The TMS320C25 signal processor is capable of running real-time AC motor models using 32-bit internal accuracy. In this paper, we present two computationally efficient discrete-time AC motor models and their assembly language implementations. These two approaches result in different speed and accuracy characteristics in computing real-time estimates for the magnetic flux and current of the AC motor. In a multiprocessor configuration, the signal processor based AC motor model considerably reduces the computational load of a host CPU executing the core of the motor control algorithm.