抄録
In parallel computing, it is important that each core is worked with good efficiency. However, it is difficult to achieve high performance in parallel computing, especially using Unstructured Mesh. In this paper, Cashe Oriented Domain-decomposition Method for parallel computing in MPI parallel environment on Unstructured Mesh is presented. A feature of the method is dividing the computational domain by the size of the cache memory in each core and renumbering of element number. Applying the method, the case of computation using 8 cores on compressible flow around a Concorde like airplane was able to achieve speed-up more than about 24% in comparison with a conventional method.
