Effects of Non determinism on the Predicted Speedup of Scheduling Low Level Computer Vision Algorithms on Networks of Heterogeneous Machines

Defining an optimal schedule for arbitrary algorithms on a network of heterogeneous machines is an NP complete problem. This paper focuses on data parallel deterministic neighborhood computer vision algorithms. This focus enables the linear time definition of a schedule which minimizes the distributed execution time by overlapping computation and communication cycles on the network. The static scheduling model allows for any speed machine to participate in the concurrent computation but makes the assumption of a master/slave control mechanism using a linear communication network. We investigate the limitations of the static scheduling model based on statistical descriptions of the model parameters. Using statistical models, an approximation of the schedule length density function is derived. This statistical model is used to establish better approximations of schedule length.