A scheduling algorithm for pipelined data path synthesis with gradual mobility reduction

Abstract

We propose a scheduling algorithm for the pipelined data path with resource constraint. The algorithm first checks the possibility of scheduling in the case of being assigned to the earliest step or to the latest step among the assignable control steps of all operations. If it is impossible to assign an operation to those steps due to resource constraint violation, the algorithm does away with those steps, that is, reduces the mobility of the operation. The scheduling algorithm is iterated until final schedule is obtained. If the final schedule is not obtained, even though there is no operation to reduce a mobility, we select a proper operation to reduce the mobility using the current scheduling state that is represented by parameters. A 16 point FIR filter and 5th order elliptic wavefilter are used to illustrate the scheduling algorithm.