There's the other very important factor: If you're going to be optimizing your calculations for PS3s, you need to have a bunch of them set up in the lab. Which means your grad students can use them for videogame tournaments. Good research comes from happy grad students. Happy grad students come from the Pande group...
Ben
· 1 year ago
Do they really have PS3s in-house at the Pande group!?
Lena
· 1 year ago
Yup. They're primarily running / displaying actual simulations. But there at least used to be a tradition of Friday evening gaming on them as well :-)
charlesju
· 1 year ago
Yeah I definitely agree that there are a lot of wasted CPU cycles out there. I think Folding@Home does some cool stuff to utilize this power, but there should be ways to leverage this effect even more.
One interesting, albeit, kind of ridiculous, startup idea I read about a couple months ago was a startup that leverages Flash to do distributed computing when users play their games or watch their videos. Stuff like this is pretty exciting and interesting, there will definitely someone with a smart idea to start a killer business idea around this in the next couple of years.
AnthonyPhan
· 1 year ago
If only we could harness the power of people utilizing social networking sites like Facebook/Myspace.
Ben
· 1 year ago
That's a pretty awesome idea actually...
Spark05
· 1 year ago
I would like to comment on BOINC portion. For the average person to not feel daunted by the process and make it as easy as checking your email I would recommend GridRepublic (www.gridrepublic.org). GridRepublic is a nonprofit working in collaboration with BOINC and is known as an account manager. They make using BOINC very simple by only having one log in to control all the projects you contribute to and help you find more projects and allow you to manage one to an unlimited amount of computer you own from a single website.
Also as of December 17, 2008 BOINC has added GPU computing for CUDA enabled graphic cards.
Ben
· 1 year ago
Jonathan, thank you for commenting. That is very interesting and a good resource to share. And I was not aware that BOINC had just begun support of CUDA. Will they be extending this to cover ATI's CTM/Brook/OpenCL (whatever standard it is they're using these days?) standard soon?
Spark05
· 1 year ago
From my knowledge BOINC is in discussion of how to accommodate future additions of support of other processing types and my guess is they would be supporting AMD(ATI) GPU's and standards at the request or interest of scientists using/wanting to use BOINC and/or as they get more programmers contributing.
They are always looking for more people to help code in new features, enhancements, and testing/debugging. See here for development projects: http://boinc.berkeley.edu/trac/wiki/DevProjects
Brandon Sos
· 1 year ago
Don't remember how I stumbled on this blog, but that's pretty interesting. I recently started learning visualization techniques for gene expression data/gene networks/pathways etc, and was thinking o sweet! giant gene network! Though I realized these examples need a lot more computational power after reading the article. I remember in 03... or 04 using a distributing computing method by SETI. It was a screen saver that utilized CPU power when the screen saver went on. I uninstalled it though, seemed to slow down my comp to much after running a bit. Other then that I've seen a distributed computing program for brute force recovery of passwords haha
All Comments
If you're going to be optimizing your calculations for PS3s, you need to have a bunch of them set up in the lab. Which means your grad students can use them for videogame tournaments. Good research comes from happy grad students. Happy grad students come from the Pande group...
One interesting, albeit, kind of ridiculous, startup idea I read about a couple months ago was a startup that leverages Flash to do distributed computing when users play their games or watch their videos. Stuff like this is pretty exciting and interesting, there will definitely someone with a smart idea to start a killer business idea around this in the next couple of years.
Also as of December 17, 2008 BOINC has added GPU computing for CUDA enabled graphic cards.
They are always looking for more people to help code in new features, enhancements, and testing/debugging. See here for development projects: http://boinc.berkeley.edu/trac/wiki/DevProjects
If your curious here's a couple good progs for network visualization.
http://cytoscape.org/
http://www.ingenuity.com/