Each day, trillions of trillions of trillions of processor cycles go to waste. As a computer sits idle, the processor is still processing instructions and using idle processes to fill the cycles not used by programs running on the computer.
Distributed computing, or more specifically Grid computing, aims to harness the unused power of normal users' computers when no one is using the machine. Instead of running a program on an immense, expensive, and limited-speed supercomputer, scientists developed a way to spread a large workload over hundreds of thousands of smaller, end-user computers. Some projects have found this approach to be quicker: a small group of scientists ran a distributed program on 10 of their own machines at home and finished in a week what would have taken their university's mainframe two weeks.
There were some computers from the currently-under-construction Econ/Business lab in the storage area of Hoyt. So much potential processing power was sitting the shelves that something had to be done. Thus, Folding@Westminster was proposed.
Information Systems permitted the UNIX lab to use nine machines that were unused and sitting in storage. These nine machines are running Folding@Home, a distributed computing project that simulates the folding of proteins with the human body. The project seeks to understand how proteins fold and how sometimes they do not fold correctly, thus resulting in various diseases such as Alzheimer's, Hodgekin's, and some forms of cancer.
Our setup is intended to run as long as Information Systems is not in need of the machines which were loaned to us. Each machine is a Dell Optiplex GX270 with a Pentium 4 3.0 GHz hyperthreaded processor, 512 MB of RAM, an nVidia GeForceFX 5200 graphics card, and an Intel Gigabit (e1000) NIC on an i865 motherboard. The processors hit somewhere around 5950 bogomips.
The initial proposal was written by Colin Dean in the summer of 2005. html pdf
There are multiple versions of the Folding@Home software, but the one in which most students and staff would be interested is the screensaver for Windows. When you are away at class or otherwise not at your computer, the screensaver turns on and displays graphical renderings of the protein on which it is currently working, as well as actually processing the protein.
If you would like to install Folding@Home on your own machine, please read the Download section of folding.stanford.edu. You can see our current work units below and our team scores here. Our team number is 45740 if you would like to join our effort. On campus, you'll need to check the "Use Internet Explorer Connection Settings" in the connection settings tab of the F@H configuration, since the program needs to go through the proxy to get work packets.
Statistics are updated hourly automatically by each machine. We no longer have the cluster, statistics have been disabled.
We also have some other machines on campus running the screensaver and the updating program.
These stats are updated randomly, most are every 30 minutes.
If you have questions about our setup or would like help installing Folding@Home, email cdean at cs dot westminster dot edu.
We've been mentioned in a number of newspapers (The New Castle News, The (Sharon) Herald) and are in tomorrow's edition of The Holcad. I've disabled the Work Unit Statistics section of the page since we only have two machines left processing and the script displaying that information hasn't ever really worked the way it should. Within the next few weeks, the GetStatusInfo utility should be ready for public use so that people running Folding@Home on their machines can automatically report their status periodically.
We are now at five machines, novem was needed. We're just about to break 1,000th place. Folding.extremeoverclocking.com has our team stats.
Down to six now, tres was needed by InfoSys.
We're down to seven machines as of today. Information Systems needed two of the machines back, so sex and septem were it. We plan on sending out an all-campus email with information on how to install F@H on Windows so that students can help in the effort. Colin found a slight bug in the progress meter script, so he'll be trying to fix that next week.
