Can 100,000 CPUs tackle fundamental barriers?

[TeaLeaf]

If you have a spare hour and want to understand a bit more about how protein folding occurs and the theory behind the programs that we download and run then I thoroughly recommend listening and watching this lecture given by Vijay Pande (the leader of the F@H program at Stanford).  

The lecture is an hour long, but highly informative and packed with some easily understand background information that gives you a glimpse into the scientific world within which the Pande Group operates.  It is targeted at a general audience, so you don't need to be a molecular biology post-graduate to understand what he is talking about.  

Part of theory talks about the potential future use of GPUs rather than CPUs.  Their view is that Moore's Law for CPU's (doubles in speed every 18 months) is different when applied to the floating point performance of GPUs where they predict a tripling in speed every 12 months!  That should boost the F@H scores if we get all those fancy graphics cards folding too - they reckon the GPUs outstrip the folding power of the CPU and we may see some 'live' application within the next 12 months.  They also mention the possibility of doing some mini-parallel computing using people's home network as a mini-super computer.

FYI, each WU covers about 1 nanosecond of the folding simulation.  You'll understand the relevant scale of that if you watch the lecture.

The 'theory' takes up the first 39 minutes, the 'results' start from 39:02 minutes if you are really boring and want to skip to the end to just listen to the razzle dazzle!

You can find the lecture here.  

And yes, I did watch it just now.  I got woken up by the fireworks and couldn't get back to sleep  :dummy:

TL.