In this article, a new predictive controller based on fast model predictive control has been proposed for permanent magnet synchronous motor(PMSM) drive system to solve the quadratic programming problem online in a very short period. The mathematical model of PMSM has been described in terms of state space and discretized by Euler method firstly. Then, the discrete model is linearized at the steady state point. Based on this, the optimization objective function which is used to measure the performance of the PMSM is designed properly. The approximate primal barrier interior point method is employed to speed up the optimization process. Finally, the simulation results declare that the proposed algorithm is useful for PMSM drive system. In addition, the proposed approach enables the model parameters, the horizon size as well as the weighting factors updated in real time so that it has potential to solve problems such as model mismatch in the future.