XO 1.5 Preliminary Battery Life Tests, Youtube Performance

The battery life when the laptop is suspended (but running
the backlight and the WLAN) of an XO-1 and an XO-1.5 were
compared.

The XO-1 lasted 10h 20m.
The XO-1.5 lasted 12h 15m.

WLAN wakeups were disabled on the XO-1 to even the playing
field, as the build I was using on the XO-1.5 (os48) doesn't have
them yet.

These are preliminary tests on the battery life of the updated OLPC XO 1.5, run by John Watlington of the OLPC Foundation.


All seems to be going very well - a 20% improvement is something to be proud off and I've also noticed my XO 1.5 B2 tends to do better than the XO-1 during intensive usage.
But let's look at the numbers: If you think 12 hours seems that it's too little compared with netbooks that can get 10 hours without being suspended, remember that this suspend state is a regular S3 suspend for most components like RAM and CPU but it allows the laptop to retain the image on screen. The OLPC uses a custom auxiliary chip called DCON to achieve this, one that is unique to it.
The laptop can resume from this state rather quickly, in less than 200ms. Below a video of the fast resume on the XO-1. It's hard to show the feature on video but you can see it is suspended by looking at the blinking power led and how fast the screen turns on after it stays fixed, which indicates a keypress and return from suspend.  

The battery of the XO is also of lower capacity some netbooks in order to save in manufacturing costs, whereas some netbooks have batteries of more than 5000mAh, the XO's battery is just 3100 mAh and of a low 6.5v. Typical scenarios for kids at school include times when the laptop can suspend but keep the image from a math exercise or a map, hence helping to achieve close to 10h of use from a cheap battery.

Performance has also improved, take a look at the flash video performance on the XO-1:

It's unusable and even the sound skips from time to time.
The XO-1.5, which features a VIA C7-M ULV @ 1GHz and 1GiB of DDR2, can produce a better experience. It's far from perfect but it is enought to reproduce at around 15 frames per second and the sound never skips:

Older Youtube codecs can run at full speed but the newer uses H.264 which stresses the CPU quite a bit. The chipset is the new VX855 that supports hardware H.264 decoding but that would require some support of it by Adobe - something which is not likely in a Linux environment. The XO 1.5 can run Windows but not Vista in a nice way and that's where the hardware accellerated Flash is currently working well, courtesy of DXVA 2.0.

There are numerous aspects of the XO I will be covering at a greater detail, like the reflective screen and the LiFePO4 battery, as well as a look on battery life on different usage scenarios, which will be tested when the final firmware becomes available towards the end of the year.