Design Space Exploration of Object Caches with Cross-Profiling

Abstract

Abstract-To avoid data cache trashing between heapallocated data and other data areas, a distinct object cache has been proposed for embedded real-time Java processors. This object cache uses high associativity in order to statically track different object pointers for worst-case execution-time analysis. However, before implementing such an object cache, an empirical analysis of different organization forms is needed. We use a cross-profiling technique based on aspect-oriented programming in order to evaluate different object cache organizations with standard Java benchmarks. From the evaluation we conclude that field access exhibits some temporal locality, but almost no spatial locality. Therefore, filling long cache lines on a miss just introduces a high miss penalty without increasing the hit rate enough to make up for the increased miss penalty. For an object cache, it is more efficient to fill individual words within the cache line on a miss.

A Time-Predictable Object Cache

“…The results are in line with the evaluation of the object cache based on static WCET analysis [3] and cache simulation based on cross-profiling [9]. The improvement scales to some extent with larger object caches.…”

“…To verify the design decisions we have also evaluated the proposed object cache by executing large Java benchmarks (DaCapo) with cross-profiling [9]. With a simulation of the object cache we where able to compare the high-associative organization with a standard cache organization.…”

Data cache organization for accurate timing analysis

“…Simulations with larger workloads (DaCapo benchmarks) show a hit rate of 70% to 90% for small heap caches [39]. In this paper it is also shown that most hits stem from temporal locality and not spatial locality.…”