Evaluating of Feasible Solutions on Parallel Scheduling Tasks with DEA Decision Maker

This paper surveys parallel scheduling problems and metrics correlated to and then applys metrics to make decision in comparison to other policy schedulers. Parallel processing is new trend in computer science especially in embedded and multicore systems whereas needs more power consumption to reach speed up. The QOS requirement for users is to have good responsiveness and for service providers or system owners to have high throughput and low power consumption in parallel processing or embedded multicore systems. Moreover, fairness is vital issue to make decision wether the scheduler is good or not. Using the metrics is very intricate because misleadling metrics will cause to lose performance and system utility that is why the metrics has been opted cautiously in this paper. However, satisfying all of the objects in which have potentially conflicts is computationally NP-Hard. So, tradeoff between metrics is needed. This paper indicates DEA FDH model based on linear programming that will select the optimal scheduling near to exact solution.