Research Study- Improved Algorithms for Yield Driven Clock Skew Scheduling in the Presence of Process Variations

In summary process change, conventional yield-driven clock skew scheduling can be expressed as a series of minimum ratio loop problems, so it can be effectively solved by algorithms such as Lawler and Howard. However, in this formula, Gaussian distribution of critical path delay is assumed, so it is not suitable for next-generation nanotechnology. Recently, a non-Gaussian generalization has been proposed and a modification of the roller algorithm has been developed to solve this general problem.

Another method of CPU scheduling is the shortest job priority (SJF) scheduling algorithm. This algorithm associates each process with the length of the next CPU burst for that process. When the CPU becomes available, the next CPU burst is assigned to the smallest process. If the next CPU burst of the two processes is the same, use FCFS scheduling to break the same rank. A more appropriate term for this scheduling method is the shortest next CPU burst algorithm, since scheduling depends on the length of the next CPU burst of the process, not the length of the entire process. The SJF algorithm is a special case of the general priority scheduling algorithm. The priority is associated with each process, and the CPU is assigned to the process with the highest priority. Priority processing is arranged in order of FCFS. The SJF algorithm is just a priority algorithm, and the priority (p) is the reciprocal of the next CPU burst.

In summary process change, conventional yield-driven clock skew scheduling can be expressed as a series of minimum ratio loop problems, so it can be effectively solved by algorithms such as Lawler and Howard. However, in this formula, Gaussian distribution of critical path delay is assumed, so it is not suitable for next-generation nanotechnology. Recently, a non-Gaussian generalization has been proposed and a modification of the roller algorithm has been developed to solve this general problem.

Dong Yunfeng et al. (2011, p. 3705) describe a new scheduling system algorithm which is a genetic search algorithm. Traditional genetic algorithms have disadvantages of initial convergence and mutation problems. The genetic search algorithm is an improved algorithm based on the advantages of genetic algorithms. Instead of using the mutation operator to solve the mutation problem, we use partial matching crossover to solve the initial convergence problem and the concept of the taboo search algorithm. This paper was able to find a solution to the school schedule between genetic algorithm, simulated annealing and genetic search algorithm. From testing, the time complexity of the genetic search algorithm is optimal and compared with other algorithms. However, in this test, we set the size of the population to 50 and the genetic algebra to 50. In the next experiment of the genetic search algorithm of this white paper, more genetic generation can be found.