Tag Archives: grid computing

Fancy a BOINC? Part 3 – software

Flu Virus Analysis

This is part 3 of a short series. You might want to read part 1 first.

So far in my project for turning an old laptop into a BOINC box, I’ve been over what BOINC is and prepared the hardware. In this post I get the software running and do some actual number crunching. You need an internet connection for this, my old laptop has an ethernet port which is plugged in to a spare port on my router.

I discovered BOINCpe (site is in German) which is a Windows-based live CD for running BOINC. Just download it, edit a couple of configuration files, burn to CD or make aload it onto a flash drive and you’re all set.

BOINCpe turned out to not be the solution for me because it requires a minimum 256 mb of RAM and the laptop only has 128 (which was a lot when it was made). So because I already had some Ubuntu discs lying around I used one of those instead.

Use the “alternate” install CD (obtainable from here among other places). At the welcome screen, press F4 and choose “command line only”. Answer the questions that come up and away you go.

After installation, login as the user you created during instal. Now it’s time to install the BOINC client and manager. To do that, just type:

sudo apt-get install boinc-client boinc-manager

and then your password.

After a bit of Googling, I figured the easiest way to do what I wanted to do was to use a BOINC account manager. I chose Boincstats BAM because it was first on the list. So, sign up to the site and choose from the projects list. Boincstats will try to login to projects using the same email and password you used to sign up (I had to change my SETI@Home password). For projects you haven’t participated in before, Boincstats will create accounts using your Boincstats details. How many times can I say Boincstats in one sentence?

Anyway, having done that, you need to then tell BOINC to use the account manager. So on the Linux box, type:

boinccmd –join_acct_mgr http://boincstats.com/bam/ username password

where username and password are of course replaced with your username and password. I kept getting an error message at this point. You can probably fix it by editing the right configuration file, but I just navigated to /usr/bin and ran it from there.

We’re not quite done yet though. Now back on the Boincstats website, you need to login and go to the host list page. Select the machine you’re setting up (following these instructions, it will be the only one). Then check the boxes for the projects you want to run on that machine.

Then on the BOINC box, run the join_acct_mgr command again. When you get the command prompt again, run

boinccmd –get_state

This shows what the BOINC client is doing right now. You should get a whole lot of statistics that are really interesting to boring people like me.

And that’s about it. You can attach the graphical version of BOINC to the account manager with a couple of mouse clicks and some copy/paste. In future I might play around with configuring things for better performance, but this will do for now.

Links and resources:

Fancy a BOINC? part 1 – what’s it all about?

LONDON - MAY 03: Detail of the new computer 'Cluster' is pictured at Queen Mary, University of London on May 3, 2007 in London, England.  The Cluster, a high performance computing suite, is the joint initiative between businessman Sir Alan Sugar's company Viglen Ltd and the University. Through communication with worldwide Clusters to form a network of processors called the Grid, the project aims to play a leading role in international scientific research and data analysis. (Photo by Claire Greenway/Getty Images)

Last week, among the miscellaneous debris in my flat, I found a box containing two old laptops (a Toshiba and a Dell). The Toshiba originally had Windows 98 installed and the Dell had Windows 95. The Dell doesn’t even have a DVD drive, so I’m not sure what I’m going to do with that, but I decided to set up the Toshiba to run BOINC (which I already have running on a few other machines).

Backtracking a bit: BOINC stands for the Berkeley Open Infrastructure for Network Computing. The Wikipedia entry (at time of writing, 2010.03.10.16.35 Australian central summer time) says:

The Berkeley Open Infrastructure for Network Computing (BOINC) is a non-commercial middleware system for volunteer and grid computing.

In other words, if you run BOINC you volunteer to use your computer to run software for grid computing (I couldn’t think of a way to sum up grid computing in one sentence).

Grid computing is where you take a computing task that would take a long time, break it into smaller tasks, and send each of those to a separate computer to be processed. So instead of one computer spending 20 days to find a solution, you can use 20 computers to find it in 1 day. The original grid computing task was SETI@Home which analyzes radio telescope data to look for signals sent by aliens. BOINC was developed to add some security for SETI@Home and to open the system for other projects and now hosts projects for medical research, climate science, and solving difficult chess problems.

So why run BOINC? I originally joined the old SETI@Home because I hope first contact with aliens happens in my lifetime, since the switchover to BOINC I added a couple of cosmology projects because they may help with that. It’s been running in that form on the PC at my parents’ house for years, with no maintenance. I kind of forgot about it until recently, when Ruben Schade tweeted about medical projects he runs in memory of his mother who died of cancer. That inspired me to set up BOINC on my current PCs and sign up to some medical projects. Not exactly in memory of Ruben’s mother (I never met her, or him for that matter) but as a general altruistic here’s a way I can help do good things kind of way. The aliens probably have a cure for cancer, but if we can’t find them we’ll have to come up with one on our own.

Click here for part 2 – preparing the machine

Links & resources: