Notices

The BOINC distributed computing client lets people take part in worldwide computing efforts (the most famous of which is probably seti@home).

I realise the current Nokia tablets don't have much computing power compared to a desktop PC, but how easy would it be for someone to do a BOINC client for Maemo?

It's already available on Linux (as well as Windows and Mac) so presumably that would make it easier to do a Maemo version?

http://boinc.berkeley.edu/download_all.php

Obviously the really heavy processing power of a desktop or laptop would totally outclass the current tablets, but the new Maemo devices will apparently have greatly improved hardware and if Nokia does eventually start using Maemo on their phones then it's possible that Maemo devices would one day outnumber desktop computers (or at least come close).

If there are eventually lots and lots of Maemo devices out there, being able to run BOINC on them might greatly increase the amount of computing time available to BOINC projects. That's why I'm interested in seeing if BOINC can be brought to Maemo.

...in short bursts before batteries run dry in devices never allowed to idle.

So it would be technically worthless due to the stress of the computing tasks?

What if the devices were left plugged into the charger while running BOINC?

OMAP devices achieve their long battery life by, essentially, shutting down as much as possible as often as possible. It's been termed "race to idle", which symbolizes how the devices will rapidly ramp their clock speeds up to complete computational work quickly so as to to get back to dozing peacefully as soon as possible.

Because of this, the more powerful processors in the upcoming tablets don't shorten battery life anywhere near as much as one might intuitively think. It turns out that the more powerful processors get more work done faster, so they actually get to sleep perhaps a bit more than current tablets even under slightly higher processing demands.

The "race to idle" scheme falls apart though under continuous processing demands. See, for instance, any thread here venting hate at metalayer crawler for getting stuck, sucking up all available processor cycles, and draining the battery in no time.

That would probably work fine, though it may significantly extend charging times. It will really depend upon what else the device is doing and how heavily the BOINC pushes the system though.

Currently, it's quite possible to take a fully charged tablet, plug it in, and still deplete the battery within an hour or two. You need only set the screen to stay on at full brightness and then play streaming video at full volume. Future devices may be similarly capable of using more power than they can take in.

Don't expect Nokia to get involved in making BOINC run on the tablets though. Their priorities lie in the opposite direction.

How many of these are not hampered by the lack of a FPU ? 'Cause desktops can easily surpass 100x the FPU performance of most ARM chips, which, ironically enough, makes desktops more efficient in terms of computing power/watts..

If you really want to give it a try, Debian has some BOINC stuff compiled for armel. I hear someone has a hack to run Debian stuff on the tablets.