An MPICUDA Implementation for Massively Parallel Incompressible

MPI和CUDA结合实现并行 Moderngraphicsprocessingunits(GPUs)withmany-corearchitectureshaveemergedasgeneral-purpose parallelcomputingplatformsthatcanacceleratesimulationscienceapplicationstremendously.WhilemultiGPUworkstationswithseveralTeraFLOPSofpeakcomputingpowerareavailable