I completely agree with your suggestion. That was basically what I was doing. All the information is useful in characterizing performance. So it would be good if maartennm could try starting up OO three times for at least some of the cases. In particular, it would be very interesting if, when starting within Maemo and from partition, there is a significant improvement from the 1:25min reported on the second try. You could also try comparing results after using close debian and not using close debian.

In maartenmk's test he is also utilizing Swappolube, KP47, CSSU, and AndrewX192's updated image so I am very interested in his restults because I do not currently use any of them. It is good to isolate what is having an effect, if any, on performace. For example, how do restults change if Swappolube is on or off. When I tried that in the past I did not see much improvement. But I will try again and consider checking other options utilized if the performance is improved. Also does it depend on combinations of options. For example, do you need to use Swappolube and CSSU together to see an improvement. I can't really say because I do not know how they work.

Finally, it might be good if there were "standard" files for testing purposes. For example a word document, spread sheet and ppt so that the tests are more relavent for comparison. Do such files exist?