module {
	func.func @reduce(%buffer: memref<?*f32>, %lb: index, %ub: index, %step: index) -> f32 {
		; Initial sum set to 0.
		%sum.0 = arith.constant 0.0 : f32
		; iter_args binds initial values to the loop's region arguments.
		;%sum = "scf.for"(%lb, %ub, %step) -> (f32) {
		cf.br ^for.entry(%lb: index, %sum.0: f32)

	^for.entry(%iv: index, %sum.iter: f32):
		%t = memref.load %buffer[%iv] : memref<?*f32>
		%sum.next =  arith.addf %sum.iter, %t : f32

		cf.br ^for.cond(%iv: index, %sum.next: f32)

	^for.cond(%iv2: index, %sum.next2: f32):
		; Yield current iteration sum to next iteration %sum.iter or to %sum
		; if final iteration.
		%iv.next = arith.addi %iv, %step : index
		%stop = arith.cmpi uge %iv.next, %ub : index
		cf.br-cond %stop, ^for.end, ^for.entry(%iv.next: index, %sum.next: f32)

	^for.end(%sum: f32):
		func.return %sum : f32
	}
}