Processor Allocation Using a Reservation Technique in Hypercube Computers*

Several allocation policies proposed in the literature for hypercubes provide only in-cremental performance improvement at the cost of complexity. This is primarily due to the FCFS scheduling used for job sequencing. Larger jobs tend to block the succeeding jobs which could otherwise be allocated on the system. A greater impact on the performance can be obtained by eecient job scheduling. This paper presents an eeort in that direction by introducing a new scheduling algorithm based on a reservation technique. Jobs that cannot be allocated reserve subcubes of required dimensions and enable the succeeding ones for possible allocation. The proposed scheduling scheme provides considerable performance improvement while maintaining fairness in job sequencing. Simulation results indicate that the average queueing delay is reduced by around 20%-50% in most cases.