Title: Scheduling concurrent garbage collector using adaptive scheduling technique for multi-core systems
Authors: K.V. Charan; A.S. Manjunath
Addresses: Department of Computer Science, Siddaganga Institute of Technology, Karnataka, India ' Department of Computer Science, Siddaganga Institute of Technology, Karnataka, India
Abstract: Concurrent garbage collectors (CGC) are contributing significantly to reinforce the performance of recently dominated multi-core systems by exploiting full potential of the processing resources. The software designers have turned into managed languages for developing time sensitive real-time applications due to security and execution parallelism. However, as the complexity of applications, heap size and number of cores in a system is scaled up, it raises many issues in the designing of CGC. In this paper, we examine two factors, reducing the number of garbage collection (GC) pauses and mitigating probability of stop-the-world (STW) pauses that can affect performance of CGC. We propose an adoptive garbage collection scheduling strategy which is based on the amount of remaining heap memory. The implementation results shows that adaptive garbage collection can have better real-time performance than periodic, slack-based and static allocation-trigger based garbage collections.
Keywords: runtime system; garbage collection; stop the world; STW; real-time applications; schedulability.
DOI: 10.1504/IJSCC.2017.087145
International Journal of Systems, Control and Communications, 2017 Vol.8 No.4, pp.291 - 302
Received: 11 Nov 2016
Accepted: 13 Mar 2017
Published online: 06 Oct 2017 *