Parallel Benchmarks and Comparison-Based Computing

Non-numeric algorithms have been largely ignored in parallel benchmarking suites. Prior studies have concentrated mainly on the computational speed of processors within very regular and structured numeric codes. In this paper, we survey the current state of non-numeric benchmark algorithms and investigate the use of in-place merging as a suitable candidate for this role. In-place merging enjoys several important advantages, including the scalability of eecient memory utilization, the generality of comparison-based computing and the representativeness of near-random data access patterns. Experimental results over several families of parallel architectures are presented.