A Scalable Scheduling Algorithm for Coarse-Grained Reconfigurable Architecture

Coarse-grained reconfigurable architectures (CGRA's) are introduced as flexible architectures that can efficiently execute various types of applications in a single device. A CGRA often achieve high IPC by utilizing tens or hundreds of functional units (FU's). The key technique in exploiting a CGRA is to find an optimal mapping of operations over FU's. Modulo scheduling algorithm is known as the state-of-art technique to find fairly efficient solution; however it often takes too much time and occasionally fails as the number of FU is increasing. In this paper, we propose a novel two-stage scheduling algorithm which finds out a solution within a reasonable amount of time. The experimental result presents the proposed algorithm reduces the scheduling time by 92% and finds out schedules that are as efficient as the solutions given by the previous modulo scheduler.