Hybrid MPI and CUDA Parallelization for CFD Applications on Multi-GPU HPC Clusters

Algorithm 5

The dichotomy algorithm.

(1)	for i = 1; i < partition_number; i++
(2)	a = min_XYZ; b = max_XYZ; xyz_current = 0.5(min_XYZ + max_XYZ);
(3)	while abs (dis) > 1.0exp-10 do
(4)	dis=(current_count-average_count)/average_count;
(5)	if dis>0.0 then
(6)	b = xyz_current;
(7)	else
(8)	a = xyz_current;
(9)	end if
(10)	xyz_current = 0.5(a + b);
(11)	end while
(12)	end for