Nowadays, applications under heterogeneous multi-processor embedded systems are seeking to reduce the scheduling length to improve the users’ enjoyment, however, with the decrease of scheduling length, the accompanying increase of energy consumption. Therefore, it is essential to get a reasonable balance between energy consumption and scheduling length. This problem has been studied by many scholars. However, most of the studies do not consider the running time of one task and the load of one task on each processor together. To deal this situation, we devise a time-load aware (TLA) scheduling algorithm to tackle this problem. Eventually, the result of experiment reveals that the TLA algorithm get a better scheduling length while satisfying the energy constraint requirement.