Sample C code:

int main(argc,argv)
int argc;
char *argv[];
{
   int reps;                   /* number of samples per test */
   struct timeval tv1, tv2;    /* for timing */
   int dt1, dt2;               /* time for one iter */
   int at1, at2;               /* accum. time */
   int n;
   char *inmsg, *outmsg;       /* Buffer containing message */
   int type;
   int numtasks, taskid;
   int rc, dest, source;
   MPI_Status status;

   rc = MPI_Init(&argc,&argv);
   rc|= MPI_Comm_size(MPI_COMM_WORLD,&numtasks);
   rc|= MPI_Comm_rank(MPI_COMM_WORLD,&taskid);
   if (rc != MPI_SUCCESS)
      printf ("error initializing MPI and obtaining task ID information\n");
   else
      printf ("task ID = %d\n", taskid);
   if (numtasks != 2)
   {
      fprintf(stderr, "Only using 2 tasks...continuing\n");
   }

   at1 = 0;
   inmsg = (char *) malloc(MESSAGE_SIZE);
   outmsg = (char *) malloc(MESSAGE_SIZE);
   type = 1;
   reps = NUMBER_REPS;

   if (taskid == 0)
   {
      /* round-trip timing test */
      printf("Doing round trip test, minimal message size, %d reps.\n",reps);
      dest = 1;
      source = 1;
      for (n = 1; n <= reps; n++)
      {
         gettimeofday(&tv1, (struct timeval*)0);          /* before time */
         /* send message to worker - message type set to 1.  */
         /* If return code indicates error quit */
         rc = MPI_Send(&outmsg, MESSAGE_SIZE, MPI_BYTE, dest, type, MPI_COMM_WORLD);
         if (rc != MPI_SUCCESS)
         {
            fprintf(stderr, "Send error in processor 1\n");
            exit(1);
         }
         /* Now wait to receive the echo reply from the worker  */
         /* If return code indicates error quit */
         rc = MPI_Recv(&inmsg, MESSAGE_SIZE, MPI_BYTE, source, type, MPI_COMM_WORLD, &status);
         if (rc != MPI_SUCCESS)
         {
            fprintf(stderr, "Recieve error in processor 0\n");
            exit(1);
         }

         gettimeofday(&tv2, (struct timeval*)0);  /* after time */

         /* calculate round trip time and print */
         dt1 = (tv2.tv_sec - tv1.tv_sec) * 1000000 + tv2.tv_usec - tv1.tv_usec;
         printf("round trip# %2d   uSec = %8d\n", n, dt1);

         at1 += dt1;
      }
      printf("\n*** Round Trip Avg uSec = %d\n", at1 / reps);
   } else if (taskid == 1)
   {
      dest = 0;
      source = 0;
      for (n = 1; n <= reps; n++)
      {
         rc = MPI_Recv(&inmsg, MESSAGE_SIZE, MPI_BYTE, source, type, MPI_COMM_WORLD, &status);
         if (rc != MPI_SUCCESS)
         {
            fprintf(stderr, "Recieve error in processor 0\n");
            exit(1);
         }
         rc = MPI_Send(&outmsg, MESSAGE_SIZE, MPI_BYTE, dest, type, MPI_COMM_WORLD);
         if (rc != MPI_SUCCESS)
         {
            fprintf(stderr, "Send error in processor 1\n");
            exit(1);
         }
      }
   }
   MPI_Finalize();
   exit(0);
}