Modify the program that just prints a line of output from each process (mpi outp

Question

Modify the program that just prints a line of output from each process (mpi output.c) so that the output is printed in process rank order: process 0s output first, then process 1s, and so on.

#include <stdio.h>

#include <mpi.h>

int main(void) {

int my rank, comm sz;

MPI Init(NULL, NULL);

MPI Comm size(MPI COMM WORLD, &comm sz);

MPI Comm rank(MPI COMM WORLD, &my rank);

printf("Proc %d of %d > Does anyone have a toothpick? ",

my rank, comm sz);

MPI Finalize();

return 0;

} /? main ?/

This is the original program .

Explanation / Answer

#include <stdio.h>
#include <string.h>
#include <mpi.h>    
#include <math.h>
#include <stdlib.h>

// Order of the matrix in the file
const int N = 8;

int main(int argc, char *argv[])
{
   int        comm_sz;              
   int        my_rank;             

   char* infile = "input.txt";
   char* outfile = "output.txt";

   MPI_File fh;
   MPI_Datatype filetype;

   // Start up MPI
   MPI_Init(NULL, NULL);

   // Get the number of processes
   MPI_Comm_size(MPI_COMM_WORLD, &comm_sz);

   // Get my rank among all the processes
   MPI_Comm_rank(MPI_COMM_WORLD, &my_rank);

   // printf("Openning file on rank=%d ", my_rank);
   // fflush(stdout);
     
   int q = sqrt(comm_sz);
   int gsizes[2]={N, N};
   int lsizes[2]={N/q, N/q};
   int psizes[2]={q, q};

   // Column-major order
   // int coords[2]={my_rank%psizes[0], my_rank/psizes[0]};
   // Row-major order
   int coords[2]={my_rank/psizes[0], my_rank%psizes[0]};

   int starts[2]={coords[0]*lsizes[0], coords[1]*lsizes[1]};
   char* array;
   array = (char*)malloc(lsizes[0]*lsizes[1]*sizeof(char));
  
   // Access modes:
   // MPI_MODE_RDONLY --- read only,
   // MPI_MODE_RDWR --- reading and writing,
   // MPI_MODE_WRONLY --- write only,
   // MPI_MODE_CREATE --- create the file if it does not exist,
   // MPI_MODE_EXCL --- error if creating file that already exists,
   // MPI_MODE_DELETE_ON_CLOSE --- delete file on close,
   // MPI_MODE_UNIQUE_OPEN --- file will not be concurrently opened elsewhere,
   // MPI_MODE_SEQUENTIAL --- file will only be accessed sequentially,
   // MPI_MODE_APPEND --- set initial position of all file pointers to end of file.
   MPI_File_open(MPI_COMM_WORLD, infile, MPI_MODE_RDONLY, MPI_INFO_NULL, &fh);
  
   // Example of parallel reads
  
   MPI_Type_create_subarray(2, gsizes, lsizes, starts,
           MPI_ORDER_C, MPI_CHAR, &filetype);
   MPI_Type_commit(&filetype);
   MPI_File_set_view(fh, 0, MPI_CHAR, filetype, "native", MPI_INFO_NULL);
   MPI_File_read_all(fh, array, lsizes[0]*lsizes[1], MPI_CHAR, MPI_STATUS_IGNORE);
   MPI_File_close(&fh);

   int i;
   char output[100];
   char output_copy[100];
   sprintf(output, "Chars on rank=%d: ", my_rank);
   for (i = 0; i < lsizes[0]*lsizes[1]; ++i)
   {
       strcpy(output_copy, output);
       sprintf(output, "%s%c, ", output_copy, array[i]);
   }
   printf("%s ", output);

   // Modify the array (turn ranks into chars)
   char c = 'A' + my_rank;
   for (i = 0; i < lsizes[0]*lsizes[1]; ++i)
   {
       array[i] = c;
   }
  
   MPI_File_open(MPI_COMM_WORLD, outfile, MPI_MODE_WRONLY | MPI_MODE_CREATE, MPI_INFO_NULL, &fh);
  
   // Example of parallel writes
  
   // No need for commiting the type again
   // MPI_Type_create_subarray(2, gsizes, lsizes, starts,
   //        MPI_ORDER_C, MPI_CHAR, &filetype);
   // MPI_Type_commit(&filetype);

   MPI_File_set_view(fh, 0, MPI_CHAR, filetype, "native", MPI_INFO_NULL);
   MPI_File_write_all(fh, array, lsizes[0]*lsizes[1], MPI_CHAR, MPI_STATUS_IGNORE);
   MPI_File_close(&fh);

   free(array);

   // Shut down MPI
   MPI_Finalize();

   return 0;
}

Related Questions

Navigate

• Browse (All)
• Browse M
• Subjects

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.