Modeling data migration on CC-NUMA and CC-COMA hierarchical ring architectures

Parallel computing performance on scalable shared-memory architectures depends primarily on the structure of the interconnection networks linking processors to memory modules and on the efficiency of the memory/cache management systems. While Cache Coherence Non-Uniform Memory Access (CC-NUMA) and Cache Only Memory Access (CC-COMA) are two effective memory systems, and the hierarchical ring structure is an efficient interconnection network in hardware. This paper focuses on comparative performance modeling and the evaluation of CC-NUMA and CC-COMA on a hierarchical ring shared-memory architecture.<>