King Fahd University of Petroleum and Minerals (Saudi Arabia)
1993
168
M.S.
King Fahd University of Petroleum and Minerals (Saudi Arabia)
1993
In this thesis a parallel list scheduling algorithm for scheduling a set of n partially ordered tasks on m processors of a distributed computing system has been studied. The parallel heuristic called Par-ETF is based on the sequential ETF approach. The Par-ETF algorithm was designed on the hypercube model and implemented on a transputer environment. Logical clock was used to study the performance of the algorithm to overcome the limitations imposed by the existing parallel computing environment. The time and cost complexity of the developed algorithm is O(n(log n + log m)) and O(usdmn\sp2usd(log n + log m)). It has been observed through theoretical analysis and through implementation that the parallel algorithm produces the same schedules as the sequential ETF scheduling algorithm. A number of tests have been made on the Par-ETF program. Results obtained from these tests agree with the theoretical analysis of the Par-ETF algorithm.