Controllable fair queuing for meeting performance goals
Performance Evaluation - Performance 2005
Surplus fair scheduling: a proportional-share CPU scheduling algorithm for symmetric multiprocessors
OSDI'00 Proceedings of the 4th conference on Symposium on Operating System Design & Implementation - Volume 4
Push-assisted migration of real-time tasks in multi-core processors
Proceedings of the 2009 ACM SIGPLAN/SIGBED conference on Languages, compilers, and tools for embedded systems
Predictable task migration for locked caches in multi-core systems
Proceedings of the 2011 SIGPLAN/SIGBED conference on Languages, compilers and tools for embedded systems
Static task partitioning for locked caches in multi-core real-time systems
Proceedings of the 2012 international conference on Compilers, architectures and synthesis for embedded systems
Hi-index | 0.00 |
In this paper, we describe an extension of the Linux kernel scheduler called the Practical Fair-Share Scheduler (PFS). PFS is a fair-share process scheduler designed to support real-time workloads with soft (i.e., elastic) timeliness requirements. Fair-share scheduling is a common choice for soft real-time systems since it is work conserving and ensures a minimum CPU allocation for each process. A novel aspect of PFS is its treatment of placement and migration in SMP or multi-core settings. Other fair-share schedulers have used ad hoc policies for handling these issues, often leading to underutilization and increased system lag. In contrast, PFS uses a strategy that maintains utilization without unfairly penalizing processes. We spend the remainder of this paper discussing placement and migration. A more extensive description of PFS, including source code, can be found at the download site listed in the citations.