c - MPI Scatter/Gather scope of variables -

i'm dealing mpi version of bml automaton mpi_scatter() won't work expected. read here collective communication functions every process needs copy of array, allocated space without initialization. in code there subgrid local_grid every process manipulates, , starting big grid root manipulate. mean use scatter-gather communication mpi datatype. allocate space grid , sub-grid every one, , initialize grid root. wrong?

        unsigned char*** local_grid;         unsigned char** grid;          mpi_status stat;         mpi_datatype rowtype;          mpi_init(&argc, &argv);         mpi_comm_rank(mpi_comm_world, &rank);         mpi_comm_size(mpi_comm_world, &nproc);         local_n = n / nproc;          mpi_type_contiguous(n + 2,    /* count */                 mpi_unsigned_char,    /* oldtype */                 &rowtype              /* newtype */                 );         mpi_type_commit(&rowtype);          /* allocate space 3d local grids*/         local_grid = (unsigned char***)malloc(2 * sizeof(unsigned char**));         for(i = 0; < 2; i++) {             local_grid[i] = (unsigned char**)malloc((local_n + 2) * sizeof(unsigned char*));             for(j = 0; j < local_n + 2; j++) {                 local_grid[i][j] = (unsigned char*)malloc((n + 2) * sizeof(unsigned char));             }         }         /* initialize local grids*/         for(i = 0; < local_n + 2; i++) {             for(j = 0; j < n + 2; j++) {                 local_grid[0][i][j] = 0;                 local_grid[1][i][j] = 0;             }         }         /* allocate 2d starting grid */         grid = (unsigned char**)malloc(n * sizeof(unsigned char*));         for(i = 0; < n + 2; i++) {             grid[i] = (unsigned char*)malloc((n + 2) * sizeof(unsigned char));         }         /* root */         if(rank == 0) {             /* initialize 2d starting grid */             for(i = 0; < n; i++) {                 for(j = 0; j < n + 2; j++) {                     grid[i][j] = (((float)rand())/rand_max) > rho ? 0 : rand()%2 + 1;                     grid[i][0] = grid[i][n+1] = 0;                     printf("%2d ", grid[i][j]);                 }                 printf("\n");             }         }         /* */         mpi_scatter(grid[0], local_n, rowtype, local_grid[cur][1], local_n, rowtype, source, mpi_comm_world);          ... 

program terminates correctly, 1 single rowtype row passed scatter() root process, nothing other process despite of number.

an issue comes way declare/allocate 2d arrays.

you declared 2d arrays arrays of pointers arrays (e.g. matrix rows), mpi expects contiguous layout.

for example, can replace

grid = (unsigned char**)malloc(n * sizeof(unsigned char*)); for(i = 0; < n; i++) {     grid[i] = (unsigned char*)malloc((n + 2) * sizeof(unsigned char)); } 

with

    grid = (unsigned char**)malloc(n * sizeof(unsigned char*));     grid[0] = (unsigned char*)malloc(n*(n+2)*sizeof(unsigned char));     for(i = 1; < n; i++) {         grid[i] = grid[i-1] + n + 2;     } 

and use grid[0] buffer argument mpi_scatter() or mpi_gather() same thing must applied local_grid[0] , local_grid[1]

when need free grid, can simply

free(grid[0]); free(grid); 

note not think program can work correctly if n not multiple of nproc

wiki