See my Aruba pictures.

One Laptop Per Child program presented to Aruba - The News 20091016

The design of this so called $100 laptop for kids from the developing world dates from 2006. I played with it at FOSDEM 2007, while it was still in preview. Since then it has started the minilaptop ‘netbook ‘ revolution. And while it is 3 years old, it still has many innovative concepts which are not yet common  in other laptops. Many laudatios have been written, but let me  share two features which I still appreciate: When there is enough light (or too much light for other laptops), the screen is very readable with the back-light switched off. This is a big energy saver. And while you are only reading the screen, the operating system (Fedora Linux) powers down most of the laptop components. It’s really impressive to see the power led go on and off while the screen remains powered. Netto result: between 0.3 and 8W is enough for the OLPC, while my other laptop consumes often 40W. The day I forgot my OLPC power supply, I could use another one from a dead Sony walkman with power specs (5.4W) lower than any other reported usable power supplies (see my contributions at the OLPC wiki).

Frits Hoff suggested me two current OLPC issues (mainly in the developed world) to think about: Flash support and speed improvements. Not all Flash content plays perfect. Some content plays too slow, certainly with multiple flash animations on the same web page. And one has to click on the animations to start them. The first root of these problems is the proprietary nature of the Flash format. The license of the Adobe Flash player doesn’t allow redistribution. That’s why the XO comes with the Free Open Source Gnash Flash player. But it’s very hard to make such alternative players perfect, as the Flash format is partly secret, partly available under a non-disclosure agreement, which only allows the development of Flash export functionality, not Flash playing. Those that  don’t mind such evilness and closed source software can install the Adobe Flash player. But it doesn’t matter if you are using Gnash, Adobe Flash player or another player like swfdec, it seems there are no tricks which result in  enough performance gain to play the heaviest Flash animations. Most Flash cartoonish content will play nicely, but e.g. the recent encoded Youtube movies will play with very slow framerate or not at all. The XO laptop is simply not fast enough for Flash. That’s no real shame, as Flash has performance issues on many other platforms, such as iPhones (but Youtube is providing iPhone versions of their movies. Maybe we could try to get these as default on the XO? Maybe that’s a matter of browser identification string?)

So, an interesting question is what the future will bring concerning speed improvements. The XO-1.5 and XO-2 are announced. The design of the XO-1.5 is finished and almost identical to the one of the XO-1, except for the mainboard. As it is difficult to estimate the speed improvement, I made a tabular comparison:

This shows that we can’t expect to much speed improvement from the processor. The main reason for the cpu switch is the discontinuation of the AMD Geode development. The C7-M is from the same era as the Geode. I found only one direct Geode – C7-M benchmark comparison, for CineBench, which measures both cpu and gpu speed, but the used gpu is not reported. Anyway, we can’t expect a lot of improvement per MHz, and when I look at the TDP, I guess the XO will often need to throttle down the cpu frequency (1GHz to 400MHz) for thermal constraints. With the same frequency, the C7-M cpu’s are roughly half the speed of the Intel Atom cpu’s typically used in the current netbooks. Then, I expect more speed improvement from the memory increase and the VX855 chipset, which is bigger and more recent than the cpu. VX855 gives impressive hardware support for H.264, MPEG-2/4, DivX and WMV9 video codecs, “allowing smooth playback of 1080p HD video”. That’s nice, but HD video  does not really matter now for the XO or the developing world. I guess not much Flash content will be accelerated with these codecs. But Youtube started using H.264 at least for their iPhone and HD versions. On the other hand this video acceleration will need specific driver support, and Via has not the best Linux reputation. So, there is hope, but it will need performance tests and probably lots of driver improvements.

And the XO-2? The design is not finalized yet, but it’s gonna be once again a revolution. For the moment one thinks about an ARM processor. ARM cpu’s are not x86 compatible. They are used a lot in embedded devices and smartphones, and I expect them to grow significantly in the netbook market in 2009-2010. Linux is strong on ARM; The Windows ecosystem is weak on ARM, like on anything non-x86. Yes Windows-CE for ARM exists, and Microsoft could port Windows 7 to ARM, but that still leaves a problem for all other closed source Windows software. Only the original manufacters can recompile their closed source software, and many have no experience with that. But don’t worry: the developing world is not really waiting for an import of the Microsoft Monopoly