Static use of locking caches vs. dynamic use of locking caches for real-time systems

Abstract

Locking caches are a useful alternative to standard cache memories in order to reach both predictability and high performance for multitasking, preemptive, fixed-priority real-time systems. Two schemes of locking cache are possible: static and dynamic use. Both schemas present a high degree of predictability and like-cache performance. But these two schemes are not equivalent. Each one performs better for systems with particular characteristics. This work show that static use presents a greater degree of predictability than dynamic use, but dynamic use offers better performance for the major part of the cases. Systems are grouped as a function of the relationship between cache size and code size, allowing a fast and easy prediction about the gain or loss of performance given by each use of locking caches.