Abstract: A consensus protocol based on local information feedback is proposed for the fast consensus problem of first-order, continuous, multi-agent systems under a directed network. The consensus and fast consensus problems of these systems with single or double gains are analyzed based on matrix theory and the time-domain analysis method, respectively. By increasing gain, this systems can converge to a common value rapidly. Furthermore, the gain cost needed for double gain systems is less than that of single gain system at the same convergence speed. Simulation results show the effectiveness of the proposed protocol.