New paper #63: Accelerating Molecular Dynamic Simulation on Graphics Processing Units

We're happy to announce a new paper (#63 at http://folding.stanford.edu/English/Papers).  This paper describes the code behind the Folding@home GPU clients, detailing how they work, how we achieved such a significant speed up on GPUs, and other implementation details.  

For those curious about the technical details, I've pasted our technical abstract below:

ABSTRACT. We describe a complete implementation of all-atom protein molecular dynamics running entirely on a graphics processing unit (GPU), including all standard force field terms, integration, constraints, and implicit solvent. We discuss the design of our algorithms and important optimizations needed to fully take advantage of a GPU. We evaluate its performance, and show that it can be more than 700 times faster than a conventional implementation running on a single CPU core. 

Also, this software is now available for general use (for scientific research outside of FAH).  Please go to  http://simtk.org/home/openmm for more details.

Comments

[ProDigit]

So, does this work on Intel and AMD/Ati cards, or only on Nvidia cards?

[Steve]

Vijay who do we contact if we folded a work unit, it submitted and it wasn't counted on our overall score? I left my PlayStation crunching a 1,000,000 piece work unit for almost two days and even though I got no submission error from my PS3 (the number of failed to submit units didn't increase) the unit wasn't tallied. That's an awful big unit and long amount of time to crunch it for it not to be counted or used.

[susato, foldingforum.org moderator]

Steve, support requests, including questions about particular work units and servers, should be posted to the general support forum www.foldingforum.org where a large community of expert Folders is available to answer your questions. ProDigit, the answer to your question is yes - and much more detail is available in the GPU/ATI section of www.foldingforum.org.

[Jaak Ennuste, Estonia Donates]

GPU's are the future of massive parallel computation in terms of price/performance and watts/performance.

Jaak, estoniadonates.wordpress.com/

[-jcc]

Following the link to the paper leads me to .interscience.wiley.com - they want money for the paper!

Hey - I donate my CPU power, so why should I pay to read that paper?
Please make it available to download elsewhere for free.

regards,
-jcc

[JediLodge]

I agree with JCC. I leave my GPU system running to support the program. I am very happy to contribute the computing power. I am myself successfully exploring the GPU usefulness in DB analysis. I am contributing my system time as part of that study. It is very unexpected that I now have to pay for this paper.