Quite a bit of enthusiasm seems to be building for ARM’s upcoming processor for netbooks and other lightweight computing devices. The Cortex-A9 is promised to have substantially better performance than the current crop of AMD processors, and a video released by AMD ARM gives a pretty convincing picture that the Cortex-A9 will have comparable performance to the Atom. Watch the video after the jump.
As an EETimes article points out, “ARM set up both systems to run the same operating system, the same browser and put them on the same high-speed corporate network and checked for responsiveness to the end-user. The Intel-based netbook seemed to beat the development board to display different websites most times, but not by very much. When it is considered that the netbook has a graphics processing unit and the development board does not, that is surprising result. Even more notable is that the Intel-based netbook is running at 1.6-GHz clock frequency the dual-core Cortex-A9 development is running at 500-MHz.”
As ARM is surely going to keep pointing out to everyone, they don’t have to beat Intel’s raw performance to make a big splash in this market, because for these kinds of devices, speed isn’t everything, and their promised power consumption advantage will surely be a major selling point. Of course, alternative OS fans will be cheering ARM on because Microsoft won’t be able to buy their way into the market the easy way with bargain-basement XP licenses like they did with netbooks. Devices based on the Cortex-A9 will likely use the new crop of mobile OSes from Google, Intel, Nokia, etc, not to mention the innovations coming out of unknown upstarts.
Of course, performance is going to be important. The current crop of ARM chips doesn’t have quite enough horsepower for a pleasant computing experience on larger mobile devices. But this video is certainly cause for hope that exciting new devices are on the horizon.
dual core cortex a9 at 800mhz use only 0.5 watt
so the 500mhz use less
2Ghz use 1.9W
a quad core and 8 core of the cortex a9 will go out…
powerfull small device can be created
More vaporware?
I’ve been waiting for there to be ARM notebooks or netbooks now for two or three years now, and frankly, so have many of the other readers of OSNews and other tech sites, unfortunately no one wants to sell them to us.
Until I see an ARM netbook that provides all the features, the Linux OS and the great prices that the ARM fanboys have been claiming are possible I’m done with them. Every year we hear about the great things that are possible, but no one seems to want to deliver on those promises to the average geek. I’m tired of it.
–bornagainpenguin
Blame OEMs scared of doing anything new because of the recession. The HP Slate is a prime example of the incredibly cautious PC industry. Make it half-arsed, and if it fails, put the blame on Microsoft.
It is also the stronghold Intel and Microsoft have on the OEMs, it says a lot that you can almost find ARM everwhere where Intel does not have a grip on things. On the PC side you cannot find it.
It is the same reason NVidia does not get a foothold with the superior ION chipset compared to the inferior Atom offerings on Intels side.
Also add to that that Windows on the Netbook and Notebook side of things is a huge seller. You cannot run Windows you create a nieche product, period. Unfortunatly, I would love to see a Tegra netbook pulling off 1080p with stressing the processor 10% , that is possible, but blame Wintel that we cannot get it.
You guys….. You really have no idea what you are talking about… do you?
The problem is the chip manufacturers not the ODM / OEM.
This is the case:
Intel and AMD knows how to sell chips to the computer / PC industry. They do so very well and and effectively. An atom incl chipset can be had in low quantity from $30 to $50 depending on specs.
The ARM cpu manufacturers are a bunch of amateurs comparatively. They service three markets primarily:
– Industrial Control (High price small volume)
– Networking (Low price high volume)
– Mobile Phone (Mid price, high volume)
The problem is the Industrial control segment, if they start to service the PC industry it will kill off their premium pricing on the Industrial Control segment and none of the suppliers has been willing to do so until now. Marvel is now trying to avoid this problem by working directly with a few Chinese suppliers to make some ODM products that cannot be used in Industrial Control and therefore will protect their lucrative profits in this segment.
Basically that CPU shown here has a current price that makes it completely uninteresting to use as a general PC. And that is the problem, unless you are willing to sign a contract for 250K quantities (mobile phone) these CPU’s are not available to you at a price that is even remotely competitive with an Atom chipset. As long as an ARM CPU alone (just the CPU chip) costs more than twice of a fully assembled Atom based motherboard but will only have about half the speed, there is no ODM in the world that is going to give that product a second thought.
Slambert666 explained…
That’s fine, but then could they kindly shut the #### up about how wonderful their vaporware is?
–bornagainpenguin
So you are calling TI amateur? Premiums or not, the chips that they make for the industrial control market are definitely different and, mostly, have military grade specs. Lifetime of those chips has to be incredible. Most Intel produced “crap” dies at a quarter of the lifetime of those chips.
do you?
I can see the logic in your argument but there’s 3 flaws.
1) The chips you use for industrial control are completely different from the chips you use in netbooks. ARM has 3 series of processors, The A (Application), R (Real Time ) and M (Microcontroller) series. For industrial control you’ll probably use an M series or maybe an R series. For a netbook you’ll use nothing less than an A series. The processors going into the two segments do not compete with each other because they have completely different requirements.
2) ARM processors are made by everyone and their dog, there’s lots of companies all competing with each other. If one of them attempts to sell them at a premium you just go to someone else and get it cheaper.
3) ARM SoCs (System on Chip – they’re not sold as just processors), tend to cost rather less than ATOM chipsets. ARMs are cheaper to make and are made in larger volumes than ATOM.
1) Industrial control is of course are wide range of applications. But dude I’m not talking about microcontrollers here. A single core (A8) 400MHz Application processor costs $25 In 1K quantities. The chips that can go into general purpose computers cost from $89 to $250 in 1K quantities (14 to 26 weeks delivery time).
2) Yes you would think that, but prices are amazingly similar.
3) please see 1 and 2.
Please let us (me) know where we (I) can get those cheap ARM CPUs that you are talking about. One thing is theory another is reality, in the real world Atom costs half with twice the performance (and I’m being extra nice to ARM here).
By the way pricing schemes are just the beginning of the problems. The whole industry is infested with NDA, Application specific pricing, personal contacts as the only entry point etc, etc.
You’re looking at it from the wrong side. The pressure is put on OEMs from consumers who, if given a device with a UI that looks and behaves like Windows, expect it to work with all of their peripherals and run all of their Windows applications.
In order to break that mold, you need a new form factor and a new interface so that consumers’ expectations will change. The tablet form factor, which is not new, but is being revitalized, will suffice for the former. The latter, however, is a challenge. OEMs have nowhere to turn for an ARM OS that provides a fully touch based interface and provides a new platform for touch based applications. It simply doesn’t exist in usable form yet. At least, not outside Cupertino.
This is why Apple felt no need to preempt Microsoft or CES with the announcement of their tablet and is delaying until the end of the month. Nobody else has the software needed to compete with what they have the ability to offer right now. They will present a device that probably has an ARM processor, runs a modified version of iPhone OS, and provides a modified CocoaTouch API for building tablet sized applications. My bet is that it will be a massive success and that everybody in the PC world will be scratching their heads wondering what the hell went wrong.
Of course, it’s easy to see what went wrong. Microsoft spent two decades building a fortress on top of a mountain called the desktop PC. The Linux folks have been chipping away at the walls of that fortress for a decade. They’ve just started to break through only to discover that the entire mountain is crumbling underneath them.
But don’t despair. If you’re excited about dumping Wintel and embracing ARM, then just wait a little while longer. The opening salvo has been fired and the response will come in a few weeks. The Wintel round of general computing is coming to an end and the next round is about to begin. The players this time: juggernauts from Cupertino and Mountain View.
I was “fixing” a pair of Vista laptops (for friends of the family) just recently. Vista64 didn’t meet your criteria.
Vista64 did have a UI that looked and behaved a bit like Windows, but neither laptop worked with all of their existing peripherals and neither ran all of their existing Windows applications.
The owners of the laptops had to buy a new printer (to share between them) and new software (each).
PS: “Fixing” a Windows machine, BTW, basically means booting a linux LiveCD and using it to save all of the user’s data, then booting the Windows install or recovery CD and wiping the disk, and re-installing all software, running Windows updates for a few hours, and then finally restoring all of the previously saved user’s data. It takes a couple of days or so (depending if you have to hunt on the web for drivers, using the Linux LiveCD again), versus 20 minutes or so for the equivalent operation with Linux.
Edited 2010-01-08 09:33 UTC
My counter-anecdote is that I recently “fixed” a workstation that is used for structural design with AutoCAD. I installed Windows 7 on it to find that it wouldn’t work properly with the Wacom tablet. This is a case where the tablet was worth more than the whole computer so we reverted back to XP Pro.
Obviously, the requirements I mentioned don’t hold across the board. After all, many people have successfully switched friends/family over to Linux without any issues. But these things happen on a case by case basis and are usually handled by an experienced person such as yourself who is able to provide support afterward. Doing it on a mass scale as an OEM is a completely different story.
No argument from me. The ability of most Linux distributions to preserve /home during a reinstall is a major time saver. Of course, I’ve never encountered a Linux user who needed help with a reinstall– they are usually capable of handling it on their own.
Please don’t mistake me as a Windows advocate. From where I sit and with all things considered, I pretty much place Windows and Linux on equal ground these days. In other words, being completely honest, I find them both equally disappointing with OS X being only slightly less disappointing.
That isn’t a counter-example … that is an entirely similar occurrence as my example.
The problem is that the Windows market heavily depends on binary backwards compatibility with an ancient architecture … i386.
This fact is rapidly forcing Windows into all kinds of increasingly awkward corners from which it won’t be able to back out.
There are problems with moving to 64 bits. There are problems in updating Windows to a new version. There are problems in retiring the positively ancient and hideously unworkable and non-standards-compliant IE6 browser. Older hardware devices no longer have working drivers. Etc, etc.
It will be impossible to move the existing Windows applications library to a new architecture. They are wedded to i386 for life.
This represents a great opportunity for alternative software on newer and more capable architectures and devices.
The main opportunity that ARM takes advantage of is the high overhead of power consumption required by the complex i386 instruction decoders and sequencers.
Edited 2010-01-08 10:54 UTC
It’s a counter example because while your users were willing to buy new peripherals and software to deal with a new OS, mine wasn’t– which was my original point.
I don’t disagree. This is Microsoft’s choke point.
The problem is that there is nowhere to go. Right now, the lack of Windows on ARM is not hurting Windows. It is hurting ARM.
The point of my original post is that users have expectations based on familiarity. Dumping a traditional Linux distribution on a netbook or tablet plays into those expectations and leads to disappointment when they don’t hold up.
Apple has shown with the iPhone and will show again with their tablet that it is possible to build a new platform from the ground up and present a user experience that breaks with expectations from the past. Users won’t expect the Apple tablet to work with their printers or run their office suite. And they won’t expect it to be compatible with Windows software.
This is what ARM needs to move forward– a whole new experience that allows users to view the device as something different. Transplanting the traditional desktop, only with the Linux bits underneath instead of the Windows bits, simply isn’t going to fly.
By the way, the large collection of open source software that can be recompiled for ARM is not really a Linux advantage for two reasons: 1) most of the useful software can run on Windows as well, providing an equal playing field and 2) the interfaces in that software are not usable on small devices, particularly when touch is the prime method of input.
That said, I do think that when the time comes, these ARM devices will run Linux and not Windows. But as I stated before, they won’t be GNOME or KDE desktops and they won’t run OpenOffice or any other desktop software. They’ll be something different and those that aren’t will be relegated to niche markets.
I would debate that very strongly. AFAIK there are more ARM CPUs sold than x86 CPUs.
Two things wrong with this: (1) In the initial round of Linux on netbooks, OEMs did not offer traditional Linux distributions. It was pathetic offerings such as constrained versions of Linpus or Xandros, and they were terrible compared to a traditional Linux distribution. To this day it is difficult to find an offering of KDE4 on a netbook, but it works on netbooks beautifully. (2) There isn’t any “disappointment” other than the impression of such that the Microsoft-aligned media is trying to portray. meanwhile, about a third of netbooks ship with Linux.
Two things wrong with this, too: (1) Windows on netbooks wasn’t a new platform, in fact it was a last-gasp revival of an old platform, but it did sell OK, and (2) Linux doesn’t need to run Windows software on netbooks (no-one is going to run Autocad or Photoshop on a netbook), as long as it has a familiar GUI and is data-compatible. KDE4, Firefox and OpenOffice can be made to fit this bill beautifully.
Well, that would work.
I don’t see why not. I personally have installed Linux on four different netbooks, two are dual-boot, and all four users are perfectly happy with the Linux. They browse they web and use Office documents without a problem. 33% of the current netbook market says you are probably wrong on this point.
If one can run the same userland software on Linux on ARM and also on Windows on Atom, then you are far better off running it on Linux on ARM. No existing malware will run on Linux on ARM, and all of the userland software we are talking about is pre-installed for you (you don’t have to hunt the web for it and install it yourself after purchase of your netbook), and it is all also updated by the one system-wide auto-updater.
Not a problem unique to Linux. having said that, check out Plasma for KDE 4.4 on a netbook:
http://aseigo.blogspot.com/2009/04/plasma-on-netbooks.html
http://en.wikipedia.org/wiki/File:Screenshot_of_KDE_4.4_plasma-netb…
http://www.liliputing.com/2009/12/kde-plasma-netbook-interface-demo…
Why on earth not? They are doing that already, right now, for 33% percent of the market:
http://www.liliputing.com/2009/02/1-in-3-dell-netbooks-runs-ubuntu-…
http://www.dell.com/content/topics/segtopic.aspx/laptop-mini?c=us&l…
http://blogs.computerworld.com/a_for_dells_new_ubuntu_linux_netbook
http://www.zareason.com/shop/product.php?productid=16216
33% of netbook sales, and 100% of the soon-to-come smartbook class of machines, isn’t a niche market.
http://en.wikipedia.org/wiki/Smartbook
Edited 2010-01-08 12:32 UTC
Lemur, your post is of sufficient length.
I’ve seen several technical and non-technical friends of mine try Linux — some of them quite eagerly, really wanting it to work — and pretty quickly run into That One Application from Windows that they just have to have, that has no native Linux version and won’t work under WINE. A non-technical friend of mine in College (a Theater major) got Ubuntu installed on his own, and, with the help of a working-programmer friend of mine, got the NVIDIA binary driver installed and working, as well as WINE (for reference, this would’ve been two or three years ago, so, we’re probably talking about Ubuntu 6.10 or 7.4 or so). And he could not, for the life of him, get City of Heroes — to which he was addicted — to work in WINE. He switched back to Windows, unhappily. And this guy really bought the Free Software ideology, and really wanted Linux to work for him.
Several more technically minded friends of mine got Linux installed — one Ubuntu, one Slackware — only to quickly figure out that there’s nothing they can do in Linux that they can’t do in Windows, but lots of things they can’t do in Linux that they can do easily in Windows.
(Well, at least, for the set of tasks they wanted to do. There are plenty of things that I like to do, that are much easier in Linux than in Windows: that’s why I’m a happy Ubuntu/Slackware/Fedora/Sidux user.)
That’s pretty much the point here. People have a certain set of expectations for a desktop OS: whether they’ve been formed around Windows or not is irrelevant, as is whether Linux distributions can be made, with effort, to meet those demands. Linux distributions can’t supplant Windows on the Desktop, or even compete with it, even it is technically better (which I’d say it is), because, even if people wouldn’t exactly say “I want specifically Windows as my Desktop OS,” if you asked them to enumerate what they do want in a Desktop OS, their list of requirements would narrow the selection down to only Windows pretty quickly. They want Windows software to work, and work flawlessly and integrate well. They want Steam and MS Office and WoW and their 15-year-old, 16-bit DOS timekeeping application. Ditto for their hardware, be it current, bleeding-edge or ancient.
Equally, new platforms are only going to succeed where that’s not the case. OS’s other than Windows can work on tablets — if tablets take off — because people don’t use tablets for the same tasks they use Desktops for, and therefore don’t expect or want to use the same tools on tablets that they use on Desktops. Nobody wants MS Office on a tablet (well, they probably do, but accept they can’t have it), because there’s no great way to input and manipulate large amounts of text on a tablet with a touch-screen keyboard (unless handwriting recognition’s improved dramatically while I wasn’t looking). Nobody wants to try to play WoW or TF2 (hell, or Starcraft) on a touch-screen tablet, without a mouse or keyboard. On a tablet, people just want to finger-drag web pages around, watch youtube videos, and read e-mail. With a little work, Linux could be made to do those things well, so Linux, if it’s fixed up a bit, could really work on the tablet. Gnome or KDE specifically, probably not so much, but some other, purpose built environment on top of X, probably so.
Basically, long, crazy, ignorant, poorly-worded, stating-the-obvious post cut short, alternate OSs will only work in settings where people don’t want Windows applications. And tablets may be such an environment.
So will smartbooks be such an environment. One can’t get Windows on a Smartbook at all.
For netbooks … you would have to name a netbook-style application that isn’t available on Linux in order for your argument above to be valid. A netbook isn’t a desktop. To empahsise this point … be aware that Linux has 33% of the netbook market even with Microsoft giving away XP Home. When Microsoft starts trying to charge appreciable money for a workable version of Windows 7 (i.e. not the starter version), and starts trying to charge more money for Word 2010 than the entire cost of a Linux netbook hardware puls complete software … then I think your argument has entirely run out of legs.
For whatever actual real-world uses anyone would want to put a netbook, Linux fits the bill perfectly well. When you buy your $299 Linux netbook you already have all the required applications ready to go, pre-installed for you. (Remember also that a Windows machine is delivered only with the bare OS and some next-to-useless 30-day trial versions). Your Linux netbook will remain good as you use it too, because it won’t succumb to viruses or malware (which are all targetted at Windows).
A full version of Linux on a netbook or smartbook makes perfect sense.
You don’t mess about with Wine, because the native Linux applications that are already installed are far better than trying to stitch in any foreign Windows stuff.
The native browser in Linux stomps all over IE. Have a look at just one application that would be included, ready to go, with perfect applicability to use on an netbook in combination with a digital camera or a camera-smartphone:
http://www.digikam.org/drupal/about?q=about/overview
http://www.digikam.org/node/341
Here is another that is perfect for a netbook with a smartphone or a portable music player device:
http://amarok.kde.org/
Even the lowly calculator can have great usability for a netbook:
http://www.speedcrunch.org/en_US/index.html
Then, of course, there is OpenOffice which is every bit a capable replacement for Microsoft Office, and it runs just fine on a netbook or a smartbook class machine:
http://why.openoffice.org/
You get all this with the as-purchased netbook, you don’t have to spend hours post-purchase chasing it all down (and possibly catching a trojan in the process).
What is even better is that all of the native Linux applications are included within the scope of the one system-wide auto-updater.
http://en.wikipedia.org/wiki/Update_Manager
You get all this when you buy your Linux netbook. No more to pay. In contrast, with a Windows 7 netbook, you are easily looking at tripling the purchase cost to try to bring the machine up to the same standard.
Edited 2010-01-09 05:45 UTC
boldingd did a great job of explaining my other points in terms of user expectations and why replacing Windows with Windows-look-alikes will lead to problems, but I’d like to take this particular point a bit further.
There are two fundamental issues with the desktop paradigm.
1) It doesn’t map well to human thought. The brain is primarily spatial and associative and we work most efficiently when we decompose complex problems into simpler pieces. The desktop model works against these processes. It forces us to store data in hierarchical structures and provides functionality through large, complex, monolithic applications. People who have used computers for years have retrained themselves to work around these issues to the point that they feel natural. Those that aren’t accustomed to computers tolerate these issues because they don’t know any better.
2) Translating a tangible model like a desktop to a computer is a poor way to make use of available technology. It brings with it some familiarity through relation to the physical world but also drags along the limitations of the physical world. The desktop paradigm limits us to folders, documents and applications that operate on them while a computer really has the ability to offer a much richer model of creating and consuming content and data.
For example: assume someone has sent you an email with an image attached. You’d like to modify the image and then send it to someone else. Currently, we must open the image in an external editor like Photoshop or the GIMP, apply our modifications, save it to a file, compose a new email message, attach the modified file and send it. This process represents a direct mapping of what we would do in the physical world ignores the ability of a computer to make the process more efficient. Why can’t the computer allow us to edit the image directly in the body of the email using context sensitive image tools and then forward it to the intended receiver? It can, but the desktop model doesn’t work that way.
Can we modify the desktop to work that way? Sure, by providing layer upon layer of hacks that attempt to implement features like the one I’ve mentioned, but they will all break down because the abstraction will not be pervasive throughout the OS. At some point, the desktop metaphor will leak through and it will all fall apart. We need to stop thinking about documents and folders and applications and start thinking about content and data and tools. This goes back to the discussion about sync applications for music players and your suggestion that people can just open up a split pane view in Dolphin and drag their music files or folders to a device. People don’t want to do that nor should they have to. People want to sync specific play lists or genres or songs within a certain range of ratings or some combination of the above along with other factors. They don’t want to manage files and folder hierarchies. Basically, they want to tell the computer what to do– not how to do it.
Now we come full circle to the original topic of the thread which is Wintel vs. ARM. Like it or not, the desktop paradigm is Wintel. The two are irrevocably the same. The massive computing ecosystem that has been built around Wintel and the desktop for the last two decades is possibly the most ingrained and expansive global system in human history with the probable exception of the internet. This can and will be broken, but it won’t happen by replacing the bits piecemeal. You can’t provide a better desktop than Windows and you can’t provide a better x86 than x86. This is not because Windows and x86 are the best, cheapest or most efficient products of their respective classes, but because use of anything else puts you at a severe disadvantage when participating in the global computing ecosystem. Whether you feel this is unfair due to illegal bundling, anti-competitive practices, user lock in or anything else is irrelevant. It is what it is.
What you can do is provide a better computing experience wholesale and do so in a way that user expectations for participation in the ecosystem change. Regardless of your feelings about Apple’s developer policies and closed nature, the iPhone has done this on a massive scale, and Apple will do it again with their tablet. Google will do it with Android and in a different way with ChromeOS. These types of devices and operating systems will scale up quickly to swallow the desktop and with it, the Wintel crown and any that chose to follow that paradigm including GNOME and KDE. This will happen because these devices provide a simpler and more humane method of use and interaction. People tolerate their computers but they enjoy their iPhones and Android devices.
The future is the web. The future is natural input: voice, touch, gesture. The future is content. The future is a large set of small, standardized, pervasive, interoperable tools that allow us to create and consume that content. The future is computing that takes advantage of all available resources to improve our lives.
The desktop has no place in that future.
Edited 2010-01-09 04:59 UTC
Lets just say that I completely disagree with you on your “cannot replace Windows” thinking. Of course it can be replaced. I am doing that right now, as are 33% of netbook users, and 100% of OSX users.
Now for your other point:
I think you have seriously missed a trick about the functionality, programmability, adaptability and flexibility of KDE4’s Plasma.
http://arstechnica.com/open-source/reviews/2009/07/kdes-new-plasma-…
http://aseigo.blogspot.com/2009/07/plasma-in-kde-44.html
http://nepomuk.kde.org/
Here is a paper (warning, PDF file):
http://morpheuz.cc/netbook-paper.pdf
Try to get your head around the concepts of Plasmoids, containments, activities, multiple workspaces, the social desktop and the semantic desktop.
http://en.wikipedia.org/wiki/Semantic_desktop
http://en.wikipedia.org/wiki/Social_Desktop
http://en.wikipedia.org/wiki/OpenDesktop.org
http://en.wikipedia.org/wiki/Plasma_%28KDE%29
This is all highly appropriate for use with a netbook/smartbook.
It will take all of the wind out of your sails if you do manage to grok it all.
BTW: The iPhone is not the only smartphone. People are already bitching about Apple locking it down too much.
Edited 2010-01-09 06:19 UTC
We will agree to disagree on this point then. Whether Windows can be replaced on ~9% of PCs, ~8% of which enjoy direct ISV support from Microsoft is not quite demonstrative of an ability to replace it on a mass scale.
I’ve been following Plasma since its inception and I am very familiar with it. I’ve applauded the KDE developers several times in comments on this very site for the work that they’ve done. It’s a small step in the right direction, but it’s not good enough.
As I said, I’m not interested in layers built on top of the traditional desktop, no matter how well done they are. Let me know when the KDE team ditches the desktop and focuses solely on Plasma. Until then, KDE continues to suffer from the desktop abstraction leak.
I also mentioned Android devices. The iPhone is just the first smart phone that the general populace actually wanted to use and could use without a manual or a handy geek nearby. It turned the entire smart phone market upside down literally overnight. But that’s not all. It’s also a general computer that fits in your pocket and has over 25,000 applications available, including productivity apps and a healthy game collection. Ignoring that would be foolish.
Edited 2010-01-09 07:05 UTC
So you are saying that you buy the myth that Linux is only 1% of the market?
Happily there are people who are not quite so easily deceived as you:
http://blog.linuxtoday.com/blog/2009/05/1-linux-market.html
Edited 2010-01-09 10:12 UTC
Kroc replied…
I don’t care whose fault it is, I’m just tired of being teased with excellent, potentially paradigm changing technology no one is willing to sell me. Either start selling or shut the #### up, is what I say. At this point like the fox in Aesop’s fables I’ve decided these grapes are sour. ;(
–bornagainpenguin
It is quite possible that an ARM-based netbook will be vaporware for a few more months.
For one thing, the concept behind the word “netbook” is still being refined in the minds of the marketing people, the techno-press, and the users. Not much momentum will be gained, especially for a non-X86/Windows platform until an unified concept emerges.
In my context, working in an office with a strongly locked-down work computer, a “netbook” would be a relatively inexpensive device (~$200), GSM and/or WIMax capable so that it can connect to the world, with a decently sized screen, a workable keyboard, and an integrated cursor navigation (my personal preference is for the trackpoint rather than the touchpad). I have yet to be turned-out by the “touch screen” such as the iPhone/iThouch
Do I really care which combination of CPU and OS it would run on? Not really – as long as it can open attachments sent to me by friends and relatives and generated using Microsoft products, gets a decent browsing experience, and can deal with on-the-spot minor editing of photos or even short videos. Being able to read eBooks and eMagazines confortably would be nice too.
Be was quite ahead of its time when it funnelled its limited resources on BeIA. This was implemented in hardware by Compaq (iPAQ) and Sony (eVilla). Unfortunately, the concept never really caught on even though BeIA ran on a X86 based hardware. The costs, size of the unit, and fear of anything non-Windows might have been additive factors for non-adoption by the mass.
We will just have to see…..and hope for a strike of genius in finding the just-right combination of features.
you’re just oblivious to what is happening, that is all.
tegra 2 will ship more than 10 million units this year across more than 10 different products.
what did you think the apple tablet was going to use.
Luminair trolled
Are you kidding me? @_@;
I complain about vaporware ARM netbooks and other computing devices and your idea of an answer is to point to even more vaporware?
–bornagainpenguin
Buddy… Calm down. Looks like you have some personal problem with ARM. Or you have no idea how ARM working model is different from Intel. Other than this I dont see how you call these cortex series a vapourware. Or you have no idea what vapourware means…
Cortex-A9 is in the customers hands. ARM’s customers or not mobile users. ARM’s customers are companies who build SoC around it. Companies are using it already. So, it cannot be vapourware.
What about the Touch Book from Always Innovating?
‘course it don’t use the Cortex-A9, but at least it got ARM inside…
http://www.alwaysinnovating.com/touchbook/
Edited 2010-01-08 07:47 UTC
The Atom is a cheap, slow CPU released over a year ago. The voice in the video says “extremely competitive”, but to me, it looks like the Atom was noticeably faster in every case… not that either would be unacceptable for web browsing.
Well there are two factors, the ATOM used also the integrated graphics accelerator for direct rendering this gave it a speed boost probably the one you could see and the cortex was using unaccelerated frame buffers for pulling the same off.
Also the cortex was running at 1/3rd of the MHz the ATOM was running, give it the same speed the Cortex probably blows the ATOM out of the water despite the missing video acceleration (which you can get btw by various chipsets like the Tegra on Cortex)
The Tegra runs circles around anything Intel has to offer on the Atom side of things. It really is a wonder that you cannot see it more often than in a handful of devices, the Tegra blows everything else in ARM land out of the Water.
That may not be true but it is yet to be seen as ARM has a MALI GPU that will coincide with the a9 platform…it should be capable of matching tegra
The last link is probably the most interesting…
[official page at ARM]
http://www.arm.com/products/multimedia/graphics/mali_hardware.html
[has video]
http://www.engadget.com/2009/10/21/arm-shows-off-its-mali-mobile-pr…
[some speculation about proformance]
http://forum.beyond3d.com/showthread.php?t=48319
Well, did you listen to th video or read the article?
That intel processor is running over 3 times faster than the ARM they’re showing here, and is also offloading graphics to a GPU. That ARM is only clocked at 500Mhz (its not production silicon — ARM itself doesnt’t do silicon — nVidia’s Tegra 2 has an A-9 running at 1Ghz already, and I imagine speeds of 2Ghz are possible if you give up some ground on power consumption.) The ARM core was also rendering in software, so that fact that it was *only just* slower in all but one case (the one that was more significant was an image (ie rendering) -heavy site.) at 1/3rd the raw clockspeed of the competition and with the rendering overhead, is pretty amazing stuff.
When you start to think that we’ll be seeing dual and probably quad-core A-9 parts in the 1.5-2 Ghz range before too long, it starts too look like a real threat to intel.
I’ve been saying it since the netbook craze started — ARM is going to make inroads into the computer market starting with netbooks, then through ultra-lights, net-tops and “green-PCs”.
ARM has serious momentum now, and because ARM owns the mobile processor segment (PDAs, phones and nearly all small portable devices) and people are demanding “the full internet” in these devices, ARM is getting tons of serious software (Android, Flash, etc) — Once people get used to doing web-browsing and email and the small things they do every day on non-intel, non-windows devices the wintel stranglehold is going to start freeing up. There’s nothing Intel can do to stop ARM in the mobile space — the software legacy is all ARM code in that market, making them the entrenched player.
No other player has gone up against Intel working from the ground-up like ARM has — PPC, MIPS, SPARC, and even Itanium went head-to-head in either the server or home markets, and never achieved a consumer-oriented “home territory” in the way that ARM has.
If things go well for ARM, I wouldn’t be surprised if, in 10 years from know, the major decision point in buying a new PC will no longer be Intel vs AMD, but rather x86/x64 vs ARM.
Ravyne posted…
Could I interest you in some ocean-front property in Arizona? Or perhaps a nice bridge in New York? Cash only and in small bills please…
–bornagainpenguin
Hey, that could happen if the San Andreas fault line expands a little, then drops CA off into the Pacific.
I’d say it has a better chance of happening…
Yes, I watched the video. The Atom CPU used here is a relatively old, cheap, and slow CPU – and it performs significantly better than the ARM system. It takes some time before the ARM system begins to renders anything. The Atom is already known to be slower per clock than the Pentium 3/M and Core CPUs – 1.6 GHz Atom N270 performance is said to be similar to a 800-900 MHz Pentium M.
What CPU will be in retail Intel netbooks by the time any of these ARM netbooks reach retail? The next-generation Atom netbooks (with the Atom N450) appear to be already shipping…
My HTC HD2 has a 1GHz Snapdragon ARM CPU and it is more responsive than my wife’s poxy Aspire One 150 netbook.
But rendering to a 4.3″ display.
In the video, the Atom was ruuning at its standard 1.6 GHz and it was assisted by an Intel GPU. The dual-core Cortex A9 development board in comparison had no GPU to assist with rendering, and it was clocked at only 0.5 GHz.
So you are left to imagine what the performance of the dual-core ARM Cortex A9 system would be if you were to triple the clock speed and add a GPU. Alternatively, you could imagine what the Atom netbook would be like if you removed the GPU and divided the clock speed down by a factor of three.
Then compare.
Edited 2010-01-08 00:59 UTC
As I said, the Atom is a slow CPU – slower than a Core 2 Duo running at 600 MHz in at least one benchmark, and comparable to a 800-900 MHz Pentium M. In the video, the ARM system takes some time before it even begins to render – so I doubt the impact of GPU acceleration here. The Atom model (N270) that was probably used just became obsolete as well (with the N450 as its replacement).
No. I don’t want to compare hypothetical future systems; I want to compare systems that actually exist. I’d rather compare mass-produced systems available in retail, but there aren’t any such ARM netbooks yet.
I think you forgot that the Atom is running at 1600 MHz and the ARM at 500 MHz. I the ARM will probably run 800-1400 MHz when it comes out.
Also, the Atom has a video card with 2D acceleration, the ARM development board does not have this, but the video states any real incarnation will have one.
So this comparison can’t be taken too seriously.
I did not forget it. I mentioned that Atom is a cheap, slow CPU. It performs much fewer instructions per clock compared to most modern x86 CPUs; slower than a Core 2 Duo clocked at 600 MHz or a 900 MHz Celeron M.
And yet it seemed to take the ARM system some time before rendering even began, compared to the Atom system.
Last time I checked, AMD were on the x86 business, not Cortex. The linked article don’t mention AMD.
For record, current version:
“a video released by AMD gives a pretty convincing picture that the Cortex-A9 will have comparable performance to the Atom. Watch the video after the jump.”
Excuse the typo.
Impressive for a less that half processing power but some questions do stand.
1 – What is the configuration of those machines?
Tough the configurations should be the same, as ARM would not use optimized ARM against non-optimized Intel code
It should be reminded that also reduces the reference to the existing/absence of a graphic card (using VESA)
2 – what are the timings?
Yes they are similar. But the eye is not a good clue of what is happening when the starting point is missed.
Anyway… for a 500MHz CPU… it is impressive.
True, the result is less impressive.
But taking into account everything… It can be an interesting match.
These were in equal conditions
But those will not be so equal on the final products.
It would then be more interesting!
You all wonder why we don’t see Netbook with ARM CPU or that we get flashy demo at CES but get nothing else after?
It’s not from Intel and it can’t run Windows. If Intel where able to put pressure on OEM to not do business with AMD, they can do it with ARM too. And lets add Microsoft in the mix. Those ARM Netbook can’t run Windows and it’s loss revenue for MS. They too can put pressure on OEM to avoid ARM.
That’s just my guess….
Windows Embedded CE does run on ARM,
http://www.microsoft.com/windowsembedded/en-us/products/windowsce/d…
Yeah, because we all know that everybody use WinCE on a laptop…
Lets talk about a real OS please.
moondevil retorted…
Perhaps so, but that still suffers from having a lack of software apps available to it. Unless things have radically changed (always a possibility) it is still quite difficult to get software for WinCE compared to WinMo, and even assuming this difficulty has been surpassed you still won’t be able to do the whole ‘click, next, next, next, damn toolbars, next’ install users are used to. That and dealing with confusions between processors and compiled versions for those different processors…
Supposedly the versions of Linux that work on ARM do not have these issues (everything is in a repository) and any commercial offerings will presumably be working as hard as possible to create an app store and build up a community around the device. Supposedly. But as I said earlier, I no longer believe we will ever be able to buy these sorts of devices for general use. We’re just going to keep on being teased about them over and over…
–bornagainpenguin
That’s where .NET comes in for MS.
n.l.o posited…
Not sure I follow, could you please elaborate?
–bornagainpenguin
Microsoft could insist that all programs be written in .NET and compiled to their .NET virtual machine. They just have to write a JIT engine for ARM and then .NET programs could run on Windows CE ARM systems.
This does nothing to help against the millions of people who will try to run their favorite not-.NET program. In fact, it will cause more confusion, and may hurt Microsoft worse.
hufman explained…
Okay that makes some sense I guess, if they had done it several years ago, and had the ability to run on multiple architectures in mind as a feature from the start. Too bad they haven’t and it looks to be far too late to be making plans for such now. Oh wait…vaporware! What am I thinking…I was concerned over nothing, because there is never going to be a consumer version of these devices available for us mere mortals…
hufman noted…
Yeah, that’s what I was thinking. And like I said in a post before I saw this one, backwards compatibility has traditionally been a religion at Microsoft–losing it would be a real blow to all those people with older software that is unable to be updated–and no virtual machines are not an answer, especially on this type of mobile device!
–bornagainpenguin
PS: Thanks for answering my question.
They already have. Its called .NETCF 3.5 (compact framework). Windows mobile has this already and windows ce 6 has it i guess since it has silverlight.
Also I found out that if you use SharpDevelop to code your app and you use strictly .netcf controls you can compile it once, and that same executable would run on your desktop and winmo phone (provided the form is the same size as the screen then you end up with a scrolling form).
All Microsoft need to do is enforce .net programing on windows and encourage .netcf programing and they would be able to switch to any arch they want without needing to worry about the dual binary executables.
You can already run .Net code on Win CE with a desktop profile. I’ve done it. It’s a version of the Compact Framework, but porting desktop code to the Compact Framework is fairly straight forward. It’s the Win Mobile version of Compact Framework that is a PITA to port UI to.
That’s very GOOD.
You know, I would not be overly surprised if MS doesn’t do what Apple did/does and have their OS build and run on multiple platforms… that we just don’t know about.
Tuishimi speculated…
They probably do, but the difference with Microsoft and Apple is that Microsoft made such a religion out of backwards compatibility that to lose it now would cripple them. Indeed it is that backwards compatibility that has probably been at the heart of many of the issues we’ve been seeing over time. Trying to push Windows on a different architecture will mean losing access to all the applications the users really care about, and as well as they perform in some instances web apps just aren’t there yet for many people.
Now also try to consider how many applications of this sort are “abandonware” or were bought out by another corporation and no longer sold, or were one off corporate applications that the source is no longer available to, are classic games no longer worth the money to port… You lose all of these with the new port of Windows. The advantage Microsoft has traditionally enjoyed is gone.
Not so with Linux, since theoretically the open source nature of the applications that typically run on Linux makes it trivial to port those along with the entire operating system. Theoretically. In practice I imagine we’ll quickly discover there is all sorts of code unintentionally tied to a specific platform that will need to be reworked, but in theory this is a huge advantage to Linux, because users get access to a huge library of applications, the same ones they may already be using on their X86 desktops.
Also, look into Microsoft’s ports of WinNT to Alpha, MIPS and PowerPC processors–the reason they failed was the lack of software ports. Somehow I think Microsoft has to be factoring that into any attempts at moving their stack to a different architecture.
–bornagainpenguin
Not so with Linux, since theoretically the open source nature of the applications that typically run on Linux makes it trivial to port those along with the entire operating system. Theoretically. In practice I imagine we’ll quickly discover there is all sorts of code unintentionally tied to a specific platform that will need to be reworked, but in theory this is a huge advantage to Linux, because users get access to a huge library of applications, the same ones they may already be using on their X86 desktops.
There are already a few ARM distros, just as there are for PPC and several others. Most of the applications do compile just fine or require very minor fixes, usually to makefiles and such.
So, the software is already there, more or less, the next thing is for someone to make an actually good ARM distro and to populate the repositories with all the same packages as you’d find on an x86 distro. I have tried a few Linux distros on my iMac PPC and the distros were very unstable and the repositories actually lacked huge amounts of packages you’d normally find. I really do hope that once ARM machines actually do come out that any ARM distos would be a better job.
As for the article itself and the topic at hand (I don’t wish to make a separate comment): I have been fascinated by ARM processors already for a good while. They’ve got lots of potential, and me liking to think of myself as eco-friendly person I like the fact that they save energy so much. We still need someone to actually roll out some good, accessible hardware with an ARM processor in it. They’ve been touting such for years now and we’re still not getting any closer to them.
WereCatf pointed out…
You’d be surprised. Just because a package has been recompiled doesn’t mean anyone has tested it out to make certain it actually works. Some one has to A) Find those apps needing minor fixes, and B) take the time to fix them and pass the fixes back to the developers, then C) Test the application to be sure it actually works now…
Like I said theoretically Linux has the advantage here, but so far this is only theory because there haven’t been enough Linux based ARM devices in the hands of Joe Public to see how well this plays out in real life. As they say, once you make something idiot proof, they go out and invent a better idiot! ;P
WereCatf pointed out…
I’ve given up on them. As you said they’ve been touting the benefits for years now and we’ve not seen them available for purchase anywhere the average geek can get their hands on them. I’m calling vaporware from this point on.
–bornagainpenguin
Debian for ARM is a full Linux distribution.
I believe Ubuntu have taken a snapshot of Debian for ARM and made an Ubuntu for ARM (in a very similar way that ubuntu does the same thing for i386 and x86_64 architectures).
http://www.ubuntu.com/products/whatisubuntu/arm
You even get it from the same place:
http://releases.ubuntu.com/9.10/
– PC (Intel x86) desktop CD
– 64-bit PC (AMD64) desktop CD
– Marvell Dove desktop image
– Freescale i.MX51 desktop image
(The last two are different ARM boards).
Enjoy.
Edited 2010-01-08 09:16 UTC
Linux is already ported. Past history. I’m running 64-bit Arch Linux with KDE 4.3.4 right now, and every single application is running x68_64 natively. No 32-bit applications. However, the killer point is the Top 500 supercomputer list … it features all kinds of architectures. That list is about 90% Linux by now, isn’t it?
It isn’t applications so much as drivers. Microsoft simply doesn’t own the source code for the vast majority of hardware drivers for Windows. If they don’t have the source code, they can’t compile a new version of a driver for a different architecture.
Therefore, Windows on a new architecture wouldn’t have any drivers.
Linux is already ported. Past history. I’m running 64-bit Arch Linux with KDE 4.3.4 right now, and every single application is running x68_64 natively. No 32-bit applications
The discussion is about other architechtures, not x86 or x86_64… And having the kernel and basic userland ported doesn’t mean it’s useable by non-geeks. No, he meant that all the applications should be ported too and I agree with him.
They aren’t designed to run all applications. You really don’t want to run OpenOffice on one of these devices.
Instead of cheap ARM netbooks I’d rather see a drop in the price of smartphones. Mini-browsers for everyone.
There needs to be a port between i686 and x86_64, just the same as a port to other architectures, because i686 and x86-64 are not the same binaries. One needs to re-compile. This is a problem for Microsoft, for example, because Microsoft do not own the source code for many, many hardware drivers. Hence there is a seeming “problem” in moving to 64-bit Windows.
But it isn’t a problem, really, if you have the source code. I am already running Arch Linux x86_64 right now, and the Arch repository for x86_64 has almost every single package that the Arch repository for i686 has.
Porting to a different architecture entirely is not that much different to porting between i686 and x86_64. The executables for different architectures are different binaries … just the same.
Now there is some problem to be had in different architectures when it comes to the kernel and drivers, it must be said. Different architectures might have a different interrupt scheme, for example, or different memory management. These things are constrained to the kernel space.
But when it comes to “all the applications should be ported too and I agree with him” … yes indeed. My point is … they already are.
http://www.debian.org/ports/arm/
http://www.debian.org/releases/stable/arm/install.txt.en
http://en.wikipedia.org/wiki/Debian#Features
My bold.
That is not 25,000 packages distributed amongst the 12 supported architectures, BTW, that is that all 12 architectures each have the full set of 25,000 packages.
ARM is one of them. AFAIK, all 25,000 Debian packages are available for ARM.
Enjoy.
Edited 2010-01-08 08:59 UTC
lemur2 said…
I am not unaware of the Debian port to ARM (bearing in mind it is for an earlier processor version of the ARM, IIRC) or the Ubuntu ‘Ohh-me-too’ port of the same to more recent versions of the ARM. I know these exist. That isn’t my issue.
lemur2 said…
How well have those packages been tested? That is my point. I’ve encountered several packages now in Ubuntu (which is based on Debian so I hold the situation might occur equally across both distros) that were broken, and yet released as working because no one had taken the time to use them. Testing doesn’t always work–you have to use the applications in order to see if they work. Given the relative rarity of ARM processors in netbooks or “smartbooks” I somehow don’t think anyone has really had the chance to do so…
lemur2 said…
Enjoy what? There are no smartbooks available to the general public. This whole discussion is a farce because there will never be any smartbooks available because the ARM people would rather continue to sell to cellphone makers and no one cares about building a netbook around one.
I intend to continue referring to these “smartbooks” as vaporware until it is possible to walk into a big box store and pick one up as easily as you could stroll into one and buy a netbook today.
–bornagainpenguin
I have here a Nokia N900 smartphone running on the ARM Cortex-A8 chip which runs Maemo OS, which is very close to a traditional Linux distro. It uses X.org, GTK, Clutter, Qt, DBUS, HAL and all that sort of thing. There are a few Linux desktop apps available like Pidgin, Claws-mail and XChat. I think more will emerge but obviously apps need a bit of reworking for the small screen UI.
Well it’s not a netbook anyway, but I think it proves the feasibility of an ARM-powered desktop Linux experience. I also have the tiny SheevaPlug ARM computer, which runs the Ubuntu ARM port, and everything works very nicely. I currently use it for DVB-T capture. There is also the upcoming Boxee Box which apparently uses the nVidia Tegra system (ARM based). If you’re really dead set on an ARM netbook, you can order the Always Innovating Touchbook, but there is a backlog of orders ATM due to high demand. Another netbook that seems to run an ARM chip is the Datawind Ubisurfer, again running Linux. It is available here for example http://www.ebuyer.com/product/176955 where you will see it has some positive reviews.
I agree that there aren’t too many ARM netbooks around right now, and I’m not even convinced that there’s a massive consumer demand for machines like that which are unable to run Windows (and I’m a Linux user/fan btw). However, you’re painting it like it’s some crazy pie-in-the-sky concept, when in fact I think there’s plenty of evidence to support the feasibility of desktop/GUI Linux systems running on ARM devices. Well I guess we’ll see who’s right this time next year
There is not yet an ARM netbook with performance competitive with an Atom netbook. The UbiSurfer and Touchbook are both underpowered compared to a basic Atom system.
No but as the article says the Cortex-A9 is just coming onto the scene and THAT is supposed to provide processing horsepower more along the lines of what Atom can do.
But even with the current line of ARM offerings I think the performance can be sufficient for what is required for netbook-type uses. For example the N900 can multitask video playback, web browsing, email etc. with decent speed and responsiveness. It helps as well that a lot of video decoding and GUI animation bling can be offloaded to the GPU, so for some tasks the CPU processing limitations are not as important.
I also think that if the processing power is traded off for battery life then that would be attractive for many users.
At any rate I think it’s great to see this competition hotting up between Intel and ARM, what with AMD looking a bit shaky at the moment I was worried that things might start to stagnate somewhat.
bralkein posted…
I’m a believer! Now tell me where I can get one that will rival my eeepc 901 today. Must be able to run Ubuntu 8.04.3 smoothly, with the Gnome desktop and firefox as its browser. Must have at least 20GB of disk space. More would be nice but I won’t quibble. Oh and must get at least eight hours of battery life, my eeepc only gets about six these days but its also a year old at this point–when it was brand new its battery offered much more time.
Well? I’m waiting…
Also if you re-read my comment you’d see I was mostly responding to people who thought the listed 23,000 packages would run flawlessly without issues in the ARM architecture and my challenge mostly revolved around previous breakage in Ubuntu (debian based) packages that never did what they were supposed to do, even though the applications themselves opened and seemed to work, until the passage of time showed the application wasn’t performing its function. Until someone does some actual real-world usage of those packages I intend to take their functionality with a grain of salt.
Understand please, it isn’t that I don’t want an ARM netbook (smartbook is the name of a educational toy by vtech) or that I don’t think they can work–I’m just frustrated by my inability to see one in a store and buy it. We’ve been hearing about how good they can be and what wonders they can perform for awhile now–its time for ARM to start delivering on their promises to the average geek or shut up.
–bornagainpenguin
So you mean q1/q2 of 2010 then?
http://en.wikipedia.org/wiki/Smartbooks
http://www.liliputing.com/2009/12/ready-or-not-2010-could-be-the-ye…
http://english.etnews.co.kr/news/detail.html?id=200912280010
Enjoy.
Edited 2010-01-09 06:48 UTC
lemur2 posted…
http://www.liliputing.com/2009/12/ready-or-not-2010-could-be-the-ye…
♫ Jam tomorrow, jam yesterday, but never ever jam today… ♫
It seems we’ve heard this song and dance before. Wasn’t it supposed to be last year we’d be seeing these types of “Smartbook” devices released? Also I wonder what vtech, manufacturer of toy laptops for educational games for preschoolers thinks about the name, since they are already using it…
–bornagainpenguin
Edited 2010-01-10 05:21 UTC
Drivers are not really an issue, if you are creating hardware specifically to run an OS then you can create drivers at the same time (ie look at Apple).
Ofcourse, you will have very limited support for any *other* hardware you try to use with it such as printers. This is why standards in hardware are so important, keyboards/mice can be driven by standard drivers for instance.
Drivers are indeed an issue if you don’t have the source code.
If only hardware OEMs have the source code, and they no longer sell a given model of hardware device, there is absolutely ZERO incentive for the hardware OEM to release a new bianry driver for a new architecture and/or OS version merely for some software supplier company to profit from.
Hence, when people buy a new Windos 7 laptop, they will typically find that their old-but-still-perfectly-working printer doesn’t work (as an example that recently happened to friends of mine). They pop in the driver CD that came with the printer … only to be told that Windows 7 won’t work with an old 32-bit driver.
In Linux, this isn’t a problem at all. Linux has the source code, and the printer drivers run in userland. Unlike Windows, Linux has relatively few different printer drivers per se, and the vast array of real-world printer models are handled by plain text configuration files.
If a printer works on 32-bit Linux (i686 or i386), then it will also work on x86_64 Linux and on a number of other architectures as well. Debian supports 12 different architectures. All 12 architectures work with the exact same set of printers.
Remember … Linux has the source code.
PS: On Windows, network printers are also a problem. I have seen it happen that when one connects a new network printer on Windows, Windows tries to download binary driver dll files from the printer server machine.
This jsut isn’t going to fly if Windows moves to multiple architectures.
Edited 2010-01-08 09:53 UTC
Linux apps are in a much better state than you think, there is very little code that doesn’t compile and run on non x86 processors.
Linux had the same advantage when x86-64 came out, because linux had been running on 64bit alpha processors for several years most issues with 64bit bugs were already long fixed.
You will find that fully operational distributions with full repositories of applications have been available for arm processors for quite some time, look at the debian repositories.
Shouldn’t be that big a problem. NT was designed to be portable. It currently supports two and a half CPU architectures (x86, Itanium and x86-64) and in the NT 4 days, NT supported four different architectures. The bigger issue would probably be applications, as various folks have mentioned. Still given that Alpha-native versions of Word and Excel were available for the entire run of NT on Alpha (thus up through Word/Excel 97), it is quite likely that MS could create ARM-native versions of its apps if ARM ended up growing in popularity in the desktop/laptop market.
Thus I suspect that it is more Intel saying “we’ll raise the price we make you pay full retail price for our chips in your high-end boxen if you move to ARM in your low-end gadgets” than resistance from Microsoft that is holding back ARM.
Edited 2010-01-08 18:31 UTC
The video is a nice demonstration of what the cortex can do. I wonder though, why did they want to compare there dev board to a production machine with accelerated video? This only makes there product look inferior.
Surely there could have been better cpu bound benchmarks that would have showed off the capabilities better. I do not mean to be negative however this seems like shooting yourself in the foot. A comparison of how many hours of dvd decoding on the same battery, for example would have been more interesting to me.
They showed graphics for the same reason it’s hard for the software verification researchers to compete with the graphics researchers at a poster session.
That would be a piss poor way to handle it, to me. Anybody can make a CPU that does well with high-throughput tasks. The trick is making it good as a desktop CPU, where random things happen all the time, with low IPC, cache misses all over the place, but still being responsive.
A specific <piece of hardware>-bound benchmark is exactly what should not be used, as my real-world usage is going to be nothing like that.
A picture and short, easy to understand sentences are worth their weight in gold. I got out of this video:
1. Nearly as fast without a GPU as an established product with a GPU.
2. Running at 1/3 power of the competitor.
I say WOW. Place that into some fine hardware combined with a fine GPU and you should be smokin’, while saving battery power.
A video released by ARM.
This is ARM folks, not AMD.
Good gawd.
Edited 2010-01-08 01:56 UTC
Comparing two CPUs by their respective clock speed is only useful when the two CPUs are identical, which isn’t the case. Moreover, RISC targets require smarter compilers and the x86 GCC target is likely to produce better assembly than the ARM one just for being a more popular platform for desktop computing. So, there’s no way to extrapolate that the Cortex A9 is running “at one third the power of the Atom” just by comparing the clock speeds. Of course, we also need to keep in mind the power consumption of both processors, likely to benefit the Cortex in most scenarios.
In any case, good times are coming: This, together with China’s MIPS-based projects, should bring some needed competition to the CPU market. And that’s always good from a consumer standpoint.
Edit: Clarifications.
Edited 2010-01-08 04:07 UTC
Even comparing the same processors running at different clock rates is not really valid either, increasing the clock rate of a cpu has diminishing returns if you don’t similarly increase the clock of everything else in the machine…
A higher clocked cpu with the same bus rate will spend more time waiting for the cache to be filled with data.
Agreed. Even the speed differences shown in the video could be caused by a slower bus on the Cortex.
What’ll Apple tablets run? ARM (as iPhone) (hmm, A9+MALI or A9+SGX?) or Atom (as Apple work closely with Intel on laptops/desktops/servers)?
And what’ll the first Chrome OS tablets run? Atom or ARM again?
I only know some of the answers with some certainty.
will be
ARM is not 64 bit. Remember you need a 64 bit processor to run Notepad/play solitaire/etc
“The Cortex-A9 is promised to have substantially better performance than the current crop of AMD processors, and a video released by AMD gives a pretty convincing picture that the Cortex-A9 will have comparable performance to the Atom. ”
AMD or ARM ? Sorry if I missed something
I don’t understand the article author. The article feels like – look ma, ARM tries to beat Atom, but it hardly can. What is the point in the comparison of ARM not paired to a GPU?
What about showing following video instead?
http://www.youtube.com/watch?v=iikt3GgJJXc&feature=related
Well, if we look at the ARM Cortex A9 effort from Globalfoundries
http://www.xbitlabs.com/news/other/display/20091006045344_Globalfou…
it would seem that AMD/ATI have an iron in this fire also. The difference is, AMD/ATI GPUs now have a working (including OpenGL, DRI and 3D) open source driver for Linux.
They will have the CPU chip, the GPU, the open source OS and the open source drivers all in place.
I don’t think this will ever be true of Tegra.
Edited 2010-01-08 10:41 UTC
I’m really glad to see new ARM CPU’s. This is a good thing against Intel’s monopoly and it seems like ARMs are really powerfull. Hopefully my dream about powerfull and fanless mini computers compatible with x86 instructions [not some palmtop toys] will come true. Go ARM!
Compatible with x86 instructions???
As I see it, there is a big irony here, especially when people tout the Linux side of this all.
If ARM is to reach the consumer PC market en masse, this probably means that the Windows ecosystem needs to run on ARM, which is of course not impossible at all.
Since closed-source Windows applications are distributed as binary only, there is no point in porting Windows to ARM, because it still wouldn’t run any of the Windows sofwtare that is already available … because that is all x86 binaries.
The existing availability of ARM binaries for Windows is near zero (not absolutely zero, as there is a small amount of stuff for Windows CE on ARM).
However … most people wouldn’t even know what x86 meant. They would be told of a place on the Internet where they could get applications for their smartbook, a bit like the iPhone App store, and there would be say 10,000 applications they could get. They would be pleasantly surprised that these were all free, and dead easy to install or remove as they felt like, with no restrictions.
No problemo.
Buisnesses can give their mobile staff a cheap smartbook each, with a full set of applications pre-installed (like a super-blackberry-come-video-phone if you will), and allow said staff to add whatever else they wanted from the online repositories, without ever having a single worry about anything like this happening:
http://www.eweekeurope.co.uk/news/software-auditors-crack-down-as-r…
Or perhaps businesses could hand out to staff one of these tablets devices instead:
http://www.pcmag.com/article2/0,2817,2357854,00.asp
Edited 2010-01-10 07:22 UTC