A key design issue for chip multiprocessors (CMPs) is how to exploit the finite chip area to get the best system throughput. The most dominant area-consuming components in a CMP are processor cores and caches today. There is an important trade-off between the number of cores and the amount of cache in a single CMP chip. If we have too few cores, the system throughput will be limited by the number of threads. If we have too small cache capacity, the system may perform poorly due to frequent cache misses. This paper presents a simple and effective analytical model to study the trade-off of the core count and the cache capacity in a CMP under a finite die area constraint. Our model differentiates shared, private, and hybrid cache organizations. Our work will complement more detailed yet time-consuming simulation approaches by enabling one to quickly study how key chip area allocation parameters affect the system performance.

Processor cores in a chip multiprocessor (CMP) typically share a large last-level cache and the off-chip memory bandwidth. Previous studies demonstrate that explicit cache capacity and off-chip bandwidth partitioning can yield better overall system performance than without partitioning. However, little work has been done to study the interaction between cache capacity partitioning and off-chip bandwidth allocation. This paper develops a hybrid analytical model that takes into account the two partitioning problems together in order to capture their inter-dependence. With an elaborate case study, we show that an optimal resource management strategy would require a coordinated allocation of the cache and the off-chip bandwidth resources.