CPU Diversification: Ubuntu’s Gain, Microsoft’s Loss
The drive to create faster, cooler, and more energy-efficient CPUs has led to a diversification of processor architectures recently, with the venerable x86 facing competition it hasn’t seen in years. If this trend continues, it will assure Ubuntu and other Linux distributions a substantial advantage over Microsoft, which so far has announced no plans to build Windows for chipsets that are not x86-compatible.
Apple’s decision in 2006 to ditch the PowerPC architecture in favor of Intel CPUs meant that Windows could run on virtually every consumer-class desktop and laptop computer in the world–a goal towards which Microsoft had worked for more than a decade, as it promoted the x86 architecture over its alternatives.
But that success was short-lived, as ARM chips, like Qualcomm’s Snapdragon (which was recently previewed in netbooks at Computex in Taiwan), stand poised to attack the market share of x86 CPUs.
Dealing with diversity
While the desktop version of Windows supports only x86-compatible machines, the Linux kernel can be compiled for nineteen different architectures, according to Wikipedia. This leaves Microsoft (and Apple, which also supports only x86 hardware) at a disadvantage as the consumer-class CPU market begins once again to diversify.
If ARM and other alternative architectures prove popular among PC vendors over the long term, Windows and OS X will be locked out of a substantial portion of the market, especially on netbooks and other portable devices. In the absence of proprietary platforms, the dominance of Ubuntu and other Linux distributions is all but assured.
Granted, adapting to a more diverse range of CPUs is no trivial task for Ubuntu, which currently has official support for only x86 machines, although unofficial builds for other architectures are available. While most of the work required to port Ubuntu to alternative processors has already been completed upstream, Canonical would still be responsible for maintaining repositories for new architectures, which cannot be implemented overnight and which would require the investment of more resources.
Nonetheless, the task of officially porting Ubuntu to run on new architectures would almost certainly be much quicker and smoother than the mountain of work that Microsoft and Apple would face to do the same for Windows or OS X. This reality does not bode well for Microsoft in particular, which is hoping to use Windows 7 to solidify its growing dominance of the low-end laptop (i.e., netbook) market. No matter how much lighter, prettier or cheaper Windows 7 is than Vista, it’s useless in an increasingly significant portion of the netbook market until it runs on ARM and other alternative CPUs.
Microsoft may well decide to diversify the machines that it supports. But unless it does, Ubuntu has a lot to gain as the x86 processor loses its monopoly on the desktop and laptop market.
Follow WorksWithU via Identi.ca, Twitter and RSS (available now) and our newsletter (coming soon).
WorksWithU on Identi.ca
WorksWithU on Twitter
WorksWithU's RSS Feed
Just one point, the Iphone operating system is supposedly an ARM port of the key components of OS X. The upcoming Applet Tablet could easily be running on an ARM processor. Indeed, Apple employs one of the top ARM chip designers.
The emergence of ARM and MIPS as challengers to x86 are more of a threat to Microsoft than Apple. But this is largely an issues cause Microsoft’s core customers are enterprise and the emergence of ARM is so far in the consumer space
I was going to write a “Well actually…” post about Apple and Arm but it seems someone beat me to it.
Re above comments, that covers ARM but there are other chips. And surely the main thrust of the article is about Windows?
I think that open source software will always be quicker to market than larger corporates who, by their very structure and size, are totally unable to move at any speed.
In the commercial marketplace governments push a lot of legislation onto companies. Corporates can handle it with staff dedicated to said handling. whereas the smaller company which can normally run rings around their corporate rivals get bogged down by the sheer weight of paperwork.
Open-source, at the moment, is free of a lot of this legislation, but you can be sure that Microsoft and Apple will be consulting their Congressmen with ideas to slow the Linux companies down.
Ampers.
Microsoft isn’t the one with non-x86 issues, it is the MS customers and vendors who have this issue. MS can recompile their OS for any architecture all day long, but as long as their users cannot run the x86-only software on the new chips, the OS is useless to users.
If you could buy Windows but it didn’t run your software, or Linux which also won’t run the software but seems to be everywhere these days, which would you pick? I’ve already made my choice, lets just make sure that more follow us.
x86 isn’t going anywhere. And Microsoft will certainly not allow Linux to gain traction in any significant market regardless of what they have said in the past. If ARM chips become increasingly common and this somehow leads to an increased Linux or Ubuntu uptake, you can bet that Microsoft will show up quite quickly. It isn’t like Microsoft to abandon a market, even when all hope of being able to compete seems lost.
Although ARM might be a Linux beachead into desktop/netbook computing, I think there are still many vendors installing Windows CE. Remember, technical superiority doesn’t always mean marketshare superiority.
I am sure that whatever computing platforms become commonplace in the future, we will all have Choice in what OS we run. Isn’t that what matters?
Ultimately, if the consumer doesn’t win, no one wins.
Gee chris, in the article before this you wrote how there is too much choice in Linux. Make up your mind!
benfrank: for the record, I think choice is great. But the Linux community could stand a bit more pragmatic thinking if it wants to make Linux a platform for the masses.
I’m all for having 1000 different Linux distributions to choose from, but I think the community’s resources should be concentrated on those with the largest user bases.
Similarly, if there were 1000 different CPU architectures, and there were enough geeks to make Linux run on every one, I’d be behind that as well. But if 3 or 4 of the architectures controlled 99% of the market, it would be silly to expect developers to support all CPUs equally, rather than focusing on the handful that matter the most.
Very good points, Cristopher. In fact, Linux was well ahead of the curve (and still is) in 64 bit support. Something I would add is that there are a couple killer applications that are binary only, in this sort of smaller devices: flash and skype. For example: PS3 linux users have no flash, and thus a poor web experience. Archos (arm) users have no skype …
I still agree overall. Diversity is a killer for a monopolistic company, and fertile terrain for open source
First, Ubuntu != Linux. The fact that Ubuntu isn’t built for more than x86 isn’t really relevant as many other distros are (e.g., Debian is built for 11 different architectures including mips, arm, ppc and sparc; Fedora is build for at least ppc as well as x86).
Second, while I’m no fan, Microsoft might have a few tricks left. I believe when Windows NT was originally distributed, it was built for both x86 and MIPs chips. It’s possible they could dust off that old code (a version of which is probably still in Windows 7!!).
George: Re: “Ubuntu != Linux” . Not to speak on behalf of Christopher. But this is an Ubuntu site (which may be clear from the title), so I think it is natural for this post to focus on Ubuntu.
BTW, look at these numbers:
http://www.linuxjournal.com/content/what-your-favorite-linux-distribution-desktop
The *buntu family is kind of popular
How difficult is it for MS to port Windows to other platforms? What’s to say they aren’t already doing it & we don’t know ab it yet? This seems to be our strategy — wait for MS to NOT do something, claim it as a victory & then watch them go ahead & do that very thing, ala Netbook, then spin the passive victory & subsequent failure as FUD, or a conspiracy. What will the Linuxes do to assure this lead is not lost? Is there a plan?
I wish the Linux Foundation, or somebody, would really take the lead in in things like this & working w/ manufacturers to make the Linux kernel something that is actively, widely developed for & to improve compatibility. I would love to see them also coordinate strategy among the Linux Developers & Distros. Freedom shouldn’t mean failure. We need coordination & planning to liberate computing.
Leo: thanks for that link. I was surprised to see that even Debian has higher market share than Fedora, at least among the readers of Linux Journal. SUSE’s numbers are also considerably lower than those that I’ve seen elsewhere.
John: you make a good point, and I agree that the Linux community lost a major opportunity by resting on its laurels in the netbook market, only to be overtaken by Microsoft without a real fight.
That said, I suspect that if Microsoft were intending to build Windows for ARM anytime in the foreseeable future, we’d already be hearing about it, because the company would want to generate buzz and encourage PC vendors to include Windows in any plans for ARM devices that they might be formulating. Since Microsoft has yet to announce that it’s even thinking about porting Windows to ARM, it’s likely that it has no intention of doing so, at least for the time being. I imagine Redmond wants more assurance that ARM on netbooks is not just a fad before it invests money into an ARM port of Windows.
It’s also hard to know how easy it would be to port Windows to ARM, since we can’t see the source code. Torvalds made the decision in the early 1990s to design the Linux kernel modularly so that it would be easy to port to new architectures; since Microsoft has been so closely tied to x86 since the DOS era, it might be fair to assume that Windows wasn’t designed to be easily ported to new CPUs. That doesn’t mean Microsoft can’t port it to ARM if it wants–it certainly has the resources–but it might take a while.
But as you say, this doesn’t mean Linux should sit around assuming that Windows will never be ported. I very much agree that there should be a plan for taking advantage of this opportunity. But given the radical decentralization of Linux, and the Linux Foundation’s lack of grounding in the real world (as opposed to the Stallman/Raymond world where every end user is an expert programmer and free-software ideologue), I’m doubtful any kind of meaningful plan involving the whole community would develop. Canonical/Novell/Red Hat might be able to pull something off if they tried, however–especially Canonical, since it’s most focused on the desktop, where ARM is most relevant.
Windows is proprietary and is dependent on proprietary third party binary applications and proprietary third party binary drivers. This is why porting Windows to other CPU architectures will not be successful in capturing marketshare. Third party proprietary developers will only support Windows on x86. Microsoft realises that by concentrating development effort on x86 and not popularizing other architectures by porting Windows to them, they stand a better chance of building a proprietary ecosystem that they can control.
In the 1990’s Win NT was ported to the ALPHA cpu, and im sure the MS boys (and girls) know how to use a compiler.
And dont forget! It’s MS who after all runs the show, and if a CPU or a product will not run the friendly and familiar MS software, then they will reject it.
MS will enter the market aka. the Netbook market and people will see the familar Windows desktop and open their cheque books.
After all, the customer is always right, you can tell someone about linux all day, but if they prefer Windows they will use windows.
Until, Linux gets to the stage where is can do everything that MS Windows can do, and offer MORE with little or no worry or hassle in changing over and you’ll win huge market share.
But offer less or the same or similar but different and they stay away like its the plague.
Basing your hopes on a Linux “Lock-in” trick wont work, it did not work for netbooks and it wont work now. Sorry.
SMP
you seem to forget the Operating System _IS_ the layer between the hardware and the applications.
I know many Linux people like to “USE” their operating system.
But for more people, they use applications that run ontop and hook into the OS either by API’s or whatever (abi’s and so on).
Therefore, once you get the OS working on a CPU it will function as the layer for the hardware (ie the REQUIRED binary drivers, for the new hardware) will be handles by the OS.
Once the OS is up, its a trivial matter of running the applications ontop of the OS. If its done correctly you would not even need to re-compile the apps, but mabey you do.
Either way, im sure 95,000 people in MS would be able to work it out and get the code re-compiled quicker than Linus could go through the 5+ Millions lines of Kernel source by himself. !!.
[...] CPU Diversification: Ubuntu’s Gain, Microsoft’s Loss If ARM and other alternative architectures prove popular among PC vendors over the long term, Windows and OS X will be locked out of a substantial portion of the market, especially on netbooks and other portable devices. In the absence of proprietary platforms, the dominance of Ubuntu and other Linux distributions is all but assured. [...]
“Once the OS is up, its a trivial matter of running the applications ontop of the OS. If its done correctly you would not even need to re-compile the apps, but mabey you do.”
@darryl
What you are describing is not just an OS but an emulator, where software emulates a different processor. That is *very* expensive in terms of CPU load. As I understand it, the “alternative” processors being discussed are generally not very powerful. So I imagine if you tried to emulate x86 on them they apps would run very, very, slowly — to the point of being unusable.
For a normal (non-emulating) OS, while the apps would make system calls, most of the time the apps are executing their own native CPU instructions. The other way what you describe could work is if all of the apps are interpred or run in a vm, as I believe is the case with .NET. But in that case the apps have a performance penalty *regardless* of the CPU they run on.
In regard to the possibility of MS porting it’s entire OS at this point to another architecture, I just don’t see it happening. Where I work we’re testing the deployment of Windows 7 – yes, it’s an RC, but we’re a (un)fortunate MS guinea pig here. Windows at this point has so many integrated components, isolated compatibility functions and a million other things, that another architecture to support what’s going on in the system, would be an undertaking that would require more time than it took to supposedly develop Windows Vista.
Right now Windows 7 can’t even run 10% of our regular business applications here on basic x86_64 arch. which should be flawless.
Linux and friends are about all that will be on ARM and MIPS, and if played right, a great place to showcase a cost saving hardware ecosystem for businesses in the foreseeable future.
I think something to consider is that the OS is becoming less and less relevant, while more and more of the functionality is run from servers and played inside a web browser. Things like: does Flash run on my platform? Does skype run on my platform? These are becoming very important questions …
[...] Weekly Newsletter #143Ubuntu Weekly Newsletter #142 WorksWithUTest-driving Chrome for UbuntuCPU Diversification: Ubuntu’s Gain, Microsoft’s LossLinux Standards, And Why They Shouldn’t Matter Ubuntu GeekGRUB 2 now default for Ubuntu 9.10 [...]