WHAT DO WE GAIN FROM HYPER-SYSTOLIC ALGORITHMS ON CLUSTER COMPUTERS?