Abstract
This paper proposes a parallel optimization algorithm PDF/IHS for the minimum execution‐time multiprocessor scheduling problem which is a strong NP‐hard optimization problem. PDF/IHS is a parallelization and efficient implementation of the only practical optimization algorithm DF/IHS among those which have been proposed for this scheduling problem. In PDF/IHS, processors perform depth‐first search in parallel on a heuristically generated search tree in such a way that it is searched hierarchically from the left‐ and right‐hand sides. The effectiveness of PDF/IHS has been verified by simulation and practical parallel processing on Alliant FX4. As a result, it has been recognized that most of the problems which required a long time by DF/IHS can be solved approximately in time 1/m by PDF/IHS using m processors. Moreover, even for a problem which required a very long time or could not be solved in a practical time by DF/IHS, it has been verified that PDF/IHS can give solutions in time less than 1/m.
| Original language | English |
|---|---|
| Pages (from-to) | 54-65 |
| Number of pages | 12 |
| Journal | Systems and Computers in Japan |
| Volume | 23 |
| Issue number | 13 |
| DOIs | |
| Publication status | Published - 1992 |
Keywords
- NP‐hard practical algorithm
- Scheduling
- branch and bound
- multiprocessor super‐linear speedup
- parallel search
ASJC Scopus subject areas
- Theoretical Computer Science
- Information Systems
- Hardware and Architecture
- Computational Theory and Mathematics