Windows boss Stephen Sinofsky has ended months of speculation with the first (fairly) detailed drilldown into Windows 8 on ARM (WOA) platform, and says it should be ready for a simultaneous launch with its x86/64 counterpart. Devices running WOA will come with both a Metro touch-based interface and the more traditional desktop, …
If this works well
Intel should be afraid.
A re-written windows without bloat but with all the office compatibility on ARM could be very interesting. Office clients tend to be under the 4gb level so 64bit isn't a huge deal and bigger ARM chips are on their way.
While office on pure tablet probably isn't a massive draw, getting a transformer-type device which happens to be able to double as a light laptop with full MSOffice could be compelling enough to get significant market share via corporate purchasing. Windows on tablet also takes away momentum towards android and linux on the tablet (and desktop).
It might also have big implications for the VDI crowd. I wonder what the relative costs of headless ARM vs VMware are? It will be interesting to see if this will play out as a defensive move by MS to stave off interest in other OS's or if they will try for Apple's fat profit margin.
Of course, this assume MS can execute well. It is possible that the complexity of Windows and Office is just too great to work well on ARM and may lead to a disappointing experience.
Office plus various web-based apps is the whole of my usage when I'm out on the road, so a lightweight ARM-based laptop with huge battery life running Windows 8 might be perfect. And if it was, say, an Asus Transformer Prime and able to double as a tablet as well, that would be perfect.
I'm liking the concept a lot. As you say, the execution is the key. We shall see.
Execution is indeed the key and I hope get an impression of that shortly.
I picked up a Acer W500 tab reasonably cheaply recently (has AMD's "Ontario" cpu) and have "experienced the joys" of Win7 as a tablet os (fine os on a conventional desktop but definitely pony as a touch os). It is fairly underpowered and the battery life is nothing to boast about, in short the perfect test-bed for the Win8 cp build when it becomes available in a couple of weeks. I am quite looking forward to seeing whether or not MS are within handshake distance of having got this right, at least as far as the x86 build is concerned.
Re: "significant market share via corporate purchasing"
Did you read the article? The ARM version runs nothing except WinRT apps, of which there are currently a big fat None. There's a couple of apps that are supposedly compatible with Office document formats, but that's not "regular Office running on ARM". Also ruled out is any form of virtualisation that might have provided you with a get-out-of-jail-free card, because apparently minimising power consumption is too important to permit it even for desktop machines that are plugged into the mains.
By design, then, this *can't* replace the corporate laptop. Whether MS dare to admit it or not (and apparently Mr Sinofsky has issues here), WOA is purely a phone OS.
Did you read the article?
It comes with a native version of Office, no WinRT needed
I don't see this working well for a whole load of reasons.
I am guessing this is a way for MS to sell more locked in to the device software i.e. get them with the arm then they wont move because it would take to much time / investment to move to another vendor, then MS will also control the upgrade time line for a lot of places, "want office 100? Sure wait your on the arm version arnt you right you will need to upgrade most of your workstations to the newer model" etc.
Not to mention this will hobble a lot of systems, we saw it with the netbook fad a few years back cheep low powered systems that sell for the £200 mark that work fine for about a month then people complain about them being slow, or to small, and that badly made they make a sand castle at high tide look positively robust.
I don't see how this is going to end well for MS,.
"I don't see how this is going to end well for MS"
Well, I think that's because you haven't considered the wider picture. In case you hadn't noticed there is a trend towards ARM (or at least, seeking lower power) across the whole IT industry now. It's no longer just mobile platforms where ARM matters now.
Microsoft *have* to respond to that. They probably need to get Windows Server and Desktop on ARM too just to survive. What we're seeing here with WOA and tablets is clear evidence that MS are positioning the whole product line to be ready for the x86->ARM transition. Sure, what we've seen here is limited, won't run everything, but the constraints are now "artificial" (purely for the sake of battery life), not technical.
It is not so hard to imagine that they could role out a desktop / server orientated version (where there is at least mains power to use) without too much difficulty. Remember that Linux is already there, and whilst MS haven't worried too much about the Penguin on desktops, Tux does do extremely well in server land. MS make money on servers too, and they want to carry on doing so I'm sure.
It seems clearer that there are going to be many sources of hardware - TI, Qualcomm and NVidia are involved - so there would seem not to be a grave prospect of hardware lock in. A bit like the PC market. And that can only be good for consumers. There maybe a good prospect of desktop ARM hardware sooner rather than later. There already is ARM server hardware at HP.
As for performance, I think that the days of ARM being too slow are already gone. And if you don't think they are, the quad core 2GHz 64bit DDR3 parts that are being talked up now should address your concerns. I mean, it's not that long ago that people were dreaming of such performance in their desktops! The smartphone revolution has shown that there is plenty of performance in ARM, and plenty more to come.
Maybe ARM is the future..
I think Intel are woried. For nearly 30 years, Intel have basically had a monopoly in commercial computing. Yes, there have been competitors from time to time (AMD, Cyrix, even NEC for a while), but they've largely been forced to follow what Intel is doing just to remain compatible. There are other architectures (68000, Power PC, SPARC etc), but they were often relegated to niche markets
Now. thanks largely to phones (and, apparently, in particular the support of Nokia), ARM chips are becoming better known. Not to mention the fact that they are increasing in processing power, while using a lot less energy than Intel chips.
Yes, they are only now approaching the 2GHz level now, whereas Intel went past that level a few years back, and are now approaching 4GHz, but most applications don't need anywhere near that speed.. 2GHz is more than sufficient.
Even in Data Centres, I've been told they are looking at ARM for parallel computers. While you aren't going to be running a massively parallel computer on a battery, if it's got a couple of hundred CPUs, the power savings (and therefore savings on the cost of running) could be significant.
There already is desktop ARM Hardware
"There maybe a good prospect of desktop ARM hardware sooner rather than later."
Sorry, been we've been using desktop ARM hardware continually since 1987 in the RISC OS world.
We've had desktops using ARM2, ARM3, ARM610, ARM710, ARM7500FE, StrongARM, XScale, ARM9 and Cortex A8, with newer ARMv7 based boards on the way.
Can't argue with that, but then I'm hardly likely to! Consider that my two now ancient systems hold up pretty well alongside my very much newer machines with much more recent processors, operating systems and apps, even to the point where there are still apps on RISC OS that work as well, if not better than, stuff that is being bundled out on Windows. I'd be lying if I said that I used RISC OS all the time, but for an OS designed in the late 80s and machines dating from the mid 90s (a couple of Risc PCs to be exact), they hold up well. That's one reason why such folk as Intel should be scared, having dropped their input into the ARM, and one reason why Microsoft are finally taking it seriously.
Shame Acorn never survived, but the curse of Boland saw to that.
Clock speed is meaningless
Clock speed (i.e. Ghz) is a meaningless measurement of processing speed when comparing different architectures. Some CPUs perform more calculatations per clock cycle than others, so a 3Ghz chip could perform less calculations than a 2Ghz one within the same timeframe. This is true even within Intel's own family of chips over the years.
I won't be buying it.
It's Android tablets for me.
These look to be locked down tight, too. It will be interesting to see if they take off. And of course anything to avoid having to talk about the mess that is WP7 at MWC.
Office apps should run pretty well, I think.
They spend most of their time just sitting there, waiting for user input, so that's not an issue. And if Apple's iPad 2 can run intensive apps like iMovie, there's really no excuse for a modern ARM (most likely a quad-core design, with NVidia graphics judging by Sinofsky's own words) device not to be able to cope with a bloody word processor, a spreadsheet, a note-taking app and the other apps that come in that software bundle.
However, the proof will be in the execution and marketing. Microsoft haven't had a lot of success with the marketing side of things, so I hope the recent head-rolling operation in that department has paid off.
Competition is good, though I suspect ARM devices running Windows 8 will be mainly fighting the likes of Android if the licensees make their usual mistake of starting a race to the bottom.
Race to the bottom?
You say that like it's a bad thing :P
I for one welcome our new cheap-o fondleslabs!
"Office apps should run pretty well,"
Anything will run well if it's only lightly loaded. My experiences editing 200+ page Word documents, with multiple editors, loaded with tables, diagrams and photos still make me shudder especially as deadlines neared.
Large Excel spreadsheets were so difficult to use for rapid data analysis that we bought JMP which could handle the data size AND was quick.
So maybe with quick new ARMs - we'll see
I remember running Office 4.3 on Win3.1 a 486. Editing large documents, spreadsheets, it all just worked.
I also recall running Office 97 on NT 3.5.1 on a Pentium. Again, worked fine (and was even compatible with all work documents until the 2k7 migration a couple years ago).
Race to the bottom?
I don't see the WOA tablets being 'cheap-o'. The hardware may be, and the WOA ODMs may have to compete on price alone as the specs are controlled by MS. But MS will demand its licence fees for W8 and for whatever they sell as 'Office'.
There were some Windows 7 tablets being sold here. They did actually dual boot into Android or W7 but the price only included a 'trial' licence for W7 and for Office. After 30 or 60 days you had to go an buy a real W7 and Office. I wonder how many didn't notice that ?
WOA is me
The last time I remember the acronym WOA being used in the compueter industry was for the ill-fated Wang Office Assistant. This was a light-weight sort of PC terminal with a version of MSDOS with no real compatibility. As a developer for Wang at the time we were 'encouraged' to write for this misfit - will never do that again. WOA is me seems destined to repeat history as as a .Net developer I am very skeptical about any thing that involves me other than writing in a good cross platform tool - even that is hard enough! If Xamarin develops MonoWOA I might think of porting some of our apps otherwise Mono, MonoTouch and MonoDroid will remain my sandpit!
"but left in standby, for weeks it is claimed"
What's so special about that ?
We have an old laptop running OpenSUSE 11.4 that's left on during the day and sleeps at night - it's hardly ever turned off it might be restarted once a month usually because it's been left unplugged.
Nice to know you are doing your bit for global warming.
It is used all day - if we leave for more than a short time it's put to sleep as are the desktops if they are on. Only the low-power Atom file/odds&sods server runs all the time
Re: Left in standby for weeks...
..."and not plugged in", is the bit that I think was implied, but missing from that statement.
It sounds a bit like they are saying 'Yes we do an Arm version that runs office'. It doesn't do much more compatibility stuff but if covers that particular base. However if you are looking to run all your familiar Windows apps on an ARM device this isn't going to do it.
A bit like an iPad then, just with more of the full blown OS and less of the phone.
I think that MS may well be onto something here trim a full OS down slightly, rather than beef up a phone OS. Personally I'd love to be able to run CentOS or Fedora on a tablet, because I don't like Android, but Windows would be a contender - especially if, as it appears, the tablets will be subsidised to some degree.
It's often easier to scale up than scale down though. Especially if the OS you're scaling up has never had a WIMP based interface, you don't have any legacy apps with mouse and keyboard input to worry about.
Personally I wonder if the dual personality thing is a good idea. It will be a master of none if it's not done right.
Thanks to the works of MSFT
anything with the "Windows 8 Approved" sticker will, by mandate from the Kaiser of Redmond, have no option to disable secureboot. As much as we'd love to install Cent over a win8 fondleslab, we won't be able to. Otherwise, you can do it very soon with this guy: http://devworks.thinkdigit.com/Tablets/KDE-Plasma-Active-Coming-in-its-Very_8645.html
Looks like Intel were correct and the WOA won't run your legacy apps, so apart from Office everything else for WOA will need to be recompiled. So could come down to the same problem MS have with Windows 7 phone, it maybe good but lack of apps will stop people taking it up and because of lack of sales no one wants to write apps for it.
And still i dont see the need for a touch screen full office suite, i would rather gnaw off my face than do any extensive typing on a touch screen keyboard.
"...everything else for WOA will need to be recompiled."
So moving to a completely new architecture with a totally different instruction set requires a recompile? Who could possibly have seen that coming, eh?
The alternative would have been an x86 emulation layer, which would undoubtably have run like a total dog.
Re: "need to be recompiled"
Er, no. It will need to be *ported*. WinRT is not the Windows API that all those legacy apps use.
Some reasons to be worried about ARM
These are my predictions for Windows tablets
a) The tablet will be locked down with a signed bootloader, i.e. no rooting or flashing for you.
b) The tablet won't run existing x86 binaries limiting what you can actually do with it
c) Past experience with other Windows ports suggest very few vendors bother to port native apps to other architectures. A tablet might benefit from .NET apps but is no use for legacy or cross platform apps.
d) Microsoft haven't made it easy for developers of native apps either by producing an equivalent of LLVM so that a single binary could run on different architectures.
e) It will almost undoubtedly be locked into a Windows equivalent of an app store. I do not believe they will permit unsigned apps to be installed or executed on a tablet.
On the plus side, a tablet that doubles up as a mobile desktop with a familiar MS office apps, browser etc. is probably attractive to some people but if you can't install the apps you know and love on top, perhaps the question is why bother at all.
(b) isn't much of a prediction, since MS have said it from the start and this article stated it in no uncertain terms ;)
For (c) & (d), my understanding is the the native "Metro" apps will work on both platforms or have some kind of one-touch compile in order to support both (very conceivable, since these are almost certainly .NET-based, so technically architecture-agnostic).
(a) and (e) sound spot-on to me. I think MS will go much more for the Apple walled-garden approach, with a little of Google's "run unsigned apps" option - mostly for developers. This would fit in with an evolution of WinMo and WinPho.
Definitely interested, though. I think a light, low-powered, "true" Office tool has a market waiting for it.
"Limiting what you can do with it..."
Well thats music to many many corporate IT depts the world over.
A remote device that the users have little choice in the junk they can install on it but has all the main apps the corporation uses for work.
"These are my predictions for Windows tablets"
By a strange co-incidence, these are also observations of Apple tablets, given a few name changes. Everyone always wants consumer lock-in at every level.
Only this time round, MS would appear to have stiff competition from elsewhere, so the pressure to have them open up their systems a little will be significantly reduced.
Try a chord based touch keyboard input, then you can type with one hand while gnawing off the other. You will need your face so you can gnaw.
How often will it crash?
WOA has desktop but no desktop apps?
>>Devices running WOA will come with both a Metro touch-based interface and the more traditional desktop ... all other apps on WOA have to be Metro-style WinRT.
That seems a bit contradictory; you can access the traditional desktop but all the non-office apps will only be visible from the Metro desktop? Did I misunderstand?
I see no problem with lack of app compatibility between Intel/Arm... if you use WinRT you can simply build twice to the two targets in the same way we do for 32/64bit builds?
That's what shocked me as well
If El Reg didn't mistype that, it means that the only applications that run on the ARM desktop would be Microsoft's. No cross-compiling of third-party applications to ARM.
That surely can't be right... Right?
"... means that the only applications that run on the ARM desktop would be Microsoft's."
On the desktop, yes, it's looking that way
" No cross-compiling of third-party applications to ARM."
Third-party apps will need to be METRO apps. But on ARM machines, which are likely to be keyboardless tablets, this is a GOOD thing.
Developers will be able to recompile their 60,000 Windows Phone apps with just a few clicks, theoretically, and distribute their apps through the new Store.
Tablets are full capable of being docked. Indeed one attraction of a Windows tablet is the prospect of running metro when you lug it around but being able to dock it and having a more traditional experience with mouse and keyboard.
Therefore it makes no sense to deny traditional apps. When someone parks their tablet in the stand they may reasonably want to be able to use the device with apps other than MS Office.
Windows Phone apps...? Seriously?
Okay, I guess it's great that 60,000 phone apps can now be run on a desktop OS. But what about the MILLIONS of applications already available for Windows? Can't Microsoft see what an opportunity is being wasted here? Personally, this would have been the single reason for me to even consider buying a Win8 tablet...
MILLIONS applications? Bah!
... all of which are unsuitable for tablet use, thanks to fiddly GUI elements, keyboard shortcuts or keyboard input required now and then. No, but thank you, no.
I'd rather congratulate Microsoft on courage to force developers to adapt to programming tablet UI and turn down these millions of apps. Yes plenty of them are useful, or even necessary for some, but if they aren't usable on tablet - though, they shouldn't be used on one, until adapted.
One of winning points about Apple environment is consistency. If you dump everything + their dog, no matter whether or not it fits tablet use, on new Microsoft tablet, it will lose any appeal it might had. For those who really want to run everything + their dog, I'm sure there will be Atom tablets to choose from.
There's a difference between "Windows Table" and "Windows on Arm"
WOA "denies" traditional apps, i.e. has the locked-down desktop with Explorer + Office.
The Intel/AMD-based tablets will have the full current desktop - but presumably at the expense of battery life, form factor size, etc.
"But what about the MILLIONS of applications already available for Windows? Can't Microsoft see what an opportunity is being wasted here?"
Er, that's why you would buy an x86/64 tablet rather than an ARM one. Only buy an ARM tablet if you don't need to run legacy apps. It's not hard.
You make a valid point, although I've been under the impression that even a non-optimized version of a program ported to an ARM tablet would still use less power compared to its x86/64 tablet counterpart because of the savings inherent to the architecture.
What is expected to happen is that legacy Windows apps will be ported to run under the METRO GUI, which will then allow them to run on both ARM and X86/64 within the new touch-based GUI. MS is moving away from the old desktop towards METRO and ideally will eventually drop the desktop entirely, I would have thought, but that step is years away.
“The conventions used by today’s Windows apps do not necessarily provide this"
Translation: we now realise the Win32 API is crap, and we're taking the opportunity to stop developers using it.
I think you mis-understood
WinRT is a successor to Win32 API, not just a library for Metro GUI. It is possible to write WinRT applications to run on Desktop, simply all Metro functions will fail. Things like sockets etc. will work just fine, apparently it is even possible to do GUI using XAML, with caveats.
As for the larger picture, I see this as a means MS employed to remove some of legacy mess from Windows API, which is long overdue. If that forces developers to do extra work, tough. MS is playing risky game here, since some developers might just switch platform instead, but this is not necessarily bad thing.
Re: WinRT is a successor to Win32 API
"some developers might just switch platform instead"
You don't say!
You already *have* a version of your program that runs on *current* Windows box. Migrating to WinRT (and dropping the WinAPI version) *reduces* the number of customers you can sell to by a such a large margin that it is a ridiculous proposition. So, if you *are* willing to re-engineer your legacy app (which, if you are a successful company, probably has years of cruft built into it) then the choice is to port to some cross-platform API that might let you target Linux, Macs and Windows desktops, or to maintain the original code and (in parallel) a WinRT version for those ARM-based Windows phones.
You'd have to be out of your tiny little mind to contemplate what MS are asking.
Well of course there is higher barrier of entry on tablets, or would you prefer for vendors to just dump all their old desktop cruft on it? If vendor is incapable of refactoring their software, good riddance to it! Market for x86 desktop software is here to stay, x86 tablets will be on the same market, and no one is being forced to write exclusively for ARM tablets (as for the phones, I don't care).
As for the last part, if you never wrote or worked with modular software with GUI code decoupled from other parts, then I'm not surprised it's beyond your imagination. There are plenty of good and average programmers doing just that, every day (look at consoles/PC games). If you happen to be a programmer and dislike refreshing your skills, perhaps it's time to change job?
Woa! Said snowy!
I have to say that this WOA looks like it's had a considerable polish as compared with WP7.