OLPC XO-2 Laptop to Use ARM Processors

   
   
   
   
   

Yes, I called it two months ago. I said OLPC is looking for a non-X86 architecture for the XO-2 laptop, probably ARM, where several providers can provide the processor. And look what Agam Shah says:

One Laptop Per Child is set to dump x86 processors, instead opting to put low-power Arm-based processors in its next-generation XO-2 laptop with the aim of improving battery life.

This is fantastic news for the current best ARM processor providers which are Texas Instruments, Qualcomm, Freescale, Samsung, Broadcom, Nvidia and other ARM licensees can all now work with OLPC with a clear goal to produce totally workable ARM based laptops reducing power consumption, reducing cost and optimizing the features so what comes out is a device that works very reliably and is even easier to use.

Texas Instruments has announced 45nm process to be used for Cortex A8 processor later this year, and the Texas Instruments OMAP4 platform that is due to come out next year will feature ARM Cortex A9 technology which supports multiple cores. Surely some of the other ARM processor providers have the same types of plans for the current and the upcoming ARM Cortex based solution to be released in the coming months.

Intel surely feels this one right in their teeth. Look at the latest round of Netbook designs, they have all added a layer of bloat (new gimmicks) to keep prices of Netbooks around or above the $400 for another year. Shrinking profit margins means the market has to change completely. Focus on larger volumes, more focused functionality, longer battery life and lower margins. OLPC again can push the industry in the optimal direction.

Google needs to help officially port Android to ARM laptops, call it Android Laptop edition or something and have it be a goal that it should run Google Chrome type of browser with all Google Gears and necessary multimedia perfectly. This is basically all that people need. You can add support for Adobe Air and Microsoft Silverlight if you want. The basic idea is to make sure that there is a perfect platform for the most basic Internet Applications access.

Sure thing, officially asking Microsoft to make Windows 7 for ARM Laptops is an amazing feat. I don't know how Windows really works, but it looks great that Microsoft seems to have learned the lesson after Vista, that more bloat and higher hardware requirements is not the way forward. That Microsoft definitely needs to slim down and optimize their OS to work on cheaper hardware if they want to stay in the game.


XO-2 going ARM over x86

All those ARM chip providers should announce their support of the OLPC XO-2 project and they could start right away showing whatever prototypes that they have ready running experimentally builds of different types of embedded Linux be it the upcoming Ubuntu for ARM, Angstrom Linux or even best running adapted Android Linux implementations.

Then current market leaders in the Netbook segment can one after the other announce that they too will be ready to be part of the ARM paradigm change. Whoever of an interchangeable mix among Asus, MSI, Acer, Gigabyte, Hasee, Quanta and the other giants need to come forward and confirm that yes, they do have working prototypes of the ARM laptops using some of the different available options and that they will be interested to use XO-2 technologies as soon as possible.

I'm hoping someone could confirm if some type of Mobile WiMax on 700mhz white spaces spectrum would be possible, including even some type of Mesh networking on those networks to even extend those networks and make them cheaper to build in a peer-to-peer decentralized fashion.

Base stations for the future of wireless broadband networks should be setup by anyone just as anyone can setup a WiFi router in their homes. Future WiFi routers should beam out hotspots for their whole neighborhoods using those Mobile WiMax on 700mhz TV Spectrum kind of technologies. Reserving all the 700mhz spectrum worldwide for private TV broadcast monopolies does not make sense.

Finally, Barack Obama and other world leaders need to embrace XO-2 on many levels. Not just the idea of using technology to improve education worldwide, but also to optimize the way the whole society uses technology. Cause when you look at it, 99% of the energy consumption for computers, laptops and data servers goes to waste on empty cycles and bloatware. If we can regulate the use of energy for PC and Laptops, we can also save the environment in the process, now that it should be a priority to connect the remaining 5 billion people to the Internet.


Stay on top of the XO-2 laptop: subscribe to OLPC News via RSS Feed, Emails or Twitter.


Related Entries

20 Comments

The basic idea is to make sure that there is a perfect platform for the most basic Internet Applications access.

GNU/Linux as it is struggles at the best of times to provide compatibility for basic web technologies. The most likely to succeed GNU/Linux Flash experience is on x86 machines running Adobe's plugin. Unfortunately Flash is used extensively on the web, and Gnash is a piece of garbage.

Also, you dump all over MS for "bloat". One of the biggest causes of wasted CPU cycles? FLASH! What runs poorly on older computer? FLASH! Normally when a web browser is open and a CPU core is pegged at 100%, flash is to blame.

If these things can be bought very cheap (

Sure thing, officially asking Microsoft to make Windows 7 for ARM Laptops is an amazing feat. I don't know how Windows really works, but it looks great that Microsoft seems to have learned the lesson after Vista, that more bloat and higher hardware requirements is not the way forward

Windows 7 won't run on ARM, and certainly wouldn't be able to run x86 Windows applications without emulation. ARM chips are low in power in both electrical, and performance.

Microsoft isn't the only one adding "bloat". The requirements for mainstream desktop GNU/Linux distros is slowly creaping upwards. OS X requirements creep upward. As it is, 2GB of a Mac OS install is printer drivers and another 2GB is languages. Most of which aren't used by most people.

Whoever of an interchangeable mix among Asus, MSI, Acer, Gigabyte, Hasee, Quanta and the other giants need to come forward and confirm that yes, they do have working prototypes of the ARM laptops using some of the different available options and that they will be interested to use XO-2 technologies as soon as possible.

Just like how people came clamouring for XO-1 technologies? Oh that's right, other companies blew past OLPC in no time, as OLPC continued to have broken "hi-tech innovations".

Also all we've seen from OLPC is a cardboard mockup of a dual screen unit. Asus has had dual screen test models in the real world.

I've seen Asus dual screen test model thing last week, it is basically a mockup (shown only after OLPC announced the design of the XO-2) of two touchscreens connected to a computer doing all the work under the table.

Adobe has announced that full Flash 10 will be supported on ARM Cortex processors within the next few months. There are already some products like the Archos 5 and Archos 605 series that are ARM based and support full Youtube Flash full screen support using the multimedia core of the processor. Basically the workaround is pulling out the video file from the Flash stream and display that using the multimedia DSP core, and not trying to decode video frames using the ARM processor itself. Adobe was basically forced by public pressure to provide full Flash support for ARM processors soon. (until then, I would say, it might have been under some kind of conspiracy theory that flash support has been weak and late on open Linux platforms especially for embedded Linux platforms)

Most of the flash used on the Internet is for video playback.

I am sure that Microsoft could make Windows 7 or Windows 8 for ARM processors. It's just a question of them being forced to do it. Just like Microsoft was forced by OLPC to delay the end of Windows XP and to massively lower the cost of licencing for Windows XP on cheap laptops to third world countries.

OLPC has in fact created the Netbook segment. If the leading Netbook makers didn't use many of the OLPC XO-1 technologies, that kind of was cause XO-1 was built like an Apple product even though even the XO-1 hardware mostly is open source. Still the XO-1 manufacturing and all that had specific exclusive manufacturers for the different parts and thus it has so far been too hard for the whole market to use the Pixel Qi screen, the WiFi Mesh networking or the DCON main processor shut-down and WiFi packet forwarding on standby features. This is how XO-2 project will change, that OLPC will focus more on making the design fully copiable by the whole market. And not just remain an inspiration for the market, not just remain a movement that pushed Intel around. But instead OLPC will become a sort of a consortium of companies that can all work on the same open hardware and software goals. And the result being much cheaper, better laptops, faster.

@Charbax:
"I am sure that Microsoft could make Windows 7 or Windows 8 for ARM processors. "

Nope.

To have a portable OS you need:
1 Modularization
2 Kernel developers who understand the kernel
3 Kernel developers who understand the target architecture
4 Driver developers

If you want to do it for profit:
5 A market that pays

Of all these requirements, MS only has money and Windows CE. I cannot see how they could compete with Linux, which is fully developed on Arm.

1 The Windows kernel seems to be the opposite of modular, with even the web browser interfering in kernel space. Windows is simply welded to x86 interrupts and instructions on all levels. Personally, I think Windows HAL is a joke.

2 MS have hemorrhaged good programmers and developers. Reports have told us for a decade that MS developers hardly understand the code they are working on. It is layers of cruft over legacy code that is held in place because nobody dares to change it. Most applications (Office) are reported to be completely dependent on the way the compiler lays-out memory (hence the enormous delays for Mac Office). For Arm they would need to develop a new compiler.

3 There would be hardly any Windows kernel developers who have Arm experience, nor can we expect many Arm developers who have Windows kernel experience. Windows CE is a different operating system. With the MS "need to know" policy, it will be difficult to "share" all that information.

4 MS were almost unable to write the drivers for the x86 XO-1. They do not write drivers, they have them written by others. Now they must come up with drivers for ALL hardware on an Arm architecture.

5 Windows-on-Arm would cost several king's ransoms. And that for a market with no buying power. XP-on-netbooks has a margin of dollars ($3?).

Undoubtedly, there will be a Windows 8 for Arm. But that will be WinCE with a Windows API. It won't run Windows software and will share nothing with the x86 version but it's name. But it will be there and forced on the netbooks irrespective of quality or user preferences.

Btw, I have never used WinCE. Is it really as bad I read in the reports?

@John:
Why do school children in the developing world absolutely, definitely need Flash? To watch YouTube? Can nothing useful be done on a computer without YouTube?

YouTube did not chose Flash because it was the only option, or even the best consumer choice. Any other format would have been more convenient for the user.

I do not see why the OLPC should hold the children in the developing world ransom to YouTube.

Winter

The Windows kernel seems to be the opposite of modular, with even the web browser interfering in kernel space. Windows is simply welded to x86 interrupts and instructions on all levels. Personally, I think Windows HAL is a joke.

If you think Windows is full of cruft and welded on parts, have a look at GNU/Linux. It still uses device names, and parts that made sense 30 years ago and are completely irrelevant today. Programs are so dependant on tons of libraries that you must hope that an application is available in your distro's package manager, because GNU/Linux has brought DLL hell to a whole new level known as "dependency hell".

X-Window is the biggest weld on cruft I've ever seen. Especially with Gnome or KDE further welded on top. X-Window has a place in the config for the resolution. But Gnome has another. Ever seen how GNU/Linux handles auto-mounting of drives? Very rickety at best.

Most applications (Office) are reported to be completely dependent on the way the compiler lays-out memory (hence the enormous delays for Mac Office). For Arm they would need to develop a new compiler.

Have you seen how the FOSS savior Open Office lays-out memory? It just uses all of the system memory for no apparent reason.

Yes they would have to develop a compiler, etc. You talk as if Microsoft has never ported their OS to other platforms. NT was available for a handful of machines including DEC Alpha. Most modern Windows versions are available not only for x86, and x86-64, but also Itanium. Itanium is a completely different platform.

However just because they could, they probably won't port Windows 7 to ARM. I assume to keep a low cost point they will have 256MB or less of RAM, and small SSDs, which just doesn't make the Win7 experience possible. But the big thing is all the programs out there for Windows desktops won't work on it, so there's no incentive. I think it's a good move going ARM and they should have done it before.

Btw, I have never used WinCE. Is it really as bad I read in the reports?

It's not terrible, but it's not very good, and there aren't a whole boatload of applications. This is a market space that GNU/Linux could gain in.

I do not see why the OLPC should hold the children in the developing world ransom to YouTube.

The original article talked about making it "perfect platform for the most basic Internet Applications access."

Unfortunately Flash classifies as that. Also, right now OLPC relies on developers developing in the relatively obscure Sugar environment. If you allow the thing to read flash, java, javascript, etc. then people can develop for web environment which is relatively known and familiar for developers. There are a number of online educational activities that use these technologies.

@John:
"If you think Windows is full of cruft and welded on parts, have a look at GNU/Linux. "

But it IS modular and you can remove all the cruft you dislike. Actually, Linux runs both my TV set and my wireless router, as well as phones, "real" computers, CERN and Google search. Except for the desktop computers, I have nothing at home that could run XP.

Moreover, Linux does run VERY nicely on ARM.

@John:
"Have you seen how the FOSS savior Open Office lays-out memory? It just uses all of the system memory for no apparent reason."

But OO.o file formats do not depend on the compiler memory lay-out as MS Office does. And the XO uses Abiword, not OO.o. So this is a complete non-sequitur.

@John:
"NT was available for a handful of machines including DEC Alpha. Most modern Windows versions are available not only for x86, and x86-64, but also Itanium. Itanium is a completely different platform."

They couldn't manage MIPS, and Alpha was only temporarily available as a result of one of many MS' illegal actions. It was part of a settlement for wholesale infringing DEC copyrights when developing NT.

Still no ARM, MIPS, PowerPC etc. Itanic is just another Intel CPU with legacy mode to support bloated software. There will certainly be no netbooks using the Itanic.

@John:
"But the big thing is all the programs out there for Windows desktops won't work on it, so there's no incentive."

My point completely.

@John:
"If you allow the thing to read flash, java, javascript, etc. then people can develop for web environment which is relatively known and familiar for developers. There are a number of online educational activities that use these technologies."

Logical and indeed commendable, although I would scrap Java from the list. But at that level, they do not need to stick to the latest Adobe Flash.

Winter

Finally, Barack Obama and other world leaders need to embrace XO-2 on many levels. Not just the idea of using technology to improve education worldwide, but also to optimize the way the whole society uses technology. Cause when you look at it, 99% of the energy consumption for computers, laptops and data servers goes to waste on empty cycles and bloatware. If we can regulate the use of energy for PC and Laptops, we can also save the environment in the process

Regulate the use of energy consumption for computers? Seriously? I think this needs some clarification. The way I read that is: "We need the government to restrict how and when we can use our computers."

Most estimates say computing technology (residential + industrial) consumes around 3-5% of the nation's electricity - and those stats are from 2001. By now its much greater.

In fact, California estimates that jumbo flat screens, which due to their size, consume more electricity than CRT's, will spike electricity use by 3% in that state in 2008 alone. So they're trying to get flat panels to use less electricity through efficiency standards.

That doesn't read "how and when" that says use whatever you want, just make sure "it" is energy efficient.

The regulation could be for the power consumed to do certain browsing tasks. Power consumption for standby modes, power consumption for the wireless networks.

If you want to buy power consuming Intel products, then fine, you may, but I think people should pay more taxes on those compared to optimized ARM laptops that consume much less power.

Basically just as washing machines that use a lot of water should have a higher tax rate than washing machines that save water. Same thing should be ior even in many countries already is about the taxation of the polution level of cars.

Same thing about the effeciency of data servers and cloud computing networks. Increased incentives to the manufacturing of lower power Laptops and computers, with more environmentally friendly materials will help speed things up for this change to lower cost laptops worldwide.

I agree that power efficiency standards should be applied to computers, and electronics in general. Data centers use a non-insignificant amount of power. And if you drop the computer heat load, you also drop A/C loads.

First place to start is the out of control amount of power consumed when a device is "off". I've seen computers use 5 Watts when "off". ACPI, WoL, etc shouldn't need that much. My cable box uses 14W when it is on. When powered off it uses... 14 Watts.

Next, look at technologies like Speed step which underclock the processor. First available only on laptops, but are only now becoming available on desktop computers. Why? They should be standard on everything.

Carrying the theme even further, even an underclocked performance processor uses a lot of juice. How about having a multi-core processor where some cores are optimized for low power consumption (eg: Atom) and other more powerful cores are available for peak loads. When the computer is just idling along, as most do most of the time, the computer uses more power efficient cores.

As an example, my EeePC uses 12-14 watts when idling, browsing, etc. An external 3.5" harddrive uses 10 Watts. A desktop computer then should not exceed 24W when idling. Many use at least 60W.

Y'know, it's extremely nontrivial just to have a jump in computer performance every generation. Government enforced power consumption standards aren't going to help the field progress, especially since we're gradually starting to flatline on GHz.

As with code, it's far more reasonable for hardware manufacturers to get something new and shiny that works, get it out to consumers for continued money, THEN optimize it for resource (power) consumption.

Look at netbooks. The makers of Atom and ION are already taking existing technology and optimizing it for lower power usage, and these chips can only get more popular. The portable market drivers efficiency, and with enough volume, soon even cheap desktop machines will be using netbook technology. This is all happening naturally without regulation.

Add into this the fact that most machines only draw as much power as a lightbulb or two, and it starts looking like it would be a large mistake for the government to start trying to regulate this.

But it would be irresponsible to completely ignore power consumption, especially since computers use a non trivial percentage of total power used. One only needs to look at cars and see what happens without fuel economy or emission standards.

Some things like computer standby power wouldn't hamper performance, but will help when people turn their computer off, to know it's actually off.

And look at companies like WD who are trying to make lower power consumption hard drives (and other companies with SSDs). There's a lot of spinning platters in data centers, and a drop in component power could spell big savings.

jose mentioned power draw of large screens. You can't go solely based on what the tag says. My 40" Sony LCD has a nameplate of 200W, but uses 150W when actually running, and 7 watts when "powered off"

You are partly wrong and partly right. As to the market needs glam, yes, truth its easier to sell a common product if its rewrapped into another shiny paper.

As to the breakthroughs in power-efficiency, You are either skipping the truth on purpose or You could get more info on arms. The story goes like this, yes on intel platform the current clock-rates are the max the market can get, and yes again no chance for more power-efficient computing with nvidia or amd or intel gfx and intel cpus, however the less known truth is, the x86 itself in idle with gfx card on eats like 5-10x more power than the most power efficient arm cpu avail on the market just today.

What I can foresee and I do believe strongly in this is either x86 cuts off the anchor of the past or arm crushes x86's market in half (more seriously its gonna be 30/70 for arm at first with possible 70/30 by the end). If energy and performance mean something to You support the arm architecture with the consumer choices and spreading the good word.

We've been discussing this possiblity for a few months over at the community forum

http://www.olpcnews.com/forum/index.php?topic=4088.0

As to Microsoft porting Windows 7, I don't know much about how possible it is. However, later this year a whole stack sub-$200 ARM netbooks is going to be coming out. I don't see how Microsoft can leave a whole new market segment to Linux and Android, so it is going to be under a lot of pressure to come up with something for ARM.

@eduardo montez:
"I don't see how Microsoft can leave a whole new market segment to Linux and Android, so it is going to be under a lot of pressure to come up with something for ARM."

Yes, and it will be WinCE (under any other name).

Expect some heavy handed marketing and many "infomericals" and "editomercials" singing the praise of WinCE against that pirate copycat Linux.

Winter

"Yes, and it will be WinCE (under any other name).

Expect some heavy handed marketing and many "infomericals" and "editomercials" singing the praise of WinCE against that pirate copycat Linux."

I wouldn't be at all surprised if you are right about this. However, I don't think it would get very far. The big selling point for Windows is all the applications that run on it, and so I suppose the Microsoft commercials would try to subtly trick people into thinking that CE can do that.

However, people will be buying their ARM netbooks from oems and retailers, and they would get in big trouble (returns, lawsuits, FTC action) if they didn't make clear that Windows CE doesn't run desktop Windows applications.

I just read "OLPC Set to Dump X86 for Arm Chips in XO-2"

http://www.pcworld.com/article/161112/olpc_set_to_dump_x86_for_arm_chips_in_xo2.html

What strikes me is the way the OLPC just tells the world
"We want to support MS Windows, but they cannot deliver".

I cannot remember the last time I heard this from a hardware maker.

The times they are a changing.

This is also a great way to say to MS "We will not support you!".

MS have now a dual boot promise on obsolete OLPC hardware, and no support on the new hardware. Maybe the OLPC did not really sell out?

Winter

Hey, I still have a 9 year old ARM Ipaq Pocket PC. Works great. But you're right, not much programming available for it.

As a programmer I would love the X0-2 with ARM if only:
+They include floating point coprocessors, so it frees the programmer from the fixed point hell.

About electronics power: last time I went to the electronics store they had 600euros big screens that consume(you can look the back side electrical power) more than 400Watts!!!

That's it, every time you watch TV you are wasting a lot. Nobody seems to care. With desktop, 100-200Watts graphic cards and bigbig screens seems to be what people want.

jose, It depends which one they will choose, cortex without the fancy dandy 3d imgtec logic shall be fine, although its still by $30 or so more than arm926, I think personally they should go for cortex based in first run, while working on the removal of the fp code from software to deliver the last (proper) version of xo2 based on arm9. But thats just my humble wish.

As to the gfx cards, there is really no choice (pretty much same story as with intel). For now the times are changing, so hopefully in 1 year or so it will be much easier to dump the intel and nvidia off the closest cliff.

Charbax,

If this isn't a sign of the times:

"In addition to the previously supported i386 (32-bit x86) and AMD64 (64-bit x86) ISAs, Ubuntu 9.04 will now natively include an ARM installation for ARM-based MIDs and low-end netbooks operating on the ARMv5EL and ARMv6EL-VFP architectures."

http://www.geek.com/articles/mobile/ubuntu-904-due-april-23-include-native-arm-port-20090414/

Winter

XO Tablets for Sale

Buy Your XO Tablet on Amazon.com
OLPC is selling the new XO Tablets on Amazon.com for just $149. Buy yours today!

xo-tablet-amazon.jpg

Discussions

Recent Comments

Community Forum

Close