GRAPH MODELS OF COMPUTATIONS IN COMPUTER SYSTEMS.
Abstract
A directed graph is used as a model for computational tasks. This model shows the precedence constraints, concurrency or mutual exclusiveness between subtasks. The directed graph is transformed into an acyclic graph in such a way that the mean path length of the original cycles are left unchanged. From this graph, mean path length measurements can be performed. Associated with an algorithm examining the structure of the graph, lower and upper bounds on the number of processors referred for maximum parallelism can be determined. A one-shot 'a priori' scheduling with different urgency rules is presented and tested on some example graphs. It presents a large saving of computational time compared to previous experiments. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1968
- Accession Number
- AD0678753
Entities
People
- Jean-loup E. Baer
Organizations
- University of California, Los Angeles