In another thread, ianbook221188 was kind enough to post the presentation that Pande gave at the GTC 2010 conference. I realize that most don't have the time to watch the 45 minute lecture so I thought I'd post highlights. The information below was taken from Pande's presentation available at http://nvidia.fullviewmedia.com/gtc2010/0923-b-2007.html. Please let me know if there is anything I need to change.
 
The following shows the difference between computational chemistry and physics. When determining the path of a spacecraft, you know the location and speed of each of the planets. With chemistry, you can calculate the location or the velocity of each particle but not both. To do so, you have to use statistics to determine the probability of interactions with neighboring atoms and the forces acting on them including the EM and strong forces and the molecular bonds that holds the atoms together in the proteins. The polarity of water molecules also affects the folding properties of the proteins and makes the statistical analysis even more complicated.
 

 
The following compares the different clients with the number of units and the combined teraflops. It's interesting the low tflops that the cpu client is producing compared to the output of the PS3 clients. The gpu client has the least number of units and highest output.
 

 
The following shows how far we've come. Currently, we're processing folded molecules in the low microseconds. Eventually, we will start calculating proteins that take milliseconds to fold and will take a thousand fold processing power to do the calculations. Given Moore's Law, how long will this take? A year... ten years...
 

 
The following map shows the locations of computers involved in the folding project. Notice that the dots are comparable to major urban areas across the world. Personally, I didn't realize there were so many in Europe (and who is folding in Bermuda?).
 

Pande's point is that we are currently producing 4 pflops on 21,500 gpus. There are a billion computers in the world and a million gpus would get us eflops of computing power. Fold on! 

yer probably should of made a new thred for it, definitely a need to watch even if like me you don't understand half of it, it got some good info in it that show why we need to keep folding on

Does anyone know where I can get an autographed photo of Vijay Pande?  I want to put it above my computer on the wall.

brcromer

Does anyone know where I can get an autographed photo of Vijay Pande?  I want to put it above my computer on the wall.

Speaking of which, I was wondering while listening to him at the conference if he gets harrassed by admirers or folders who are aggrevated about the points system. The disadvantage of distibuted computing  is that he's more visible to the public. I doubt they had an autograph line following his leacture.

id like to mention that there is a red dot right were muskie lives 

What, you mean, like my hoody?
 

 

Been wearing that one for a good ol' while. May get myself another at some point.

definitely cool.  remember though, that most of those CPU's are single core clients.
 

That's awesome geekiness!

Yeah I was just playing.
 
This is some really cool information though.
 
 

Yeah UK isnt really t-shirt weather but I do try to wear them. Had to resort to a hoody as it's a little warmer.
 
I'm also using EVGA logo + folding team, etched into my DangerDen Torture Rack SR-2. Permission asked and granted.
 
I mean its all positive advertising for them. If you asked permission to wear a hoody that said something like:
 
EVGA logo + fails
 
I'm pretty sure they'd turn their noses up at that one ;)