GPUGRID

GPUGRID is a volunteer computing project that harnesses the massively parallel processing power of consumer graphics cards (GPUs) to perform full-atom molecular dynamics (MD) simulations of biological macromolecules. Based at the Barcelona Biomedical Research Park (PRBB) and led by Dr. Gianni De Fabritiis, the project focuses on understanding the molecular mechanisms of diseases at atomic resolution.

Molecular dynamics simulation numerically integrates Newton's equations of motion for every atom in a molecular system — typically 50,000 to 500,000 atoms including the protein, surrounding water molecules, and ions. Each timestep advances the simulation by only 2 femtoseconds (2 × 10^-15 seconds), so reaching biologically meaningful timescales of microseconds requires billions of individual steps. GPUs, with their thousands of parallel processing cores, are ideally suited for these calculations — a single modern GPU can be 50–100 times faster than a CPU core for MD simulations.

GPUGRID was a pioneer in applying GPU computing to distributed biomolecular research, adopting NVIDIA CUDA technology early and running the ACEMD molecular dynamics engine developed by the same research group. The project's simulations have contributed to understanding how drugs bind to their protein targets, how ion channels open and close, and how mutations cause proteins to malfunction in disease states.

Research from GPUGRID has been published in leading journals including Nature, Nature Chemistry, Science, PNAS, and the Journal of the American Chemical Society. The project demonstrates how consumer gaming hardware can be repurposed for serious biomedical research, achieving simulation timescales that were unattainable just a decade ago.