You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
MagIC now tests several global communications frameworks: mpi_alltoallv, mpi_alltoallw and mpi_isend, mpi_irecv, mpi_waitall and picks up the fastest at the initialisation stage
Remove the allocation of Plm, dPlm arrays when SHTns is used
Reduce the memory allocation of linear solves by only allocating spherical harmonic degrees which are on processor
Reduce the number of global MPI transposes
Improve the OpenMP coverage in update* routines and in the global MPI transposes
Reduce the number of SHT when Boussinesq approximation is used: l_adv_curl=.true. needs to be set