It would be cheaper and faster for them to buy up the rights to something critical for ARMbooks, then sue the crap out of anyone trying for a work-around. What exactly that "something" is would be the hard part.

>It would be cheaper and faster for them to buy up the rights to something critical for ARMbooks, then sue the crap out of anyone trying for a work-around. What exactly that "something" is would be the hard part.

Exactly so. Windows 7 on ARM is kind-of possible, but nowhere near practical. It would have no applications and a paucity of drivers and utilities.

Eg. Microsoft has a cross-compiler for ARM (Windows CE runs on ARM, for example) ... but does Adobe? Do hardware OEMs (who write Windows drivers) have a cross-compiler?

So, what would Microsoft sue over? What would they be able to claim to license? Remember ... it has to be something that Microsoft owns. Hmmmmm. I seem to be coming up awfully short here ... Linux owns cross-platform, not Microsoft. Linux on ARM outnumbers Microsoft software on ARM by thousands to one or more.

Elephant in the room ... Microsoft does not own all of the source code on a typical Windows desktop.

Drivers ... written by OEMs. Codecs. Adobe photoshop. All sorts of applications and utilities. FOSS software such as OpenOffice, VLC and Firefox.

How are Microsoft going to get all these people to co-operate and port their stuff to Windows on ARM? To what end for them?

If Microsoft produce Windows for ARM, it will have an OS without any applications.

The drivers should not be a problem, either MS owns them or holds an exclusive license.
The toolchain should not be a problem, they have the Posix overlay
(from back when they tried to out-mingw, mingw) - that allows them to run any *nix toolchain.

And - unmentioned yet in these comments - embedded XP - -
If I was in the decision making chain - that is what I would pick to dust-off and port,
not do a refresh of CE.
Of course, they might want to change the name (XP) - so that they can announce, again,
the death of XP. ;)

>The drivers should not be a problem, either MS owns them or holds an exclusive license.

No. The larger bulk of hardware drivers for Windows are written and owned by the hardware OEMs. Printer drivers is a classic example ... written by the printer manufacturers, they come on a CD with the printer. They are not part of Windows per se. Drivers for a new architecture are a **HUGE** problem for Windows. The fact that Microsoft does not own significant parts of the drivers codebase is illustrated by the driver problems that Vista had, and still has for hardware that is no longer in production. It is also illustrated well by the fact that 64-bit Windows (which also requires new drivers) is quite scarce still.

>The toolchain should not be a problem

True. Microsoft does seem to have a cross-compiler for ARM.

But you missed the elephant in the room. Again this has happened.

Adobe Photoshop ... not owned by Microsoft. Adobe has the code.

Likewise Flash players. Likewise for PDF viewers. Likewise for Dreamwaever. Adobe has the code for these too.

In a similar fashion ... there is a long, long list of applications ffor Windows that are not part of Windows itself. Third-party applications. Microsoft does not have the code.

So ... Microsoft ports Windows to ARM. Fine. It compiles and runs ... but it runs Notepad, Calc, Wordpad and Paint.

Microsoft won't embed its own actual applications such as Office ... Microsoft wants to charge you heaps extra for that. Adobe won't port its applications to Windows for ARM, nor will any other software vendor until there is a decent reason to do so.

How will Microsoft get such a market on a new platform? Existing x86 applications won't run. A good percentage of hardware won't have drivers. Application software vendors are going to have to do a porting effort, similar to Microsoft's own port of Windows itself. Users are going to have to re-buy these new ported copies.

Meanwhile, right there on the new ARM netbook machine as it is sold, there is a full copy of Ubuntu desktop with all applications already installed for you ... at zero cost.

i dont think that win 7 on a arm processor is a good idea, linux ok but windows? no way i think. windows ce ok but i dont think that ms will do this. we wont see so much arm netbooks that it is profitabel for ms.

Folks, take a look forward and we could see something emerging here.

Most of you are looking at what it would take MS to get an ARM port working and regardless of how easy/difficult it would be technically this still needs resources internally and reasonably specialised ones When stating the "only" thing needed is driver support underplays the impact of that - remember Windows Vista!!

But back to the trend I mentioned.

Netbooks as we know had a significant impact on MS, before they got into the game there was a meteoric rise of netbooks and almost all of them not running Windows !! MS reacted and the tables have turned in terms of market share.

Now look forward ...
Netbook prices are rising and whilst they wil still be very competitive against the traditional alternatives they could be the target for ARM powered netbooks as ATOM powered netbooks were for traditional Laptops.

Basically why couldn't the surge happen again for ARM powered Netbooks?
Then MS would HAVE TO respond. Now unlike the easier technical challenge of the ATOM chip it certainly won't be an overnight fix for ARM chips.

So for those that support MS lets hope they did keep backwards compatibility because lets face it if we have to rely on CE or NT the comparisons between Linux and Windows will produce a hilarous raft of laughter and glee from the open source folks.

Oh and market share could be impacted permanently

Not that this is a bad thing but worth thinking about...

If they're not already working on it, I can't imagine them getting one ported and tested in the 3 year (now) product cycle. The NT 3.51-era elements, such as Kernel/User/GDI and applications, are all set up to be converted to RISC, so it's mainly a matter of rewriting the kernel to the ARM instruction set and recompiling the C++ code. But there's still a ton of other kernel-level code - mainly the new driver model - that would probably take quite an effort to port.
But then, Apple maintained a port of MacOS on Intel for years in secret while they ran on Motorola, so it's not inconceivable.

Exactly right. The current Windows codebase is known to build in x86, x86_64 and at least parts on Itanium. Most of the low level stuff built at one point on MIPS, Alpha and PPC. So if they have been smart and maintained the ability to portability when writing new code they could do a port to a couple of specific target machines in months. If they have let the portibility go to heck or wrote all of the Vista stuff assuming x86 then they would be pretty much boned.

No, the problem is the same one that accounts for PPC, MIPS, Alpha, Itanium, being essentially dead and x86_64 only becoming a viable port as the average machine is shipping with more RAM than the x86 port can address. Nobody ported any applications to the non x86 targets. Not even Microsoft's own apps division, hence DEC's FX!86 emulation layer to let NT/Aplha run Office... at speeds worse than a cheap Pentium of the day and everyone got the impression Alpha was crap and it died.

But one thing every other port had was the advantage of being more powerful than x86. ARM isn't, most implementations have severe RAM limits, etc. Unless Microsoft has a cross compiler chain (and I have seen no sign of one) even finding good developer workstations is going to be a challenge. And the end target is going to be little netbooks which will probably top out at $400 max. How many 3rd party applications houses are going to get excited about expending a lot of effort to capture customers they will see as 'cheap bastards?'

Any bets on whether we see Adobe port anything other than Flash and perhaps Reader? Yea like people will pony up $999 for Creative Suite on a low spec machine costing a third of that? Make the round of the key software people would be wanting Windows to be able to run and count how many would be likely to port. Unless Balmer threw a chair at em I doubt Microsoft's own apps division would bother.

And that leaves .NET as their potential saving move. Except they don't write much with it and few 3rd party shops are releasing 100% .NET products.

This is imminently doable. Windows NT, the grandfather of every version of Windows since 2000 (Me excepted) was fully supported on the MIPS, Alpha and PowerPC processor architectures. Microsoft knows how to do this and how to make it work, the problem is getting good ARM-version Windows drivers for the rest of your small PC hardware and getting a good number of Windows apps recompiled for ARM (although .NET-based applications should run just fine).

Short answer: No.

Long answer: Nooooooo.

It would be a major undertaking, the expense of which would never be recovered - the market for low power devices might be large, but the margins are too tight for Microsoft to make the kind of money they would need to make for the venture to be worthwhile.

We might see a new version of Windows CE/Mobile updated to look and feel more like Windows 7, but that's about it.

another trick would be the move the CE platform into closer compatiblity to x86, so that a recompile is what it takes to port something over.

thats basically what it takes to port a linux app over, so why should it not be possible with a windows app?

This one is so thin you can see right through it. Even if Microsoft could port 7 to ARM, it would take them 5 - 10 YEARS to do so. There are so many millions of lines of code that Microsoft doesn't even know what it all does anymore (this is not an exaggeration). Even if they could, ARM is not as powerful as the current processors used in netbooks. I expect they will be in that same 5 - 10 years, but for now it would be too slow. I am sure Microsoft would like to have the capability, but I doubt we will see it happen.

Why would they need to REWRITE all those lines? They just compile the C/C++ codebase with ARM compiler.

and here i was under the impression that cortex can go head to head with atom...