people very cross. ARM makes up a majority of the GPS/PDA/gadget market.

Although, it would be cool if ARM bolted on an x86 emulation layer, ala Tansmeta.

Waste of perfectly good ARM instruction set CPU cycles.

What's the point of running 7 or Vista on an arm device or processor anyway, it's too heavy and would run like crap anyhow. Absolutely pointless to even try.

ARM devices need a slim 'n trim (re: non-fatty American) OS to be of use to anyone but the priestly patient and dead.

MS in vesting heavily on new microkernel architecture (project Singularity) which will be the next Windows Mobile and when evolved enough enw Windows platform. No point in investing on old technology when you have something better in the pipe.

Or Arm isn't powerful enough to run 7. Because it's not. It's not even powerful enough to run anything for a user. Arm is for embedded systems, not PC's.

This is a smart move. All modern OSes can be easily ported to any architecture, and Windows is no exception.

However, Microsoft does not want to deal with all the problems associated with running architecture-specific software on a different architecture--let alone an embedded one--that would need to emulate another architecture to even run the majority of software out there.

The obvious choice is to simply not bother with that can of worms to begin with. Can you imagine all the complaints from people that their ARM-processor based system perform poorly, or do not run their favorite app? What would even be the point?

Leave ARM to OSes which easily allow for their software to be recompiled for that architecture. Desktop Windows is not one of those OSes; that's what Windows CE is for.

WHERE ARE YOU TO MAKE SOME SENSE IN ALL OF THIS???

Recently the Samba project effectively documented SMB - it seems MS didn't know how it works and had to wait for someone else to tell them so they could add a few bits in Vista to stop Samba working properly!
No Arm - of course not - that would require knowledge of the inner workings of Windows - something its becoming more and more clear that MS have lost somewhere. It seems they bought the lie they sell about computing being easy themselves. Hoisted by their own petard!

Windows is not a compilable OS like Linux.
Even Intel64bit was hard work for Windows.
Likewise Aix is PowerPC only.
Hard luck for Microsoft, in a way, but that is how DOS/Windows was built from the start, and who knows, perhaps it gave them the head start they live on today.
PS. try to find a processor architecture where Linux does not run.

c:\win7 make arm

Windows 7 could be compiled for ARM. And it is incorrect to say that zero windows apps will run on it if it were.

.NET apps essentially compile at runtime based on the processor architecture that it finds.

The main hurdle i see is intel. Can you honestly see Intel allowing anyone to disrupt the atom train?

We will see windows 7 on arm only if android etc gains traction in the netbook market.

<windows <linux s because Linux is an open source piece of software and windows is not!?

Windows is not a compilable OS like Linux.

What is that supposed to mean???
And how do you think does windows manage to run on a CPU if it is not "compilable"?
How the hell do you turn windows source code into machine code without compiling it???
Every piece of software must be compiled from source code into machine code for a CPU to run it. Unless of course this piece of software is written in an interpreted language like Python. But we don't write an OS in an interpreted language, do we? :)

Even Intel64bit was hard work for Windows.

It was not. Use your grey matter.

Likewise Aix is PowerPC only.

Aix is power chips only is because IBM needs to sell power chips. And not because it can not run on Intel.
For exact the same reason OS X is legally (not technologically) tied up to apple hardware.

Hard luck for Microsoft, in a way, but that is how DOS/Windows was built from the start, and who knows, perhaps it gave them the head start they live on today.

Rubbish !!! How old are you?

PS. try to find a processor architecture where Linux does not run.

Maybe that's because Linux is an open source piece of software and windows is

I am surprised noone drags up the fact that WinNT 3.1 was written for Intel i860 and MIPS R4000 to avoid being trapped into x86 thinking. Releases were made for x86, Alpha and MIPS and later came PowerPC and Intergraph ported to Clipper (and maybe SPARC internally?). Today, Windows NT runs on Itanium. No, there should be no serious problem compiling Windows NT on a RISC chip.

As others pointed out, the applications and the UI are the problem. If you would want to run something like WinNT 3.5 instead of WinNT 6.x that'd run like greased lightning...

Karoly's comment is spot on, but I doubt that given the huge changes since 3.x,4.0 it would be easy to build for ARM. It is a huge codebase, and perhaps has serious bitrot given that they haven't kept honest and portable...

Do they even *have* a clean source backup of the old versions any more?

(Maybe if they have an old enough codebase as a base set they should open source it community style a la OpenOffice or Java).

@Andy Allen... Yes they have a clean code base, it is compiled everyday for both the Itanium and AMD64 and EMT64 platforms in addition to older X86 architecture. (Do you not realize that these are different architectures?)

Yes Windows NT can (Win7) could be easily ported to any architecture, as it is the basic design of the NT architecture with its HAL and multi-layered Kernel API set. In theory NT is easier to port to various hardware platforms than even Linux or other Kernel technologies.

(Windows NT 4 was available for PowerPC, Alpha, RISC, as well as the x86 architecture. Win2k would have had a full Alpha version if Compaq didn't pull the plug on the Alpha after aquiring DEC.)

The problem with porting Windows to non x86 platforms is 3rd party applications, plain and simple.

Joe Smoo buying an ARM based Win7 Netbook will not 'get' why Adobe Illustrator won't run on it, as its is compiled for the x86 or EMT/AMD64 architectures. They would 'choose' Win7 to run software they run on their desktop, and this won't happen without virtualization.

So if you are looking at NO application support, you have a couple of choices.

The first one would be a virtualization environment for 3rd party x86 applications, and in doing that, you are mitigating any 'performance' or 'battery' gains of the alternative architecture. So that would be pretty stupid, and worthless to the architecture.

Even a translation virtualization like DEC and Microsoft used on the Dec Alpha versions of NT that would translate the binaries to a more native format each time they were run would still face performance issues when you have MILLIONS of 3rd party software designed for the Windows Platform.

The second option is to leave the architecture as a 'blank slate', and if you are going to do this, Windows CE or Windows Embedded CE or a Win7 Embedded variant with a custom UI would be a better solution, as the users would expect only .NET and native CE compiled ARM applications from running properly.

Of course this brings us to the Linux is better argument, as there is source for most 3rd party applications.

Well, this is not entirely true, and secondly it also illustrates how 'few' applications are widespread or commericial on the Linux platform.

A few thousand or even a hundred thousand Linux applications recompiled doesn't even dent the millions of Windows applications available by 3rd parties.

If people really think MS made this decision to help Intel, they forget, MS tends to hate Intel about as much as any company, as MS's AMD support has been spot on and MS even moved to a fork of Windows (after 2003 Server) running on a tri-core PowerPC for their XBox 360 - abandoning Intel all together.

(Intel testified against MS in the monopoly case, and was less than truthful in their testimony, trying to get in out the fallout of a broken MS.)

I can see MS expanding to ARM and other architectures more than they do now, as the need for a lower end Windows CE version is no longer needed as hardware catches up.

It will also be when virtualization of x86 3rd party software can be done easily, either via the architecture or a MS NT Subsystem/VM variant.

However, that won't happen until the Win7 Embedded gets done and MS looks to Win8 to flip over to more .NET and fully VM the Win32 subsystem.

So put down the NT can't do that rants, and put down the MS loves Intel crap, both are insane...

Of course you have to compile even assembler. The thing is that Linux is basically C, what we call a "high level language" (even if it makes me laugh to some extent).
So for Linux you basically need a C compiler for the architecture you need to compile to.
With windows it is completely different, it is not written with a high level language.
So yes, it is a hard effort to move Windows to some other arhitecture.
Pleas read more, there is a lot about it, even at Mirosoft, I am sure.

My age, well I was a programmer for 35 years, and I started before Micro-Soft was born. Still active but rather for fun.

I know this will not convince you but if it was easy to compile Windows for other arhitectures, do you not think it would have happened and been a part of Windows for the last 25 years.

Also remember that when Intel built the "rather unhappy" Itanium Linux was used to test it. (not Windows).
Can you run Windows on a Itanium.
Also the Chinese used Linux to develop their processor.

PS. do not kill your self if there is, perhaps, something missing in Windows.

Yes you can run Windows on Itanium.

And there have been versions for a number of different processor platforms over the years.

Cutler and the boys from DEC made sure the thing was easily portable, so if MS really wanted to port min-win to ARM, they could.

However, as others have already pointed out, porting the O/S core is not the problem... it's the 3rd party apps and the various add-ons for the O/S that would be the issue.

What's the point of porting to ARM and then have no applications for people to use on it?

I would guess that MS view the ARM as a target for WinCE because it fits better with their view of cloud computing for the ultra-low powered market.

The only reason to run Windows is to play commercial 3D games but most of them need a much more power hungry 3D chip than any of the small machines have. Even W7's GUI is designed assuming hardware 3D & a big LCD. So from a utility standpoint, it's pointless.

Under the hood, the first version of NT was very clean thanks to Dave Cutler (of DEC's VMS fame) leading the project. But as soon as all the MS coders raised on nothing but x86 started working on it, they started filling it up with Intel-isms again. Notice that even the Intel Itanium versions of Windows don't have all the features of the x86 branch. And of course, they require new compiles of all the apps too. So don't think that W7 on an ARM would magically be able to run Excel, let alone Left for Dead 2.

Free Software has the advantage that it's designed to be recompiled so you can make all kinds of changes to the hardware and still use existing software. In fact, closed source has been a major frustration for Intel as well as other chip designers because no matter how great your innovation, if MS won't/can't support it, it's pointless to deploy. Hopefully, we're coming to the end of such times and we'll be able to set hardware designers free to make real breakthroughs.

Do some people remember the Japanese version of Windows 3.1 for the NEC PC-98?
It was for a Japanese-only hardware platform and required separate version of the OS along with all applications!
There was endless confusion of the end-user, who had to make sure they bought the correct version of Windows, and Office, and anti-virus, etc...
Microsoft probably learned their lesson at that time.