Microsoft is set to unveil a new incarnation of Windows that runs on ARM chips, according to a report citing two people familiar with the company's plans. Bloomberg reports that Redmond will announce an ARM-friendly version of Windows early next month at the Consumer Electronics Show (CES) in Las Vegas. According to Bloomberg's …
Well, OK then.
This is a pretty huge facestab for intel. That MS are willing to ditch their usual dance partner like this would indicate just how desperate they must be.
It wasn't all that long ago that MS bullied Asus into canning their ARM based smartbook after all.
How times change!
However, MS already has "Windows" for ARM, it is called Windows Phone 7 née Windows Mobile.
I'm surprised it's taken so long, historically (and even today) windows has been available on many architectures, not just Intel. Also proper windows on tablets is a much better idea than win phone os or other embedded windows.
Not the first time, that relationship has been a little less than close for some years now.
There have been quite a few cold shouder moments in the recent past, but I reckon the real kick in the nuts was the Itanium / EPIC vs. AMD64 debacle. Looking at what Intel did at the time, you have to bet that they were banking on MS going with EPIC as the future platform of choice. It must have been a serious "et tu Brute?" moment when the AMD64 versions of Windows shipped.
Available does not mean usable
The only non-x86 architecture on which windows proper was reasonably stable was Alpha. There it worked. That was not the case with other platforms.I had the chance of playing with Windows on Power at some point in the 90-es. Average uptime - 2h. This probably says it all.
RE: Available does not mean usable
"I had the chance of playing with Windows on Power at some point in the 90-es. Average uptime - 2h. This probably says it all."
Sounds pretty good for Windows!
... so that's how it works. Business decisions are all acts of desperation. I never was quite clear on that; thanks for the revelation.
Talking of Windows on esoteric architectures - NT on Alpha was rock stable. In an intranet environment you could have half a year of uptime with ease even on a heavily loaded system. Compared to that getting a day of uptime from a loaded Power or MIPS running windows NT was like winning the lottery. It got dropped at NT4 for a reason and the reason was that it did not work.
Well I don't think you should blame microsoft for the itanium debacle. Have you ever considered buying an itanium? Ever looked at the price or the die size of a chip?
AMD64 allowed to cheaply cross the 2GB memory limit of 32bit windows that and marketing was the demise of the itanium.
Same is true about windows on alpha, awesome platform, used it for development for a couple of years and it really was rocksolid and outperformed the intels we had at the time. But our customers all used intel and then amd came along and the rest as they say is history.
Oh, where to begin.
MS aren't ditching Intel. No one's going to have to throw away their Intel PC and buy a new ARM-based PC to run Windows 8 or any other existing branch of Intel Windows.
Windows Phone 7 has nothing to do with Windows for PCs and laptops, other than a brand name.
If anything, this is good news for ARM as it can only aid their expansion into server hardware. If MS get Win32/64 running on ARM then it makes the possibility of a version of Windows Server one day running on ARM one step nearer.
Also, it proves that WinCE and everything that was borne out of it over the last 15 years was crap. The iPad is exactly the kind of device that MS's WinCE was intended for, yet there's absolutely nothing running it (under whatever modern name they'd care to use) on a rival device today. Not a consumer product, anyway. Touchpads on van dashboards or used by meter readers and delivery drivers don't count.
He didn't say MS are ditching Intel, just that they are no longer wedded. This really erodes the value of the Intel Atom line. Atom's selling point was that it was the lowest power platform that can run Windows. Now ARMs will knock it off that perch.
The only reason OLPC etc used x86 was to be able to run Windows. ARM would have been far cheaper, lower power etc etc.
It does not prove WinCE is crap. WinCE runs on memory/CPU footprints that Real Windows can't and never will. Real Windows won't be coming to cell phones etc any time soon.
WinCE was never intended to be used on ipad-like devices, just PDAs, phones etc. WinCE process models just could not support large media apps we associate with ipads etc. MS have been dabbling, and failing, with tablets since the 1990s. They've always used Windows.
"The only reason OLPC etc used x86 was to be able to run Windows"
Wintel dirty tricks are what got Windows in OLPC.
>WinCE was never intended to be used on ipad-like devices, just PDAs, phones etc.
So when is a PDA not a tablet? When you remove the keyboard? I've had WinCE based PDAs (not through choice) and they were universally crap. With or without a keyboard, they've been junk. I don't see how screen size (which afterall is the only difference between a PDA and a tablet) would make that piece of junk any better or worse.
WinCE really is a terrible operating system. It's nasty to develop for. Ever tried putting it on your own hardware? And it gets so many things so very badly wrong. Even simple things like Time-of-Day it can't get right, or even consistent, especially when considering alarms in other daylight savings zones. Admittedly other OSes have had some problems, but at least in their case it's bugs not the actual design that causes the problems.
If MS put half the effort they applied to Win7 into WinCE it would be much better. The new Win Phone7 does look nicer (very nice actually, I think), but it's still WinCE under the hood with all it's terrible, terrible nasties. They've a lot to do before WinCE becomes anything like halfway decent. Perhaps even MS have finally decided that WinCE is inevitably junk and that a port of Win7 is the only sensible thing to do?
>Real Windows won't be coming to cell phones etc any time soon.
What do you mean by 'real windows'? Sure, you're not going to get a full Win7 desktop on a phone, the screen is too small. But if they took the guts of Win7, stripped them down to only that which is necessary to support a phone GUI (i.e. kernel. some essential services, etc) then I think that MS would have the foundations of a very good and varied offering. Especially if they can point to commonality across mobile, desktop and server running on intel and arm.
Other commentators have said that this can only be good news for ARM. It certainly is! Who really needs the hulking great performance of an intel chip on their desktop? Gamers do, maybe a few scientists, but that's about it. It may be that everyone else is quite content with a quad core 1GHz ARM based desktop running Windows 7 consuming 5Watts. Server operators will likely be quite content with a low power CPU coupled to the same IO hardware running their software stack - likely to save them an absolute bundle in electricity bills.
It is indeed bad news for Intel. Atom hasn't taken off anything like as they'd hoped. They're still many Watts short of breaking in to the phone market. It's especially embarrassing for them considering that they voluntarily **SOLD** their ARM development line to Marvell at a low-ish price, who must surely be rubbing their hands in glee. Intel had the opportunity to define and lead a new PC standard based on ARM, but didn't. Sounds like MS are going to do it for them.
There are many people around who don't 'care about the evidence (millions of sales) but have the attitude that everything Microsoft is crap just because it is, and you will NEVER change their minds.
These same people are either advocates of Apple (who make pretty but normally quite restricted and useless products with an amazingly poor developer experience) or some for of Linux (I tried one variant of this a year back, after 2 days of pain it would work in text mode only but refused point blank to cope with the perfectly good graphics monitor).
The reason most people have windows at home and work is that however bad it is they recognise it is still better than the competition.
But can MS make a UI that doesn't require pixel-perfect precision? Battery friendly and finger-based for me...
At this rate, Microsoft might be a competitior in this "Mobile Internet Device" arena by Christmas 2012! Nice to see Microsoft making descisions in a not-PC-space market that might actually make it something other than a terrible also-ran.
Really, how hard can it be?
M$ actually were the first major OS vendor to develop an alternative UI with windows XP media center - which actually works fine on a tablet, since it is designed to work with remotes and a simple up-down-left-right-OK logic (kind of like a DVD player).
As far as developing a tablet UI goes, some brainpower is needed, but honestly its not rocket science.
The iOS interface is no great shakes anyway, so the bar's set pretty low. It's functional by not doing much, which is not that hard to match. It actually reminds me of windows 3.1.
iOS depends on apps to do most of the tasks, which means that while it will require M$ to drive the porting of some of the windows ecosystem to tablet, the level of functionality required to achieve parity is limited. On top of which the dev's behind the 250K iOS apps have already worked out pretty much all the possible permutations a fart app's UI can take. How hard is it to copy the best one?
Yes, of course they can...
Mind you, based on previous evidence it might not be 'Windows' as we know it...
Cut down? Incompatible with any of the current range of software? I'd say those are not unlikely. However, it might provide a cheap source of tablets that can be re-flashed with whatever flavour of Linux takes your fancy...
Who says usability means everything has to be all bells and whistles with layers of complexity?
Just because Microsoft try too hard with their interfaces and provide millions of features so their sales and marketing guys have something to talk about doesn't mean that is what a good computer should be.
Then there's the millions of lines of code metric, as if that means it is any good? more likely it is bloatware.
iOS depends on apps to do everything? what are you talking about?
You do realise an operating system is basically a kernel, file system and libraries to operate hardware and develop applications. Things like web browsers, email applications and media players are applications, they are not part of the OS. Microsoft and others just bundle all that stuff for their own reasons.
Just because something is pre-installed it doesn't mean the OS won't work without it.
Makes perfect sense for MS
But Intel are really in the last chance saloon... oh well maybe there's karma after all and all these years of screwing up AMD with the protection racket are coming back to haunt Intel
Re: in the last chance saloon
Sorry? There isn't a CPU vendor anywhere who wouldn't swap with Intel right now. That's not "last chance saloon". They have 80% of the desktop market, the best process technology on the planet, and the only reason they don't make low energy processors is because the desktop market can't use them.
There's no point in making the CPU use significantly less power, for less performance, than the most power hungry device in the box. For a laptop, that's the screen, by some margin, and by the time you've put a battery big enough to drive the screen, you have spare power to drive a faster processor.
ARM traditional low power strengths make sense for headless devices like NAS or routers, and that is where we've seen them. Times are changing and MIPS-per-watt is starting to matter in servers, so we are starting to see ARM (and Microsoft) giving serious thought to putting low power chips there.
Finally, instruction decode is a single-figure percentage of a modern CPU's complexity and power budget, so the actual instruction set doesn't matter. It hasn't mattered since some time in the last century. You could build Power, x64 or ARM chips for any power envelope you choose. It's just that historically they've cornered different parts of the market and built barricades of closed source software.
So long since the demise of NTe for MIPS and PPC we forget a tiny XP embedded lives until 2016. Yet now, in times of celebration and merriment, word comes a mere decade into a new millennium of a great and new Windows Embedded Standard may yet be born into loving ARMs. A virgin birth it shall not be as we have long been screwed by the progenitor of this hoped savior of MS. So come let us rejoice and consume great quantities of eggnog without fear that I spoiled perfectly good hooch with milk or eggs. Ho Ho Ho my kingdom for a ho!
Graham, is that you?
Stay away from the booze, man!
Missing the point
>word comes a mere decade into a new millennium of a great and new
>Windows Embedded Standard may yet be born into loving ARMs
You're missing the point. Mobile devices have stopped being 'embedded' and are rapidly heading towards being general purpose computing platforms, just like a PC or MAC are today. Indeed it is only the artificial barriers raised by Google and especially St. Jobs that prevents them being almost general purpose platforms already.
MS were caught out by that - hence WinCE's continued existence - but are showing signs of having realised that and would seem to be addressing it. If they can move Win7 to ARM then I think they will have a very strong offering on mobiles, desktops and servers. What would you buy - a Win7 ARM device that lets you run anything you want, even Flash, or stick with an IOS device from Apple that will run only things that Steve Jobs approves of, and only then (most of the time) one at a time?
Would you buy a 200Watt+ server based on an intel chip, or a 20Watt server that does the same job but runs an ARM?
Unless you're playing games at home, do you really need the horsepower of an i5 just for a bit of browsing, emailing and video watching? Or would something smaller and cheaper be acceptable?
Not that ARMs are particularly slow these days - Marvell, Qualcomm and TI devices in particular are looking pretty pokey - quad cores on the way dual cores already, multiple GHz already, power levels that Atom can only dream of.
eeehm, flash on Windows/ARM???
> What would you buy - a Win7 ARM device that lets you run anything you want, even Flash,or
> stick with an IOS device from Apple that will run only things that Steve Jobs approves of, and
> only then (most of the time) one at a time?
I dunno but why do you think flash will run on Windows/ARM? Right now it is an i386-binary, not an ARM. Porting Windows to ARM is nice and doable, but porting ALL known windows applications to Windows/ARM is another adventure.
No known applications (windows/i386 that is) will run on Windows/ARM without compiling. Good luck either without the source (Luke!!) or depended on ... Microsoft in the same way you are depended on Apple.
not that big a leap
Errm, ever thought that Adobe may just re-compile it? Besides, Flash is already on ARM under Android. I don't think that Adobe would have much trouble doing the port beyond the problems they create for themselves anyway.
So if the Win ARM market takes off I don't see why app developers wouldn't be able to just re-compile and make it available. That's pretty much what has happened during everyone else's platform migration phase. There is a dependency on the app developer, and not really much of one on MS (provided they get the libs and dev tools right). Having the source would mean that anyone could do the port, but then that'd likely be beyond a high percentage of Windows users anyway. HP had to work very hard to get developers to bother to recompile their HPUX apps for Itanium, but that was largely because no-one could raise any enthusiasm whatsoever for the soon-to-be-sunk and completely pointless dogs dinner of a chip that is Itanic. In fact I can't remember the last time The Reg did an article about Itanic!
I remember reading years and years ago why ARM went little endian, and the answer did have an MS angle to it. That might yet prove to be a wise choice, at least from a commercial point of view.
Judging by your chosen logo I'm guessing that your a Linux user. I use both. Linux of course is already available for ARM with all the software pretty much available and ready to roll. The LINARM pitch is clearly a massive threat to WINTEL in the server market, just because of the large power savings that are likely to be made. I think that more than anything has prompted MS to look afresh at ARM. It smacks of MS deciding that the WINTEL arrangement is suddenly no longer a sure fire winner. With WINARM Microsoft's commercial position may survive. I'm pretty sure that Intel are in trouble.
I find it very hard to imagine full fat Windows 7 on a mobile device. I predict a Windows Phone 7/ Zune OS tweaked for larger screens, which could actually be rather good.
is that the sound...
of the 'me to' band wagon... cannot wait to see a slow, insecure tablet for sale. How much ARM power will a virus scanner use i wonder.
So, a punter buys a windows tablet...
... or net book and he notices that his colleague has one the same that doesn't seem to need charging up every five minutes and starts up twice as quickly. His colleague says 'oh yes well I ditched windows and put [insert your favourite OS here] on it and it's much better. Exit one Windows licence. Next year, when tablet2 comes out and is the next 'must have' our hero doesn't even bother looking at the Windows version.
Microsoft knows that exposing users to the competition is the beginning of the end for their easy profits. That's why they where so down on the netbook manufacturers. Maybe there isn't much drawing an average laptop/desktop user to ditch Windows. After all, the hardware has been designed to fit the Software. Intel processors and current GPU's all have some kind of Windows/ActiveX optimisation. Huge batteries and massive memory/storage is considered essential.
The RISC option for NT did not sell well received because, well, if you are the type of customer to go for the alternative hardware you know about the alternative hardware. The big draw for Windows servers where that they where cheaper and/or looked familiar not that they where better.
If they do actually produce an ARM version it will surely invite users to compare the cheaper and less resource hungry alternatives. If it is true then it is a position they have been forced into. Perhaps the netbook manufactures who stayed in the Microsoft cabal are now asking why they should stick with Microsoft when the market is getting away from them. I can't see this going well for Microsoft
>Next year, when tablet2 comes out and is the next 'must have' our hero
>doesn't even bother looking at the Windows version.
Except that if we let M$ in they will stitch up the market so you won't be able to buy without paying an M$ tax
So long Intel...
It's very obvious that the dominance of Intel in the PC world is purely down to Windows.
With Windows moving sideways into ARM, I believe we are just about beginning to see the "beginning of the end" of Intel's dominance.
IMHO, despite Intel doing a frankly incredible job with the x86 hardware over the years, the x86 family is constrained by its own legacy. The "need" to be able to run code written in 1977 on a modern chip has led to inevitable difficulties for Intel. They've done an incredible job eeking out the performance that they have.
However, ARM have no such legacy restrictions. We might also see a return to more conventional architectures. Hands up who remembers the crack smoking memory access hoops we had to jump through in the early 90's with the early 286's etc. OK, it's not like that any more, but the point is, Intels chips still have that basic legacy at their 'core' (forgive the pun) that isn't there in ARM technology.
It was only a matter of time until Windows went on ARM. I guess early versions will be restricted to certain hardware/chipset combinations only, but once the 3rd party market gets a hold, you'll see 8 hour battery life Windows laptops.
One wonders how (or even if) MS will get things like Office running on ARM - once the OS is up and running, fine, but what about the apps?
Windows Server on ARM would be nice, but what about exchange, SQL Server et al? I should imagine it would not be a trivial 're-compile' to get those apps running. Perhaps they will rely on some sort of virtualisation? An ARM emulating an Intel?
Interesting times. And a decent spike in ARM share price already on this news.
it ought to be
>I should imagine it would not be a trivial 're-compile' to get those apps running.
It ought to be. As they work out why its not, they will find out why their products are not maintainable.
At last... sort of
MS have not ported Windows to the ARM before because the couldn't - they don't know enough about their own code to get it to work on ARMs - hence the new incarnation. It wont be Windows as we know it but hopefully it will no longer actively discourage the manufacture of high quality netbook/netpad machines which will become available without the windows tax (so £150 rather than £200+) for handheld computing that has been a real possibility for 5 years now, and a 'where is it?' for the last 3.
We can now have machines that can run Windows - but it will become quickly apparent they'll be better off without it.
Where the foot-gun icon?
Windows tablets aren't failing because they have x86 processors. They're failing because they have Windows. The bloated, decrepit, diseased platform runs a billion forms of malware and 15 years of legacy apps that have user interfaces completely inappropriate for tablets.
Windows has been engineered in turns to be incompatible with Stacker, OS/2, Lotus 123, Wordstar, WordPerfect, Netscape Navigator and a host of different platforms and applications. If you had any idea what kind of ridiculous twists you have use to pull to that off you would understand why it's the easiest platform to write malware for. Legend has it most of the stuff in there is magic code noone dares touch for fear it will break the whole stack - without an overriding concern like a published exploit.
The company's legacy of execution on major changes is laughable. By itself this means that nobody should be getting excited about this until somebody else tries it first. More likely than an evolution of desktop Windows to ARM is that they're going to re-launch Windows CE as a tablet OS. That should be wonderful - to watch from the sidelines, with popcorn.
But yeah, if they want to survive the next decade this is what they have to do. It's possible not all of them are stupid. I was beginning to wonder.
"Windows tablets aren't failing because they have x86 processors. They're failing because they have Windows. "
But Microsoft already has a version of windows for ARM chips, it is called Windows Phone 7, so what does this mean? Some tweaks on the interface to make it work better on tablets?
The most interesting part is that it would represent a complete about face for Microsoft, gone is the pretence that 'real' windows is need for tablets, didn't Steve Ballmer say that customers had told them,time and time again, that a tablet should support all windows apps?
And if they are going to let OEMs use WP7 in tablets, will they port it to Atom for those tablets? Will they have to reinvent universal binaries? Would they work given the limited storage in those devices?
Windows on ARM
Hardly the NT family, that Win7, Vista, XP and Windows 2000 is based on. It's a long time since Alpha, MIPS and Power PC supported. ARM never was. It's now X86-X64.
WinCE, also the basis of Windows Mobile, Windows 7 Phone and Zune pretty much only runs on ARM nowadays (though it did run on other CPUs).
There are three families of Windows Embedded:
WinCE based (ARM)
Defunct DOS + Win 3.x based (x86 only)
NT Based (NT4.0 version was last to support MIPS and never supported ARM. Now only x86).
So Which version of Windows is this ARM Tablet going to run?
Windows currently support 32 and 64 bit Intel/AMD and also Itanium, it's still portable.
Server makes more sense
It doesn't make sense for MS to release another tablet/handheld version of Windows as they've already got Windows Phone 7 (after all, Apple moved iOS from phone to tablet without too many problems). Also, an ARM version of a desktop OS like Windows 7 would require all current Wintel applications to be recompiled for a "WinARM" platform (assuming there won't be enough grunt to translate x86 on the fly).
Surely the best option for MS is to produce a server only version of Windows for ARM. This would only require recompilation of a (relatively) small number of enterprise applications, quite a few of which are MS products anyway. This would give enterprises the option of selecting a Microsoft OS for the ARM powered server farms which everyone is getting so excited about. It would also allow the desktop version of Windows to remain on the client workstations.
There is currently no server market for ARM except with a few geeks at home (myself included) and let's face it, they'll almost certainly be using Linux. if big businesses want to use their hardware more efficiently you put in a virtualisation solution, such as VMware or Xen et al. The tablet and netbook market as it is today would dwarf the ARM server market. That's not to say that this won't change, but it's the state of play for the foreseeable future.
Missing the point
But surely the biggest reason people want a Windows tablet is to run their desktop apps. If you have to get a recompiled version of everything you might as well use WP7 as the base. This puts in the same boat as Android/iOS et al surely. The only advantage is that managers will see Windows and hit the safe buy option.
People won't want to run desktop apps on a platform that doesn't do pre-emptive multi-tasking
Windows 7 on ARM
Most of the posts here seem to be assuming that MS are going to port over Windows 7 (as it seen on the desktop pcs) so you will suddenly have access to the all the same programs that will run on your desktop pc on a tablet. Well that isn't going to be the case as even if MS do port Windows 7 to arm all the current programs you can run on Windows 7 desktop PC are compiled for the X86 processors so aren't suddenly going to run on an ARM chips. Sure MS could include a X86 emulator in the ARM version of Windows 7 but even on a 1ghz ARM chip this isn't going to give anything like a speedy experience and emulators are very CPU intensive so would soon drain your battery after a short time running an application under x86 emulation. Also desktop PC programs generally expect large amounts of hard drive space and memory to run which to add these to a tablet increases its cost compared to Andriod/apple devices which don't need a gig of ram and 5 gig of hard drive space to install one application.
Native Windows 7 ARM complied programs would take some time to come along thats if they ever do as its the same chicken and egg situation of no developer wants to spend time recompiling all their programs and testing them on Windows 7 ARM if there is hardly any market for it while no one will buy a Windows 7 ARM tablet if there is no software for it.
its much more likely that if MS are considering a consumer launch of Windows on ARM it would be Windows phone 7 or Windows CE the OS they choose as these already run on ARM cpus so just need some tweaking to get them to run on a tablet form factor. Although in the case of Windows CE its looking very dated now so would need some serious updating to compete with iOS or Android.
I would say that a ARM version of Windows server would be a more likely option. Often its just the core server software and 1 or 2 other programs that a server will run so much easier to get a few major server software vendors to recompile their programs for ARM than the millions of programs already out there for Windows 7
I'm wondering what proportion of modern Windows software actually IS x86 - Redmond has been pushing everyone towards .net bytecode or whatever they call it for a long time. In theory a .net framework for ARM would run all this managed code stuff.
Legacy x86 code can be handled by an x86 conversion layer / emulator that runs like poo and crashes every half hour - everyone's a winner.
Of course that ignores the ridiculous hardware requirements to install and run even the most mundane .net applications.
Now, if El Reg had been reading my blog, they'd have known this in mid-November...
No surprises then
Another day another mobile OS from M$.
Yet, somehow Ballmer and Gates like to claim that Linux is splintered.
Possibly not splintered
The point of MS porting Win7 to ARM would be so that tablets and anything else based on it wouldn't have to use the godawful WinCE. Getting rid of WinCE would surely make the world a nicer place. Proper Windows on ARM would give MS an offering not dissimilar to Linux - one API, many platforms, go compile to suit.
Ballmer and Gate's claim that Linux is splintered is surely based on the fact that there are so many different distributions of Linux, all different, all slightly not compatible. I use Linux (mostly Ubuntu) but it's irritating that the Linux ecosystem is riven by stupid and pointless differences of opinion.