An Experimental Study on Dynamic Bank Partitioning of DRAM in Chip Multiprocessors

Concurrent execution of multiple applications on chip multiprocessors leads to interference at different level of shared resources like the banks of a DRAM. There are a few studies in literature suggest that we can dynamically partition the DRAM banks among the running processes. However, a detailed performance statistics is not available in existing literature. In this paper, we made a comparative study of DRAM bank allocation algorithms between static and dynamic bank allocation algorithms on a large number of SPEC CPU 2006 benchmarks. We conclude that, though per application performance increases for few applications, there is no benefit in overall system performance.