Using GPUs to parallelize Process Studies Ocean Model

June 16, 2010

Our research group uses the Process Studies Ocean Model originally crafted by Prof. Amala Mahadevan at Boston University and now in use for multiple process studies at many institutions.

Graduate students Xianquin Yao and Eric Holmes at UMass Dartmouth are now beginning to investigate whether we can get a speedup of this model by harnessing the parallelizing potential of GPUs. Rakesh Ginjupalli and Prof. Gaurav Khanna have used CUDA Fortran for Astrophysical Black hole simulations, and we would be getting their advice  and help on PSOM optimizations. The computational power for this work will be enabled by our participation on the recent MRI granted to Prof. Bob Fisher and colleagues at UMass Dartmouth by the National Science Founfdation; see this link.