San Antonio, Texas
Oct. 25, 1995 to Oct. 28, 1995
S. Darbha , Dept. of Electr. & Comput. Eng., North Carolina State Univ., Raleigh, NC, USA
D.P. Agrawal , Dept. of Electr. & Comput. Eng., North Carolina State Univ., Raleigh, NC, USA
The inter-processor communication time is a major bottleneck for distributed memory systems (DMSs) and can be reduced by having an efficient task partitioning and scheduling strategy. The paper deals with the scheduling issues and presents an algorithm to schedule tasks onto DMSs. The complexity of this algorithm is O(V/sup 2/), where Vr is the number of nodes of the directed acyclic graph (DAG). This algorithm has been applied to some practical DAGs and the results obtained using this algorithm are very promising.
processor scheduling; distributed memory systems; directed graphs; computational complexity; scalable scheduling algorithm; fast scheduling algorithm; distributed memory systems; interprocessor communication time; task scheduling strategy; task partitioning strategy; complexity; directed acyclic graph nodes
S. Darbha, D.P. Agrawal, "A fast and scalable scheduling algorithm for distributed memory systems", SPDP, 1995, Parallel and Distributed Processing, IEEE Symposium on, Parallel and Distributed Processing, IEEE Symposium on 1995, pp. 60, doi:10.1109/SPDP.1995.530665