Analysis of Work Stealing with latency

We study the impact of communication latency on classical Work Stealing load balancing algorithm. Our paper extends reference model in which we introduce a parameter. By using theoretical analysis and simulation, overall this Makespan (maximum completion time). derive new expression expected running time bag independent tasks scheduled by Stealing. This enables us to predict under conditions given run will yield acceptable performance. For instance, can easily calibrate maximal number processors use for work/platform combination. All our results are validated through simulation wide range parameters.