A workload characterization for coarse-grain multiprocessors

Scalable shared memory multiprocessors commonly employ replication and the associated coherency maintenance of memory blocks, but differ in the granularity from fine-grain (cache-coherent multiprocessors) to coarse-grain (page-based distributed shared memory systems). Regardless of the size of coherency blocks, attaining good performance may depend on the number of copies staying small. Previous workload characterization studies of fine-grain systems have shown sharing patterns that tend toward small worker sets. Coarse-grain architectures offer scalability advantages unless sharing patterns for larger blocks are considerably less favorable. This is the question we address in this paper. Our results are encouraging in that, despite the coarse granularity, worker sets scale reasonably