GPU-Based Parallel Computing for Nanotechnology Research
Alex McGuffey · ValpoScholar (Valparaiso University) · 2012
An effective technology for parallel computing is the application of graphical processing units (GPU) to computationally intensive calculations. Present research in nanotechnology simulations requires intensive calculations that have the potential to be parallelized and may benefit greatly from GPU processing. These simulations involve eigenvalue calculations on matrices with sizes up to 7776 x 7776. GPU computing speeds up this core calculation by a factor of 2.5, saving hours of valuable research time. As the size of the matrix calculations increases, the speed up using GPU computing increases; however, at small matrix sizes the GPU actually takes longer to compute than the CPU.