Add initial parallel matmul implementation

pmatmul.f90

@@ -0,0 +1,65 @@
+! Compile with:
+!
+! $ mpif90 -fcheck=all pmatmul.f90
+! $ mpiexec -n 1 ./a.out
+
+program pmatmul
+use mpi
+implicit none
+integer, parameter :: dp = kind(0.d0)
+integer :: n ! matrix size
+integer :: i, j, ierr, myrank, numprocs, rowstart, rowend
+real(dp) :: temp_sum
+real(dp), allocatable :: A(:,:), x(:), y(:)
+
+n = 1000
+
+! Initialize MPI environment
+call MPI_INIT(ierr)
+call MPI_COMM_RANK(MPI_COMM_WORLD, myrank, ierr)
+call MPI_COMM_SIZE(MPI_COMM_WORLD, numprocs, ierr)
+
+if (myrank == 0) then
+    print *, "nproc =", numprocs
+end if
+
+allocate(A(n,n), x(n), y(n))
+
+! Divide rows of matrix A among processes
+rowstart = ((n-1)*myrank)/numprocs + 1
+rowend = ((n-1)*(myrank+1))/numprocs + 1
+
+! Initialize matrix A and vector x
+do i = 1, n
+    x(i) = 1
+    do j = 1, n
+        A(i,j) = 1.0d0/(i+j-1.0d0)
+    end do
+end do
+
+! Scatter matrix A and vector x among processes
+call MPI_SCATTER(A(:,rowstart:), (rowend-rowstart+1)*n, &
+  MPI_DOUBLE_PRECISION, A(:,rowstart), (rowend-rowstart+1)*n, &
+  MPI_DOUBLE_PRECISION, 0, MPI_COMM_WORLD, ierr)
+call MPI_BCAST(x, n, MPI_DOUBLE_PRECISION, 0, MPI_COMM_WORLD, ierr)
+
+! Compute partial dot products
+do i = rowstart, rowend
+    y(i) = dot_product(A(:,i), x(:))
+end do
+
+! Gather result vectors from all processes
+call MPI_GATHER(y(rowstart:rowend), (rowend-rowstart+1), &
+  MPI_DOUBLE_PRECISION, y, (rowend-rowstart+1), MPI_DOUBLE_PRECISION, &
+  0, MPI_COMM_WORLD, ierr)
+
+! Finalize MPI environment
+call MPI_FINALIZE(ierr)
+
+! Print the result vector y from rank 0
+if (myrank == 0) then
+    write(*,*) 'Result vector y:'
+    write(*,*) y(:10)
+end if
+
+end program