FundScience Blog

You don't usually imagine the multi-billion dollar video game industry and hoards of teenagers with eyes glazed, frantically pounding away at their keyboards and controllers contributing to solving complicated scientific questions or developing therapeutics for human diseases. But the demand for ever more intricate video games with more complicated graphics and higher definition video resulted in the development of specialized graphics cards, called GPUs (graphics processing unit). Over the last number of years, scientists and others have realized that this technology could be used to dramatically speed up non-graphical calculations, because the type of numerical operations in video games are also used in scientific simulations. This enables GPUs to perform calculations that once required supercomputer-type resources to run on the computer sitting on your desktop at a fraction of the cost and energy usage.

CDW has generously donated one of the latest generation GPUs (the NVIDIA GTX 480s pictured above) through FundScience. I have been in the process of developing code to perform GPU-accelerated simulations using OpenMM, which is also used in the popular volunteer-computing project Folding@Home. With a single GPU, we can perform simulations that would take a year or longer on a traditional CPU-based system in less than a month. With that type of computing power, we can begin to imagine doing research that was completely unfeasible just a couple of years ago.