An approach to hypercube allocation using weights /
Abstract: "A new processor allocation strategy, called weight allocation strategy (WAS), is proposed for a hypercube multiprocessor. WAS is based on the weights of the processors in the system, and provides a criterion of finding the best subcube of the requested size. The weight of a proces...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Book |
| Language: | English |
| Published: |
College Station, Tex. :
Texas A & M University, Computer Science Dept.,
[1993]
|
| Series: | Technical report (Texas A & M University. Computer Science Department) ;
93-004. |
| Subjects: |
| Summary: | Abstract: "A new processor allocation strategy, called weight allocation strategy (WAS), is proposed for a hypercube multiprocessor. WAS is based on the weights of the processors in the system, and provides a criterion of finding the best subcube of the requested size. The weight of a processor is determined by counting the number of its neighbors that are currently being used. The best subcube is the cube which has maximal weight sum of the processors in it among all the available subcubes of the requested size. It is shown that WAS is statically optimal and has a complete subcube recognition ability. The performance of WAS in terms of parameters such as waiting times is better than the previously proposed processor allocation strategies. We also apply WAS to the Buddy strategy and form a mixed strategy which has the same worst case time complexity as the Buddy strategy but has much better performance. Our allocation strategy is a best-fit approach that keeps the allocated processors as dense as possible, thus minimizing the system fragmentation." |
|---|---|
| Physical Description: | 24 leaves : illustrations ; 28 cm. |
| Bibliography: | Includes bibliographical references. |