Low-Cost Embedded Program Loop Caching - Revisited
Lea Hwang Lee, Bill Moyer, John Arends · 1999
Many portable and embedded applications are characterized by spending a large fraction of their execution time on small program loops. In these applications, in-struction fetch energy can be reduced by using a small instruction cache when exe-cuting these tight loops. Recent work has shown that it is possible to use a small instruction cache without incurring any performance penality [4, 6]. In this paper, we will extend the work done in [6]. In the modified loop caching scheme proposed in this paper, when a program loop is larger than the loop cache size, the loop cache is capable of capturing only part of the program loop without having any cache con-flict problem. For a given loop cache size, our loop caching scheme can reduce in-struction fetch energy more than other loop cache schemes previously proposed. We will present some quantitative results on how much power can be saved on an integrated embedded design using this scheme.