I've added an option to provide the MPAS block decomposition file as a namelist option. This enables MPASSIT to assign processors as MPAS would, creating an efficient decomposition strategy. This initially takes a bit more time than the previous code to create the message, but the interpolation itself is faster. Hence. the larger the grid and the more fields you're processing, the greater the advantage becomes for using this option.
This change should be backwards-compatible as users are not requires to provide a decomposition file in the namelist and it will default to the previous method if none is provided.
I've added an option to provide the MPAS block decomposition file as a namelist option. This enables MPASSIT to assign processors as MPAS would, creating an efficient decomposition strategy. This initially takes a bit more time than the previous code to create the message, but the interpolation itself is faster. Hence. the larger the grid and the more fields you're processing, the greater the advantage becomes for using this option.
This change should be backwards-compatible as users are not requires to provide a decomposition file in the namelist and it will default to the previous method if none is provided.