Evaluation of Cache Coherence Protocols on Multi-Core Systems with Linear Workloads

Cache coherence protocols are major factors in achieving high performance through thread-level parallelism on multi-core systems. Among them, the token coherence protocol is the most efficient cache coherence protocol in maintaining the memory consistency. We used linear workloads such as parallel matrix multiplication to evaluate the token coherence protocol against the directory coherence protocol using GEMS on Simics. The experimental results demonstrate that the token coherence protocol is a more flexible framework than the directory coherence protocol on multi-core systems because it can provide improved performance and reduced complexity. Moreover, we established the relationship between the shared cache size and total performance.