Improving the Configuration of the Predictable ACDC Data Cache for Real-Time Systems

Abstract

[EN] In real-time systems, analyzing the worst-case execution time (WCET) of a task in the presenceof data caches is hard. The ACDC is a data cache that provides predictability, facilitating WCET analysis.It works by granting data cache replacement permission to speci_c load/store instructions. Nonetheless,knowing how to select these instructions to minimize the WCET, i.e., con_guring the ACDC, is not trivial. In this paper, we propose four new methods to con_gure the ACDC, and compare them with existingmethods. Unlike those in previous studies, our proposed methods provide speci_c ACDC con_gurationsfor the different phases of a given task, instead of a single ACDC con_guration per task. We evaluate theWCET bounds obtained when using different ACDC con_guration methods on the TACLeBench benchmark suite. Our results show that the most complex benchmarks work better with multiple-content con_gurations,which indicates that realistic tasks may also bene_t from this kind of con_guration. The methods proposedin this study improve the WCET in more than 60% of cases, with an average WCET improvement of nearly5% and up to 50% in some cases.