Analyzing Fixed Priority Global Multiprocessor Scheduling.
| Document type: | Conference Papers |
|---|---|
| Peer reviewed: | Yes |
| Author(s): | Lars Lundberg |
| Title: | Analyzing Fixed Priority Global Multiprocessor Scheduling. |
| Conference name: | 8th IEEE Real-Time and Embedded Technology and Applications Symposium |
| Year: | 2002 |
| Pagination: | 145-153 |
| ISBN: | 0-7695-1739-0 |
| Publisher: | IEEE COMPUTER SOC |
| City: | SAN JOSE, CA |
| ISI number: | 000178662300014 |
| Organization: | Blekinge Institute of Technology |
| Department: | Department of Software Engineering and Computer Science (Institutionen för programvaruteknik och datavetenskap) Dept. of Software Engineering and Computer Science S-372 25 Ronneby +46 455 38 50 00 http://www.bth.se/eng/ipd/ |
| Authors e-mail: | lars.lundberg@bth.se |
| Language: | English |
| Abstract: | We consider a multiprocessor where hard real-time tasks are scheduled globally on m processors. Each task has a fixed priority and tasks are executed using preemptive scheduling. The state-of-the-art priority assignment scheme in such cases is called RM-US[US-LIMIT] [1], where US-LIMIT is a parameter to the RM-US algorithm. The challenge is to find the US-LIMIT that can gaurantee schedulability for as high utilization as possible. The previously best known US-LIMIT value could guarantee schedulability as long the multiprocessor utilization is below m/(3m-2), i.e. 0.33333 when m --> infinity. In this paper we define a new equation for US-LIMIT which quarantees schedulability for higher utilization values than the previous result. When m --> infinity we can now guarantee schedulability for all tasks sets when the multiprocessor utilization is below 0.37482. We also show that our US-LIMIT values are optimal, i.e. we show that there is no room for further improvement of this state-of-the-art priority assignment scheme. |
| Subject: | Computer Science\Computersystems |
