Next: Spin spiral calculations Up: Analysis of spin texture Previous: MulPCalc: k-space spin density Contents Index
MPI parallelization is available for kSpin. More detailed information follows:
FermiLoop
Although MPI parallelization is done for k-points on a grid, the second search should be done in each domain picked out in the first search. Since the second search is done sequentially only in domains necessary for calculations, the maximum of the appropriate number of MPI processes is the product of the first and second values for the keyword 'k-plane.2ndStep'. Note that the second search should cost most of computation.
GridCalc
MPI parallelization is done for k-points on a grid so that the maximum of the appropriate number of MPI processes is the product of the first and second values for the keyword 'k-plane.1stStep'.
BandDispersion
MPI parallelization is done for k-points on every k-path so that the maximum of the appropriate number of MPI processes is the maximum number of k-points among k-paths.
MulPOnly
MPI parallelization is done for k-points so that the maximum of the appropriate number of MPI processes is the number of specified k-points.