Abstract
This paper describes and analyzes the design of Mirage, an efficient and robust implementation of the Supervisor-Worker paradigm for distributing and coordinating the parallel tasks of image generation using ray-tracing, on a cluster of workstations. The Supervisor-Worker paradigm permits flexibility in the degree of concurrency in the system as well as a level of fault tolerance. Furthermore, the cluster of workstations with a Network File System (NFS) provides a flexible, cost effective, and yet powerful distributed computing environment. Image generation by ray-tracing is well suited for effective distributed processing due to its intensive computations and its ability to be partitioned into independent components. Key enhancements to the ray-tracing technique for image generation are also described in this paper.
Original language | British English |
---|---|
Pages (from-to) | 133-147 |
Number of pages | 15 |
Journal | Arabian Journal for Science and Engineering |
Volume | 24 |
Issue number | 1 B |
State | Published - Apr 1999 |
Keywords
- Distributed system
- Image generation
- Ray tracing