Applications:

GROMACS BIOCHEMICAL SOFTWARE PORTED TO ClearSpeed PROCESSOR

ClearSpeed Technology, developer of high performance, low power, programmable microprocessor solutions, announced that some of the most computationally-intensive inner loops of GROMACS, a molecular dynamics software application, are now ported and off-loaded to ClearSpeed's PCI plug-in accelerator card.

GROMACS is an open-source application able to simulate the Newtonian equations of motion for systems made up of hundreds of millions of particles. It is primarily designed for biochemical molecules such as proteins and lipids that contain a high number of complicated bonded interactions. Protein folding and molecular simulations represent some of the most compute-intensive applications. GROMACS is a broadly used tool in biological and medical research and the biopharmaceutical industry, and has additional use for non-biological molecular dynamics applications.

ClearSpeed's initial acceleration port focused on the inner loops that account for 90 percent of GROMACS run-time for large molecular systems in solution. Since water molecules tend to vastly outnumber other molecules, (e.g. a protein of a few hundred atoms in ~30,000 molecules of water) most of the run-time is spent calculating non-bonded, Lennard-Jones and Coulombic force interactions between the oxygen and hydrogen atoms of the water molecules. Additional computationally-intensive loops will be ported to ClearSpeed processors in the future.

"GROMACS ported to ClearSpeed's silicon is an important first step. Next we'll be working with ClearSpeed to contribute the ported code back to the open source community. We're anticipating further optimizations on multiple cards and ClearSpeed's next generation processor," said Vijay S. Pande, assistant professor of chemistry and of structural biology at Stanford University. Pande is also project PI at Folding@home, a distributed computing project which studies protein folding, mis-folding, aggregation and related diseases. Folding@home (folding.stanford.edu) uses GROMACS as its main compute engine.

ClearSpeed's unique parallel architecture is designed to dramatically accelerate floating-point processing, enabling supercomputing capabilities on affordable platforms including clusters and workstations. ClearSpeed now has GROMACS running on a standard PC with ClearSpeed's 50 GigaFLOPS PCI add-in card and will give a presentation on this approach at the American Chemical Society (ACS) in Philadelphia this week.

"Biopharmaceutical researchers can now exploit accelerated GROMACS in localized small node clusters or desktop workstations," said Mike Calise, president of ClearSpeed Technology. "The benefit for life sciences companies is higher productivity and lower cost of ownership. In addition to stand alone, these 'bio-box' systems can be used in a Grid configuration to rival the performance of supercomputers at a fraction of the cost, space and power requirements."