Issue No. 06 - June (1993 vol. 4)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/71.242160
<p>The authors present a new compile-time scheduling heuristic called declustering, whichschedules acyclic precedence graphs that fit the synchronous data flow (SDF) modelonto multiprocessor architectures. This technique accounts for interprocessorcommunication (IPC) overheads and considers interconnection constraints in thearchitecture so that shared resource contention can be avoided. The algorithm initiallyinvokes a new clustering method that uses graph-analysis techniques to isolateparallelism instances. When constructing an initial set of clusters, this procedure explicitly addresses the tradeoff between exploiting parallelism and incurring communication cost. By hierarchically combining these clusters and then systematically decomposing this hierarchy, the declustering method exposes parallelism instances in order of importance and attains a cluster granularity that fits the characteristics of the architecture. It is shown that declustering retains the clustering advantage of avoiding IPC, yet overcomes the inflexibility associated with traditional clustering approaches.</p>
Index Termsmultiprocessor scheduling; compile-time; scheduling heuristic; declustering; acyclicprecedence graphs; interprocessor communication; shared resource; contention;graph-analysis; multiprocessing systems; scheduling
G. Sih and E. Lee, "Declustering: A New Multiprocessor Scheduling Technique," in IEEE Transactions on Parallel & Distributed Systems, vol. 4, no. , pp. 625-637, 1993.