I been using my main desktop to do all the crunching for my self. But recently I have thought of using some thing like a raspberry pie in a small network array to do this by it self. I am wondering if any one has tried this and what was the outcome of the array? Do they have a place in the computational crunching arena or not?

Individually they are slow, but if you are on a seriously restricted power supply and speed doesn't matter then why not.

I have plenty of power for my main pc! I know that the University of Stanford built a array of raspberry pies that did something like ten thousand peta flops I dont remember how many were in that but I would think if you strung lets say five to ten you would get a couple beta flops per sec which is not a whole lot but in the grand scheme of things every flop that translates in to more research being crunched is better for the projects or am I wrong!

My only caveat would be that some projects don't have tasks available for the arm processor. CPDN, my main project is one, I can't remember which others.

My only caveat would be that some projects don't have tasks available for the arm processor. CPDN, my main project is one, I can't remember which others.

Those which support Linux on ARM are marked on the Projects page. The information isn't yet shown in the 'Attach to Project' wizard in BOINC Manager, but I'm working on it.

Raspberry pi is a great little platform for its intended purpose which is, according to the designers, to provide a low cost, platform with minimal power requirements for students in third world countries to use for learning how to program computers. When it comes to crunching numbers raspi's greatest weakness is the SD card. They are not designed to have huge amounts of data written to them. The more data you write to them the sooner they deteriorate and fail. An SD card is not all that expensive but it's a pain having to reinstall everything from scratch every time it fails. And you lose the work done on tasks that started.

You can avoid that problem by putting the BOINC data folder on a USB stick. They "wear out" too though not as fast as an SD card. You can connect an external hdd via USB but you have to provide 12 volts via an external power supply as raspi's USB ports don't supply sufficient amperage. USB 3.0 is a fairly fast interface. USB 2 is very slow compared to SATA and might slow down crunching appreciably.

Another option is use NAS for the BOINC data folder. Anything but an SD card.

That is awesome. I think if a group of us got together and put one of these mini superputer together and ran it on linux we could do some real good crunching

I have a number of projects happily humming along on my Raspberry Pi 3 B+s but for the life of me I cannot get the screensaver included in the package (sudo apt-get bionc install) to work. Can't find any instructions either. Can anybody help me get this working? TIA

I have a number of projects happily humming along on my Raspberry Pi 3 B+s but for the life of me I cannot get the screensaver included in the package (sudo apt-get bionc install) to work. Can't find any instructions either. Can anybody help me get this working? TIA

All I can do is point you to another thread https://boinc.berkeley.edu/dev/forum_thread.php?id=11102

The last post in the linked thread is out of date. Virtually none of the tasks for Climate Prediction have graphics now. There is I think just one model type that has but there haven't been any tasks of that type since CPDN came back on line. Also the tasks available right now don't run on Linux and require WINE.

The last post in the linked thread is out of date. Virtually none of the tasks for Climate Prediction have graphics now. There is I think just one model type that has but there haven't been any tasks of that type since CPDN came back on line. Also the tasks available right now don't run on Linux and require WINE.

It might well be that case that there are no projects that support the screensaver in BOINC for Linux any more.

Quake catcher network does. And this laptop has just downloaded 8 tasks for Linux from CPDN. These like the current Windows tasks don't have any graphics. A shame as the graphics did give the user some idea of what the task was doing and was useful for spotting if a task had gotten stuck in a loop, though I don't know of that happening recently, it was fairly frequent with a couple of model types in the past.

I'm using a android tablet to run SETI tasks. It's easier than raspberry pi. Have to watch the settings because there is a pronounced warn spot running three wu's.