We can safely say that the Apple iPad is received with mixed feelings by the IT-savvy community. The main problem is that the tablet is just not as revolutionary as many expected it would be. It keeps many of the limitations of the iPhone (no multitasking, tightly controlled app-store) and doesn’t provide impressive new features which could keep the balance.
Let’s compare this tablet to one of the more promising Tegra 2 tablets on the way to the market: the Notion Ink Adam. Admittedly, the Adam is not on the market yet, while the iPad is quite sure gets there soon. Nevertheless, we give the benefit of the doubt to Notion Ink (especially, considering how badly Nvidia wants to start the Tegra 2 device line). In my comparison, the Adam runs Ubuntu Linux with the Mobile Edition which has a touch oriented user interface. The first version of the Adam is expected to come with Android but since Nvidia officially supports Ubuntu on the Tegra 2, we can expect a fully working Ubuntu edition on the Adam soon enough. Moreover, for such powerful hardware as the Adam, even Android seems limited to me.
Screen technology and diplay features
Both the iPad and the Adam have ~10″ LCD displays with capacitive touchscreen technology. However, the Adam features PixelQi technology which means lower power consumption, direct sunlight readability and higher contrast when switched to BW mode for reading. The Adam will have a hardware switch for easy mode-changing (BW/low-power colour/full colour) which will help lowering the power consuption (like a wifi or 3G radio shut-off switch). The iPad doesn’t seem to have any hw switch, not even one for adjusting the brightness of the screen (meaning: unless there is an ultra-easy touch gesture for it, nobody will adjust the brigthness for ambient light conditions). The iPad specs page lists an ambient light sensor, so the OS may be able to automatically turn-down the backlight when not needed but this won’t save you as much as the power saving modes of the PixelQi screen. All-in-all, the screen of the Adam seems to be more versatile and power efficient.
The custom Apple A4 processor of the iPad runs at 1Ghz and is a single-core ARM Cortex A9 solution. The Tegra 2 of the Adam features 2 Cortex A9 cores running at 1 Ghz. Due to the exact same technology platform, I expect the Adam almost two times as powerful as the iPad and this should be very much noticable in the more important applications (e.g. web-browser). The iPad has the stock ARM Mali 50 graphics core, while the Tegra 2 includes a Geforce 9 level graphics core. Although, I am not very familiar with the capabilities of these cores, I expect the Nvidia core to be more powerful since this is where Nvidia has strong competency. Some say the Mali core is not even in the same league as the Tegra 2 graphics core but this remains to be seen. (Disclaimer: it is possible that current reports about the inclusion of the Mali are inaccurate and the iPad uses a PowerVR graphics core like the iPhone).
The iPad runs the GUI environment of the iPhone OS (an OS X derivative), while the Adam runs Ubuntu Mobile (at least in my comparison). The Adam may not support multitouch in the short term but already has support for gestures (e.g: a swipe for moving to the next image in the image browser). The iPad supports multitouch and a wide array of gestures. I expect the iPad GUI easier to use and more refined (at least for the time being, since Ubuntu Mobile is quite a young project).
The “desktop” is easy to use in both environment, very similar application startup and indicators. The Ubuntu OS will run several applications in parallel while you will be able to use only one app ata time on the iPad. (see about this later)
The iPad doesn’t run Flash but said to render normal webpages snappily in its custom Safari browser. Youtube is supported just like on the iPhone but no Flash games and no Java applets in webpages (see about interpreters later in the application section).
The Adam will run full editions of Firefox/Chrome/Opera and expected to have an optimized Flash version (Flash 10.1 coming soon) so it will be good for running web-pages with video streaming and Flash games. Java applets (rare nowadays) will work too.
Playing video / Listening to music
The iPad has no HDMI output (has a simple VGA output, max res: 1024×768) and is rated for decoding 720p videos. This is a far cry from the 3 simultanious 1080p streams of the Tegra 2 and built-in HDMI port of the Adam. The Tegra 2/ Adam offering is far more powerful and makes the Adam a viable HTPC if you want to play the movies from your tablet onto your HDTV screen.
Both have a 3.5 mm jack and speakers so hw-wise, listening to music should not be a problem. Ubuntu has powerful music player applications which are on par with iTunes of the iPhone/iPad.
The Geforce graphics core of the Tegra 2 in the Adam will be quite sufficient to play 3D games. The performance of the Mali 50 is not widely known at this point but is not expected to be worse than the graphics core in the iPhone 3GS so it is likely able to run 3G games like the demo based on the Unreal engine. I expect the Tegra 2 graphics core of the Adam more powerful.
Camera / video chat / VOIP
The iPad doesn’t have a built-in camera which is a glaring omission. The Adam will have a 3Mp built-in camera, which is more than enough for Skype videophoning. Moreover, the Tegra 2 supports real-time hardware encoding of 1080p streams to H264 so, properly written video-chat applications should work extremely well on the Adam. Ubuntu should run any Linux VOIP app compiled for ARM. I expect Skype and other open/closed source software work well on the Adam in the short-medium term.
Apple has recently lifted the restrictions on Skype and other VOIP apps in the iPhone app catalog so the iPad will have the voice part OK but you will need an external cam for video chats and currently there is no information on the video encoding capabilities of Apple’s A4 SOC.
The iPad is rated for 10hrs of use (wifi browsing). The Adam is specified to have 16 hours of wifi browsing. The Adam looks like the winner here but the iPad’s 10-hour runtime is also quite good.
The Adam’s PixelQi screen supports this activity much better especially in sunlit places. Ubuntu runs FBReader (my favourite ebook reader software) and has viewers for every kind of complex-document formats (most of them will be displayed in Evince, in case of Ubuntu). The iPad has a new reader application which is too early to write about but expected to be an intuitive reader-app (if the track record of Apple is any indication).
Both devices include GPS units, so navigation software should be available for both. Google Navigation will certainly run on the Adam, I just hope it doesn’t get blocked from the iPad app-store.
Other applications, Multitasking
While the iPhone has a huge selection of applications in Apple’s app-store, most of them will have to be tailored for the iPad for full potential. This will surely happen if the iPad becomes successful but it may happen slowly if the device proves to be less than a clear success. Application-wise, I expect the Apple iPad to be as closed as the iPhone, so you will be able to install only Apple-approved applications from the official app-store.
Ubuntu on the Adam can run any full-desktop or command line Linux/ARM software from the Ubuntu ARM repositories. This is a huge selection of software and includes powerful applications like OpenOffice, GIMP and others. These may not be optimized for the touchscreen interface but the Adam’s backside trackpad can help using them in tablet mode and in docked mode you will be able to use a USB mouse and keyboard just like with a netbook. Moreover, Ubuntu is completely free of limitations so you will be able install whatever software you want.
The iPad currently has multitasking disabled so you can run only one application at a time. The Adam has the full multitasking of Linux. The Adam’s dual-core hardware should run several applications efficiently in parallel. A good use-case for this: a Bittorrent download running on your Adam while reading an ebook (I do this quite often on my OLPC XO-1).
Apple doesn’t allow applications running in interpreters so you will never run a Flash/Java/Python/.Net-Mono application on the iPad although the hardware is sufficient for them. The Adam’s Ubuntu will run any of those applications without any restrictions. Flash 10.1 is expected to be optimized for ARM SOCs and Java 6 has an optimized version for the Cortex A9 processors so the Adam should run apps based on these technologies well. Desktop and Webstarted Java clients are quite common in the enterprise IT world so the Adam may get some love from there.
It is not yet known how much RAM the iPad or the Adam has, but based on the Tegra 2 development board, the Adam will have at least 1 Gb of RAM which is quite sufficient for running even several complex applications in parallel.
The Adam will have expandable storage by a microSD slot, while the iPad seems to have no storage expansion slot at all.
The cheapest version of the iPad is announced for $499 in the US. The Adam is expected to carry a sub-$400 pricetag ($320 is a current estimate).
The iPad hardware seems to be seriously lacking when compared to the Adam’s Tegra 2 foundations, PixelQi screen, trackpad and other features.
Application-wise, the playground is more leveled but for Linux-savvy people the choice is a no-brainer. The iPad will certainly appeal to people who want devices which “just work” and accept the serious, artificial limitations imposed on their device.
According to this faq-like post on the official Nvidia Tegra developer site, Ubuntu Linux is supported as an operating system for Tegra 2 based devices.
This is extremely important for both Nvidia and Linux in general since a lot of IT-savvy people find Android insufficient for the netbook form factor and ask for a “real” Linux on these very promising devices.
I tend to agree with this view primarily because Tegra 2 @1Ghz is a powerful SOC for a smartbook/netbook/tablet which can run a full desktop Linux with decent speed. I see no reason to limit Tegra 2 based systems to inferior operating systems like Android or Windows CE (bah). As an example, Windows CE 6.0 supports only 512Mb RAM and only one processor. Since the Tegra 2 has two Cortex A9 cores,Windows CE will not be able to utilize both. Android has no X-Windows on it so it cannot run normal Linux software, only software directly written for Android. In contrast, Ubuntu for ARM supports multiple processors, any reasonable amount of RAM and most of the popular Linux software can be installed readily from the Ubuntu ARM repositories, and even the more obscure sw likely requires only a recompile.
A simple Gnome desktop or Ubuntu’s Netbook Remix user interface may not be perfect for a touchscreen operated tablet but is very useable with the traditional laptop form factor. Some of the Tegra 2 tablets will add a pointer device as well, in addition to the touchscreen (like the Notion Ink Adam) so these machines will be easy to use with a customized, full Linux desktop.
The list of currently known Tegra 2 tablets/smartbooks.
Some performance comparisons between Cortex A9 (like the Tegra2) and Intel’s Atom:
Earlier, I wrote about the relative performance of Intel’s Atom and the upcoming smartbook processors based on ARM’s Cortex A9 here.
Continuing this line, I have found a demonstration video which shows a 1.6 Ghz Atom (likely the 270) and a 500Mhz Cortex A9 development board side by side. Both of the machines run at the same screen resolution, same memory and same operating system (looks like a stock Ubuntu with Gnome).
The video demonstrates web browsing performance with typical websites. The Cortex A9 board seems a little bit slower but not significantly, and at still a perfectly acceptable speed.
Now, the most astounding part is, that the A9 board runs only at 500 Mhz which means that its performance is throttled back for ultra-low power consumption. Cortex A8 level SOCs – the current generation – are known to run at 1 Ghz (Snapdragon, OMAP3, Armada) and the Cortex A9 Sparrow demo chip runs at 2Ghz (produced on the 28nm GlobalFoundries process).
This means that a completely doable Cortex A9 at 1.5 Ghz would have about 3 times the performance of the demo 500Mhz development board and still consume much-much less than the Atom. It would definitely leave the Atom 270 and N450 in the dust.
Moreover, the development board had no graphics accelerator at all, while finalized OEM SOCs will definitely have GPUs built-in (for example the Tegra 2 will include a Geforce GPU coupled with the two A9 cores).
The Cortex A9 at 500 Mhz is an ultra-low-power configuration and it is safe to say that it would take 1/10 – 1/5 of the consumption of the Atom which means much-much better battery runtimes.
In theory, the 1.6 Ghz Atom puts out ~4000 DMIPS and the dual-core, 500 Mhz A9 puts out only 2500 DMIPS raw power. This means that the A9 also has an architectural advantage somewhere. This may be the two real cores of the A9 versus the one-core-with-hyperthreading of the Atom. Since web browsers are by nature heavily multithreaded, the dual-A9 may support this type of application much better.
I can’t wait to see some prototype A9 devices displayed at CES.
Freescale has come up with a tablet reference design which they expect to be selling for under $200. It is not entirely clear whether they meant end-user prices or OEM production prices but $200 for a production price would be way too much for a Cortex-A8 category tablet, so I assume they meant end-user prices.
What is interesting about this design, that they have created a special, simplified user interface, which may be appealing for the target audience.
The operating system is a Debian derivative.
This machine would be ideal as an ultra mobile browsing/emailing, video/music playing device at home or when traveling.
The 7″ screen may be too small for some but it definitely makes the machine more portable.
More information here.
While I am eagerly waiting for the smartbook product introductions at 2010 CES, I am wondering what kind of performance we can expect from those upcoming ARM based tablets and netbooks. Although some of the smartbooks will be based on Cortex A8 technology, I believe only the more performant, dual-core Cortex A9 system-on-chip (SOC) designs will be really successful (see this about the A8).
Since newer Pine Trail Atom netbooks are already getting fairly good battery runtimes, the question is unavoidable for the smartbooks: will we get at least similar performance to the Atom based netbooks? New Intel based netbooks will mostly use the netbook-oriented Atom N450 chip (Pine View), so I will try to draw a comparison between this chip and the known characteristics of A9 SOCs.
Since it is extremely hard to come by good comparative data between ARM and Atom, partly because Cortex A9 based systems are not yet available for the public, this post is highly speculative and by no means should serve as the basis for purchasing your next smartbook/netbook.
By ARM’s specifications, the Cortex A9 core has an approximate raw performance of 2.5 DMIPS/MHz. It can run at 2Ghz when produced on the 28nm GlobalFoundries process. This is 5000 DMIPS/core with an expected 10000 DMIPS for a dual core setup (MP CORE version).
Since it is more likely that the first A9 SOCs will be manufactured with a 40nm process, we only calculate with a 1.5 Ghz top frequency which would yield about 7500 DMIPS.
Now, performance wise. the new N450 is only marginally better that the earlier Atom chips (5-10%, see this Anandtech article) and several discussions report that the older Atoms get 2.5 DMIPS/Mhz, a 1.6 Ghz Atom yields about 4000 DMIPS (one of the discussions). Pine View Atoms for netbooks (N450 descendants) are not planned to be made dual-core in the near future so we calculate with only one core.
I am aware that this is not a perfect comparison since DMIPS values between different architectures are not 100% comparable, but these results would mean an 80% advantage in raw power for the dual-A9.
Some more considerations:
All recent ARM chips include hardware decoders for H264 video while the N450/NM10 has no such capability. This means either choppy HD video from Youtube or an external H264 decoder chip (like the Broadcomm one or an Nvidia Ion like extension). Certain Cortex A9 SOCs promise multiple 1080p stream decoding in parallel (like the Tegra 2) without loading the general purpose ARM core. Moreover, in the case of the N450, the maximum output on hdmi is restricted to 1366×768 (1440×1050 for the analog vga out). So you can forget about viewing HD videos with your Atom netbook even if you have an external, HD monitor or TV.
Flash hardware acceleration is coming (with Flash 10.1) to all ARM machines with H264 decoders but in case of the N450 alone, there is nothing to accelerate with. So you better check whether your Atom netbook has the external video decoder or you will never watch streamed hd videos with decent speed. It is fully possible that Nvidia won’t produce a new ION chipset for the new Atoms since Intel denied access to the relevant hw interfaces. Instead, Nvidia will work even harder on Tegra 2.
Architecture wise: the new Pine View Atom remains in-order architecture, produced on 45nm while the Cortex A9 is out-of-order core (more modern, inherently more powerful than the Atom) with easy implementation on the TSMC 40nm process and the GlobalFoundries 28nm process.
The n450 SOC and the NM10 companion chip(set) still works in a 6.5w TDP while the dual-A9 SOCs are expected to work in a 2W TDP. This is massive difference and makes it likely that the same battery will last much longer with an A9 smartbook than with an N450 netbook. The N450 doesn’t need active cooling anymore (this is good news) so the new machines will not have fans but the TDP values above indicate that Atom netbooks will likely be hotter than A9 based smartbooks especially in continued use.
The third generation Atoms – which could improve the situation – will come only in 2012, seriously late compared to A9 based chips (1Q 2010).
It very much seems that dual core Cortex A9 SOCs will be at-least on-par with the Pine View N450 Atom, performance wise, and possibly overpower them by 50-80% in raw processing power. Graphics performance, end-user price and battery runtimes are also expected to be much better for the A9 based machines.
Unless, Intel comes out with much more powerful Atom designs for netbooks, ARM Cortex A9 based smartbook products may severely cut into Atom netbook sales. X86 compatibility is less of a factor in this segment, so consumers may decide based on perceived performance, battery runtime in which A9 smartbooks seem to have the advantage.
OLPC is widely known as the organization which – indirectly – started the netbook revolution by pioneering affordable, mobile computing devices. The second iteration of their low-cost educational laptop, the XO-1.5, is about to be released. “Released” in a sense, that it will become available to large scale educational projects but not to individuals or smaller, grass-roots projects. The current lean production cost of the XO-1 is at around $180, the XO-1.5 may go below that if a sufficient volume is achieved. The XO-1.5 is expected to provide full internet browsing with Flash support, ebook reading and the more traditional learning functions of the Sugar Learning Platform. The XO-1.5 will provide a Gnome or XFCE based Linux desktop in addition to Sugar. One of the goals of OLPC to make the XO be able to provide day-long battery life for students (although the current version of the hardware only provides ~3.5-4.5hrs of use with web browsing on wifi.
Smartbooks is an upcoming, low-cost consumer device category. These machines are expected to cost significantly less than current netbooks (sub $200-250 end-user price) while supporting the same functions (full web browsing, ebook reading, playing video/audio). Smartbooks will likely come with Android, full-desktop running Linux operating systems or Windows CE. Since these computers are based on ultra power-efficient ARM system-on-chip designs, they are likely to provide at least day-long battery runtimes (12-16 hrs with browsing on wifi is promised by several vendors).
As can be seen, capabilities and cost-wise, smartbooks coming in 2010 get very close to the upcoming OLPC XO-1.5. Some of the smartbook versions (based on Tegra 2 or other Cortex A9) will way outperform the XO-1.5 in processing power and battery runtime.
Naturally, the comparisson is more complex than this, since the XO has some special features, which are not planned for every currently known smartbook designs. The most notable are:
- Daylight-capable PixelQi screen for working/reading in direct sunlight. Some of the smartbooks will likely also come with PixelQi screens.
- Rugged industrial design with rubberized keyboard. There is no currently known smartbook product which puts special emphasis on usage in harsh environment but all ARM designs are fanless and it is likely that a lot of smartbooks will come with smaller capacity SSD drives so smartbooks will be likely less prone to typical laptop failures that the average netbook.
- Special educational software (Sugar). The Sugar Learning Platform is now available for a set of “normal” Linux distributions as an alternative desktop environment, next to Gnome and KDE, so with minimal effort, Sugar and activities can be made usable on smartbooks as well.
Again, it seems that a smartbook derivative should be able to play the role of an XO in an educational environment.
Two of the huge advantages of smartbooks is availability and economies of scale.They are planned to be available from a host of sales channels, most notably from 3G service operators at subsidized prices. This means, that some of the smartbooks will likely come at zero initial price, only a 2 year data contract will need to be signed. At the targeted end-user prices (sub $250) smartbooks will be immensely popular as secondary home computers and mobile companions (just like netbooks but even more) which means that economies of scale will be reached quickly and prices will go further down.
In contrast, OLPC products are not available for individuals or small projects and currently there are no major project sales at OLPC. This means that there is no easy way to reach economies of scale (unless, suddenly a lot of countries start ordering XOs, which is unlikely). G1G1 programs will not work again, the first was a success, the last one was a complete failure. (G1G1= give 1, get 1 = donate 400$, you get 1 machine for yourself and 1 other machine is sent to the developing world as your donation).
All of these factors point into one diection: If OLPC wants to stay as a relevant IT/educational effort, it should realign its hardware and “business” strategy taking advantage of the upcoming smartbook revolution.
To a certain extent, this seems to be happening. OLPC is planning to switch to ARM by 2011 with their XO-1.75 product and continue this line with the XO-3 in 2012. There is not much information about the XO-1.75 but it seems that it will be based on a Marvell ARM chip. Since Marvell hasn’t announced plans for Cortex A9 based products, we can assume, that OLPC will use one of the smartbook-oriented Armada (a Cortex A8 SOC).
Software-wise, OLPC has already improved its position by providing Gnome and Sugar as equal desktop alternatives on top of the default Linux OS shipping with the XO-1.5. This way, the XO can be a much more versatile tool when needed. Even Microsoft’s Windows XP is expected to run well on this machine, which may be appealing to some of their prospective customers.
These steps are certainly necessary but I would suggest a much more aggressive startegy:
- Stopping as much in-house hardware development as possible. Partnering with a smartbook producer, taking a polished, tested ARM Cortex A9 board design without any further customization. I would skip the Cortex A8 processors in order to ensure better performance than the VIA C7-M in the XO-1.5.
- The industrial design may be kept at OLPC but I suggest scrapping those extreme designs which has been recently circulated for the XO-3. Use a simple, 10″ touch tablet form factor or the current laptop design with a 10″ PixelQi screen.
- Using a Linux distro already customized for the said board and add Sugar only as an alternative to the default desktop environment.
- Sell the XOs to everyone in the developed world, ensure the widest possible availability. A $50 donation markup over the costs is reasonable but G1G1 style sales attempts should be strictly avoided. If sufficient sales can be generated, that will make sure that the project stays known to everyone.
This way, a lot of development costs could be spared and OLPC could come up with an ARM based XO product in 2010, long ahead of schedule. A move like this would revitalize the community around OLPC and put the organisation firmly back to the map of relevance.
Smartbooks are an upcoming line of affordable, mobile computing devices. Their use case scenarios fit in well with currently popular Java desktop applications (e.g.: Azureus/Vuze bittorrent client, RSSOwl feed aggregator…etc) and there are still a lot of websites using Java applets or webstartable Java applications for auxiliary functions (like mass upload of files). Moreover, many company intranets contain desktop Java based technologies (e.g. fast data entry forms, GIS mapping clients…etc). Due to this, affordable, Java capable devices may spur wider adoption of mobile computing within companies. Finally, the new JavaFX rich internet application framework (an Adobe Flash competitor) seems to be heavily supported by Oracle, so we can expect advancements on JavaFX deployments on public and intranet websites. JavaFX has similar Java requirements as Java applets.
For these reasons, it is interesting how smartbooks can be expected to work with desktop Java software.
Since the upcoming smartbooks are mostly based on ARM, we need to analyse how Java is supported on the latest ARM system-on-chip (SOC) processors.
Originally, ARM created dedicated Java support in their processors through its Jazelle technology. This provided hardware acceleration in a CLDC profile environment (Java edition for smartphones and other embedded applications). Jazelle provides better startup times until the Hotspot compiler can compile Java bytecode to native ARM code. If the device is memory constrained, it is possible to use Jazelle alone, without the Hotspot compiler. Some details about this can be found here.
In a smartbook environment CLDC is not enough, the full Java Standard Edition is expected in order to make the computer be able to run complex desktop programs like Azureus.
A number of upcoming smartbooks are based on the Qualcomm Snapdragon platform (which is roughly an ARM Cortex A8 level processor). A recent development (June 2009) that Java standard edition has been ported and optimized for the Snapdragon platform (details here). It is unknown if the Snapdragon uses Jazelle and the optimizations include Jazelle but it is likely since other Qualcomm ARM processors include this technology. This is encouraging from the Java perspective (the article mentions 32x improvement in application performance) but the Snapdragon and Cortex A8 processors may be generally not fast enough for complex applications.
The more powerful smartbooks will be based on dual core Cortex A9 processors (like the nVidia Tegra 2). These are generally assumed to be comparable to latest Atoms in performance. For this processor type, Sun has recently demonstrated their optimized Java 6 Standard Edition environment (2009 Oct, details are here). Cortex A9 processors include Jazelle by default and it is likely that SOC manufacturers will also include them in their end products.
Conclusion: With proper customizations by the device manufacturer, Java applications may perform better on Cortex A8/A9 based ARM systems than on comparable Atom chips because of the inherent Jazelle hardware support and the ARM specific optimizations done by Sun. The safe bet – from the Java perspective – would be on dual core Cortex A9 based systems since these will certainly have the necessary processing power for desktop applications.
The success of smartbooks remains to be seen but it is safe to say that Java will have good support on these systems and customers of such machines may expect reasonable performance from their Java applications.
Today, I watched a video about a prototype Wistron smartbook called the Pbook (see the video here) which is a Snapdragon based machine. The Snapdragon is roughly comparable to ARM Cortex A8 based SOCs like the Marwell Armada and the TI OMAP3.
The machine was running a Linux variant with a lightweight desktop environment (I believe it was XFCE).
The boot process seemed quite slow and the machine very unresponsive in general.
If the video represents the processing power of the Snapdragon realistically, I have to say, it will not be sufficient to run “real” desktop environments on Linux. This is a problem, because people will expect laptop-like behaviour from laptop-looking devices.
These machines may run Android snappily but Android is currently optimized for smartphones and thus may not fulfill user expectations. (It would certainly not fulfill mine)
It is possible, that when it comes to ARM system-on-chips, only the dual-core Cortex A9 based SOCs will be suitable for being used in smartbooks/netbooks (like the Tegra 2). However, it seems that a lot of producers have already chosen A8 based SOCs (for example the Lenovo smartbook for AT&T with Snapdragon).
Potentially, even A8 based systems can be fast enough if an optimized Linux edition is shipped with them but producers should be very careful to ship systems powerful enough to make their users reasonably satisfied with their responsiveness.
After watching this video, the picture has become a bit clearer. The video is about the commercially available (only in Japan) Sharp Netwalker PC-Z1 which uses a Cortex A8 based processor and runs a full Gnome desktop with Firefox and Abiword. Boot time is not very fast (~90secs) but since you can suspend-resume this machine, it is not very important. Firefox cold starts (first start) in about 10 seconds and warm starts (second start) in about 5 seconds. This is not very snappy but acceptable. OpenOffice Spreadsheet cold starts in about 50 secs which is rather on the unacceptable side. The Netwalker sports only 512Mb of RAM which is at the minimum for the Gnome desktop. Warm applications starts could be helped with more memory. Cold starts could be helped with more memory and some configurable pre-fetching. This is mostly in-line with my earlier assumptions.
Quite strangely, I find myself waiting for CES 2010. Not that I am a big fan of electronics shows (or planning to attend) but this CES is special since several notable companies seem to be making important product announcements at it.
One of these companies is Nvidia which is announcing the second generation of its Tegra system-on-chip (SOC) processors. Moreover, it very much seems that the product announcement will be coupled with actual end-product announcements from a set of companies.
Why is Tegra 2 so interesting?
It is interesting because it will help creating a new product category, most often referred as smartbooks. These are practically netbooks with very good battery life, integrated 3G connectivity and a very appealing price tag (sub $250). Tegra 2 processors are based on ARM’s Cortex A9 core design and are coupled with nVidia’s own graphics technology.
According to nVidia, smartbooks based on Tegra 2 should be able to play 3 Full HD (1080p) video streams in parallel, without major framedrops. This means that the machine can play a 1080p video and run some power-hungry application (e.g. a Bittorrent download) in parallel. I assume that performance wise, the dual Cortex A9 core of the Tegra 2 coupled with 1 Gb of RAM will be enough to run average netbook software snappily (OpenOffice, Skype, Firefox, chat-client). All of this without a noisy fan (ARM based systems usually don’t need active cooling) and with 8-16 hours of battery runtime.
For me, such a machine would constitute the ideal web-browser + video player + ebook-reader + bittorrent home computer. Ideally in the tablet factor. My current favourite is the Notion Ink tablet (although this company seems to have the least chance to actually deliver.)
Generally, the Tegra 2 will spur some competition in the Atom dominated netbook space. Originally, netbooks were promised as the affordable computing for everyone (started by OLPC). Unfortunately, netbooks are still not really affordable. Their price isn’t going down but rather stagnate or rise slightly. This means that there is no competition, which is likely due to Intel’s domination of the market segment (by its Atom processors).
Tegra 2 and the other new ARM based solutions (Snapdragon, Armada, OMAP) may be the long awaited competitors which result in really affordable computers for the general public.