I was reading an article about a university group called Vision Lab of the University of Antwerp who created a computer designed to do tomography calculations which out performed the CalcUA on campus costing 3.5 million dollars. The CalcUA is the most powerful super computer in Belgium. (Tomography is the analysis of 3D medical images)
To read more about this click here: http://www.dvhardware.net/article27538.html
What is amazing about this machine is that it can be constructed for about $3800 from parts found on NewEgg.com.
This got me thinking about some of the distributed computing projects out there.
AMD Phenom 9850 processor + Scythe Infinity CPU cooler 4x MSI 9800GX2 graphics card 4x 2GB Corsair Twinx DDR2 PC6400 memory MSI K9A2 Platinum motherboard Samsung Spinpoint F1 750GB HDD ThermalTake Toughpower 1500W Modular PSU Lian-Li PC-P80 Armorsuit case Windows XP 64-bit
Remember the Seti project? The search or extra-terrestrial life in the universe. This project uses a distributed computing method to use your home PC to help analyze white noise data from space in an attempt to find defined patterns that could indicate the existence of life on planets other than earth. I consider this a complete waste of time considering many factors but if your geeky enough you might enjoy participating.
Then I though about protein folding. Which has real world benefits and could result in the cure for deceases as well as help us create many advancements is drugs such as antibiotics. This can also have commercial benefits as well.
Check out the web site: http://folding.stanford.edu/
Here is a video explanation:
As it turns out they have a version of the folding simulation software that supports advanced GPUs such as the Nvidia 8600 and up as well as the ATI 3750 series on up. My 6 Ghz machine folds at 60 iterations per second. Using the GPUs to run the calculation brings this to super computing speeds of 1300 iterations per second. This can be further tweaked to run on multiple GPU cores depending on how your hardware is configured.
Link for the latest graphical client I am using:
The nvidia 8600 cards show they are not supported until you load the CUDA display driver. This is much newer display driver from Nvidia than the one currently available for download.
This worked on both my 8800Gt and my 8600GTX cards.
One note about the display page for nvidia; it will not show your user name and shows you a test protein is selected. You can ignore that. Its best to run the application minimized anyways.
For multi cpu support you copy your folding@home folder creating multiple folders. The shortcut in your start menu points to a folder in you application data folder. You can change this to point to the other folder created in the Program Files directory. Then you change the shortcut to include the "-gpu N" flag (N starts at 0)
Your extra parameters in the advanced page of the systray clients, or the advances settings of the console clients. Again N starts with 0 not 1, so your primary display is 0, the next is 1, etc. Each client if you are running more then one needs a different -gpu and a different machineD, and a different working directory, so follow the instructions for multiple clients.Here are 2 tools that can give you more insight on what is going on with your progress.
http://www.blogger.com/http://fahspy.com/index_eng.shtml// - you point to your folding at home folder. You can monitor multiple folders for multiple GPUs with this tool.
My stats are here: http://fah-web.stanford.edu/cgi-bin/main.py?qtype=userpage&username=vbiggar
MIT open courseware on the subject of Protein Folding
I'm on the DL.TV team 57391, feel free to join.