Numerical study on applicability of the shared memory type computing with the Xeon Phi processor for flows around a ship including physical models

Ohashi, Kunihide*; Onodera, Naoyuki   

The applicability of shared memory type computing with Xeon Phi Knights Landing processor is examined with the flow solver which is dedicated to the flow computation around a ship hull. The computations at the high Reynolds number with the one- or two-equation turbulence models and the free surface model, additionally, the over-set grid method are selected as the test cases. The processing speeds of the Xeon Broadwell and the Skylake processors show better results than the speed of the Xeon Phi processors. The speed up ratio is smoothly increasing until the maximum number of the core with using the Xeon Phi processor.