The influence of heterogeneous selfish tasks (or agents) on the efficiency of single machine resources allocation is studied，in which regular tasks and non-regular tasks exist simultaneously． Hence，noncooperative game is introduced to modeling such problems，and orresponding Nash equilibrium schedule and Pareto schedule are defined． The relationship between two solution concepts is investigated and a sufficient and necessary condition is given to judge whether a Nash equilibrium schedule is a Pareto schedule or not． The Price of Anarchy of Pareto schedule which quantitatively measures the loss of the system’s global optimum is analyzed． The results reveal the conflict mechanism among resource users and resource providers，and explain the influence of heterogeneous selfish tasks ( or agents) on the efficiency of resources allocation．