Collective Communications

1
Collective Communications

• Solution

2
Solution

• include ltstdio.hgt
• include ltmpi.hgt
• define N 300
• int main(int argc, char argv)
• int i, target /local variables/
• int bN, aN/4 /a is name of the array each
  slave searches/
• int rank, size, err
• MPI_Status status
• int end_cnt
• int gi /global index/
• float ave /average/
• FILE sourceFile
• FILE destinationFile

3
Solution

• int blocklengths2 1, 1 / initialize
  blocklengths array /
• MPI_Datatype types2 MPI_INT, MPI_FLOAT /
  initialize types array /
• MPI_Aint displacements2
• MPI_Datatype MPI_Pair
• err MPI_Init(argc, argv)
• err MPI_Comm_rank(MPI_COMM_WORLD, rank)
• err MPI_Comm_size(MPI_COMM_WORLD, size)

4
Solution

• / Initialize displacements array with /
• err MPI_Address(gi, displacements0)
• / memory addresses /
• err MPI_Address(ave, displacements1)
• / This routine creates the new data type
  MPI_Pair /
• err MPI_Type_struct(2, blocklengths,
  displacements, types, MPI_Pair)
• / This routine allows it to be used in
  communication /
• err MPI_Type_commit(MPI_Pair)
• if(size ! 4)
• printf("Error You must use 4 processes to run
  this program.\n")
• return 1

5
Solution

• if (rank 0)
• / File b.data has the target value on the
  first line /
• / The remaining 300 lines of b.data have the
  values for the b array /
• sourceFile fopen("b.data", "r")
• / File found.data will contain the indices of
  b where the target is /
• destinationFile fopen("found.data", "w")
• if(sourceFileNULL)
• printf("Error can't access file.c.\n")
• return 1
• else if(destinationFileNULL)
• printf("Error can't create file for
  writing.\n")
• return 1
• else
• / Read in the target /
• fscanf(sourceFile, "d", target)

6
Solution

• /Notice the broadcast is outside of the if, all
  processors must call it/
• err MPI_Bcast(target, 1, MPI_INT, 0,
  MPI_COMM_WORLD)
• if (rank 0)
• / Read in b array /
• for (i0 iltN i)
• fscanf(sourceFile,"d", bi)
• / Again, the scatter is after the if, all
  processors must call it /
• MPI_Scatter(b, N/size, MPI_INT, a, N/size,
  MPI_INT, 0, MPI_COMM_WORLD)

7
Solution

• if (rank 0)
• /Master now searches the first fourth of the
  array for the target /
• for (i0 iltN/size i)
• if (ai target)
• gi (rank)N/sizei1
• ave (gitarget)/2.0
• fprintf(destinationFile,"P d, d f\n",
  rank, gi, ave)

8
Solution

• end_cnt 0
• while (end_cnt ! 3)
• err MPI_Recv(MPI_BOTTOM, 1, MPI_Pair,
  MPI_ANY_SOURCE, MPI_ANY_TAG, MPI_COMM_WORLD,
  status)
• if (status.MPI_TAG 52)
• end_cnt /See Comment/
• else
• fprintf(destinationFile,"P d, d f\n",
  status.MPI_SOURCE, gi, ave)
• fclose(sourceFile)
• fclose(destinationFile)

9
Solution

• else
• / Search the b array and output the target
  locations /
• for (i0 iltN/size i)
• if (ai target)
• gi (rank)N/sizei1 /Equation to convert
  local index to global index/
• ave (gitarget)/2.0
• err MPI_Send(MPI_BOTTOM, 1, MPI_Pair, 0,
  19, MPI_COMM_WORLD)
• gi target / Both are fake values /
• ave3.45 / The point of this send is the
  "end" tag (See Chapter 4) /
• err MPI_Send(MPI_BOTTOM, 1, MPI_Pair, 0, 52,
  MPI_COMM_WORLD) /See Comment/
• err MPI_Type_free(MPI_Pair)
• err MPI_Finalize()

10
Solution

• Note The sections of code shown in red are new
  lines added to perform the broadcast of the
  target value. The sections of code shown in black
  are lines added to scatter the global array b
  among all four processors. Lastly, the section of
  code shown in blue is the new loop the master
  must execute to search its part (the first part)
  of the global array b.

11
Solution

• The results obtained from running this code are
  in the file "found.data" which contains the
  following
• P 0, 62, 36.
• P 2, 183, 96.5
• P 3, 271, 140.5
• P 3, 291, 150.5
• P 3, 296, 153.
• Notice that in this new parallel verison where
  all four processors search, the Master(P 0) finds
  the first location of the target, while processor
  1 finds none.

12
END