10 1 in this experiment we manually transformed the perfect benchmarks r into parallel program versions