Abstract
In recent years, the number of processor cores on one platform has largely increased, while evenly distributing jobs to every processor becomes an important issue. Most previously discussed scheduling situations were well defined in general cases. In this paper, we propose an algorithm, which is modified from Heterogeneous Earliest Finished Time (HEFT), to increase the performance of a homogeneous system. Our proposed algorithm inherits all the advantages of HEFT such as easy implementation, low complexity, high performance, and so on. In general condition, it generates less overhead for scheduling but the output performance still approximate to or even better than the recent modified version of HEFT up to 7%. The multiprocessor scheduling problems are focused on two dimensional vector graphics, and we will discuss how to estimate the processing time, determine the dependency of each sub-graph, map it onto a directed acyclic graph, and then use our proposed algorithm for vector graphics processing.
Original language | English |
---|---|
Pages (from-to) | 252-281 |
Number of pages | 30 |
Journal | Journal of Information Hiding and Multimedia Signal Processing |
Volume | 3 |
Issue number | 3 |
Publication status | Published - 2012 Dec 21 |
All Science Journal Classification (ASJC) codes
- Software
- Computer Vision and Pattern Recognition