Issue No. 08 - August (1993 vol. 4)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/71.238623
<p>The author studies dynamic scheduling of computational tasks with communication costsusing nonuniform memory access architecture. The computing model assumes that datatransfer can be partitioned into parallel and sequential parts with respect to the taskexecution. A scheduling heuristic, called least-communication (LC), together with atwo-level scheduler is proposed in an attempt to minimize the finish time. The LC selectsthe task that removes the largest amount of remaining data transfer, if no such tasks areavailable the task that has been ready to run at the earliest is selected first. The timecomplexity of LC is O(n/sub 2/). Testing the finish time of LC and first-come first-servedscheduling (FCFS) shows that LC is useful for tasks having moderate granularity andwhose computation and communication requirements vary widely for different data sets.</p>
Index Termsmacro-dataflow dynamic scheduling; nonuniform memory access architectures;computational tasks; communication costs; data transfer; scheduling heuristic;least-communication; finish time; time complexity; first-come first-served scheduling;granularity; computational complexity; multiprocessing programs; parallel architectures;parallel programming; scheduling
M. Al-Mouhamed, "Analysis of Macro-Dataflow Dynamic Scheduling on Nonuniform Memory Access Architectures," in IEEE Transactions on Parallel & Distributed Systems, vol. 4, no. , pp. 875-888, 1993.