The compute gradient module is used to compute gradient vectors for 2D or 3D scalar byte fields. The output is a field of the same dimension as the input, but each field element is a 3D vector of floating point values representing the gradient at that point.
The compute gradient module requires the immediate neighbors of each point for which the gradient is computed. Thus, each PE requires exactly one extra row or column on every side of the block of elements to be calculated. The data distribution scheme was implemented using the MPI_Send and MPI_Recv functions in the same manner as in the interpolate module.