Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.636320
Title: An analysis of some multi-microprocessor scheduling strategies for parallel programs
Author: Crespo, M. B.
Awarding Body: University College of Swansea
Current Institution: Swansea University
Date of Award: 1980
Availability of Full Text:
Access through EThOS:
Abstract:
Current trends in integrated circuit technology suggest that multi-microprocessors system will be of common use in the future. A multiprocessor architecture can be utilized in order to exploit the capabilities offered by parallel processing. A single processor computer system may not complete computations quickly enough in order to provide real-time responses. Partitioning a program into parallel processible tasks and using a multiprocessor system is one method of obtaining the required computational speeds. However, before a more integrated approach can be undertaken, the system designer must consider among other things, problems such as comnunication among the processors, suitable management disciplines for scheduling the user programs and the system resources, etc. In this thesis, we concentrate our effort in the study of scheduling strategies that could be implemented in dedicated multi-microprocessor applications. The combinatorial nature of the scheduling problem makes it computationally difficult to search for optimal schedules, since it amounts to an exhaustive enumeration of all possibilities. In fact, it is known that the problem of determining optimal schedules is NP-complete. Even in extremely restricted cases it is NP-complete. For this reason, we study heuristic strategies requiring only a reasonable amount of time to produce schedules with acceptable completion times. This has been carried out by simulating a hypothetical multiprocessor architecture. The results obtained in the simulation of the scheduling heuristics considered enable us to conclude that the exploitation of parallelism and the use of multiprocessors can positively improve the processing power of the system.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.636320  DOI: Not available
Share: