Loading...
- Type of Document: Ph.D. Dissertation
- Language: Farsi
- Document No: 43472 (05)
- University: Sharif University of Technology
- Department: Electrical Engineering
- Advisor(s): Sanaei, Esmaeil; Goudarzi, Maziar
- Abstract:
- Advances in semiconductor manufacturing technologies have enabled us to build billions of transistors on a single die. However, the increasing amount of process variation in nanometer technologies has made it inevitable to move toward statistical analysis methods, instead of deterministic worst-case-based techniques, at all design levels. In this project, we studied the problem of variation-aware task scheduling for MPSoCs. To this end, we first proposed a variability analysis framework to analyze the effect of process variation on the main parameters of MPSoCs. Then, to solve the MPSoC task scheduling problem, we proposed two metaheuristic variation-aware task scheduling method based on Genetic algorithm and Simulated Annealing algorithm. In the cases where global optimum solutions are required, we proposed an ILP-based variation-aware task scheduling method. Finally, in the cases where finding an acceptable local optimum solution in a very limited execution time is considered, we proposed a heuristic variation-aware task scheduling method that can achieve significant speedup over well-known metaheuristic counterparts, while keeping the quality of the results within acceptable tolerance. Furthermore, our proposed algorithm always converges to a feasible solution in all our experiments unlike any of the previous algorithms.
- Keywords:
- Multiprocessor System ; Optimization ; Parametric Yield ; Task Scheduling Algorithm ; Process Variation
-
محتواي پايان نامه
- view