concept vert_type in category parallel computing

appears as: vert_type
Parallel and High Performance Computing

This is an excerpt from Manning's book Parallel and High Performance Computing.

The MPI_Gatherv does the opposite operation, as shown in figure 8.4. We only need the global array on the master process, and so it is only allocated there on lines 40-42. The arguments to the MPI_Gatherv start with the description of the source with the standard triplet. Then the destination is described with the same four arguments as were used in the scatter. The destination rank is the next argument, followed by the communication group.

Listing 8.19 Setup for ghost cell exchanges in a two-dimensional mesh
GhostExchange/GhostExchange_Pack/GhostExchange.cc
30    int imax = 2000, jmax = 2000;    #A
31    int nprocx = 0, nprocy = 0;    #B
32    int nhalo = 2, corners = 0;    #C
33    int do_timing;    #D
      ....
40    int xcoord = rank%nprocx;    #E
41    int ycoord = rank/nprocx;    #E
42 
43    int nleft = (xcoord > 0       ) ? rank - 1      : MPI_PROC_NULL;    #F
44    int nrght = (xcoord < nprocx-1) ? rank + 1      : MPI_PROC_NULL;    #F
45    int nbot  = (ycoord > 0       ) ? rank - nprocx : MPI_PROC_NULL;    #F
46    int ntop  = (ycoord < nprocy-1) ? rank + nprocx : MPI_PROC_NULL;    #F
47 
48    int ibegin = imax *(xcoord  )/nprocx;    #G
49    int iend   = imax *(xcoord+1)/nprocx;    #G
50    int isize  = iend - ibegin;    #G
51    int jbegin = jmax *(ycoord  )/nprocy;    #G
52    int jend   = jmax *(ycoord+1)/nprocy;    #G
53    int jsize  = jend - jbegin;    #G