With improvements in required analysis accuracy, there also exists the tendency for analysis model elements to increase in number. Parallel computing that uses many cores and GPUs can be effective for calculating models of scales as large as these. The shared memory-type (SMP) which uses the many cores of individual nodes, the massively parallel distributed memory-type (MPP) which uses the many cores of either a single node or multiple nodes connected through a high-speed network, and parallel processing that uses the GPUs included in machines can all be used in JMAG. While the type of parallel computing and the number of cores used can be set in advance, making the above specifications in accordance with machine resources is also possible when executing the calculations themselves.
This document describes the procedure for changing settings for parallel computing when executing calculations.
Parallel computing, SMP, MPP, GPU, Executing calculation, Changing settings