Abstract
We propose efficient load balancing methods for two computational problems namely ray tracing and bottom-up binary tree computing in a distributed environment. In the context of ray tracing, we propose a variant of a static load balancing technique presented in [15] where the sampling is based on partitioning the object space. Our approach partitions the image instead and uses an efficient scheduling technique for load balancing. Computations carried out on a binary tree arise naturally in image processing and network optimization problems. Many of these problems are solved efficiently in parallel by the popular tree contraction technique [1]. In this paper, we explore the tree-contraction technique in a distributed setting using the grain packing method [9]. Implementations of our algorithms on a cluster of workstations using Parallel Virtual Machine (PVM) [6] demonstrate nearperfect load balancing.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have