GPU Enhanced Programs for Neuroscience Research
In addition to Beowulf systems, today's most cost-effective hardware are graphics processing units (GPUs). Thus, computationally intensive scientific software potentially will benefit by migration to GPU compatible code. We are currently porting NEURON to GPUs via C for CUDA. Other, similar projects are also being considered. In consultation with IS company personnel, a GPU compatible version of NeuroJet was constructed at the University of Virginia.