CS 662: Theory of Parallel Algorithms
SR Matrix Examples

1. Matrix Multiplication
   1. Mesh - Heartbeat
   2. Prescheduled Strips

global sizes
	var N := 6
body sizes
	getarg(1, N)
end

resource point
	op compute(right, down: cap point)
	op rowValue(value: real)
	op colValue(value: real)
body point(i, j: int)
	import sizes
	var aij, bij real 
	var cij: real := 0.0

	proc compute(right, down)
		fa k := 1 to N ->
			receive rowValue(aij)
			receive colValue(bij)
			cij +:= aij*bij
			send down.colValue(bij)
			send right.rowValue(aij)
		af
	end

	final write(i, j, cij) end

end point
resource main()
	import sizes, point
	var grid[N, N]: cap point
				
	fa i := 1 to N, j := 1 to N ->
		grid[i,j] := create point(i,j)
	af

	#connect the grid together
	fa i := 1 to N-1, j := 1 to N-1 ->
		send grid[i,j].compute(grid(i,j+1),
									grid(i+1,j))
	af
	fa i := 1 to N-1 ->
		send grid[i,N].compute(noop,
									grid(i+1,j))
		send grid[N,i].compute(grid(N,i+1),
									noop)
	fa 
	send grid[N,N].compute(noop,noop)
	var A[N,N], B[N,N]: real
	
	readMatrix( A, B) # code not shown

	# Feed the processors
	fa i := N downto 1, j := 1 to N ->
		send grid[1,j].colValue(B[j,i])
		send grid[j,1].rowValue(A[j,i])
	af

	
	fa i := N downto 1, j := 1 to N ->
		destroy grid[i,j]
	af

end

global sizes
	var N := 10
	var PR :=2
	var S: int

body sizes
	getarg(1, N) 
	getarg(2, PR)
	S := N/PR
	if N mod PR != 0 ->
		write("N must be a multiple of PR")
		stop (1)
	fi
end

resource mult ()
	import sizes
	var a[N,N], b[N,N], c[N,N]: real
	sem done := 0, continue := 0
	var start := age ()

	process strip(p := 1 to PR)
		const R := (p-1)*S + 1
		fa i := R to R+S-1, j := 1 to N ->
			a[i,j] := 1.0; b[i,j] := 1.0
		af

		V(done); P(continue)

		fa i ;= R to R+S-1, j := 1 to N ->
			var inner prod := 0.0
		af
		fa k := 1 to N ->
			c[i,j] := inner_prod
		af
	end
		process coordinator
		fa i := 1 to PR -> P(done) af
		fa i := 1 to PR -> V(continue) af
	end

	final
		#code here to print results
	end