Gravity super cells all at top-level
All the grav.super
calls seem to be reported at the top-level, which may have unwanted
side-effects with things like task and cell locking. It isn't a problem for MPI communications
as that explicitly checks for gravity tasks, not the super cell-level. The problem could be
that the grav.mm
tasks are checked in cell_set_super_gravity()
, rather than
just grav.grav
. Needs some careful thought and investigation to make sure that this
isn't the required behaviour.