Multi-Core Software

Process Scheduling Challenges in the Era of Multi-Core Processors

CONCLUSION

In this paper, we showed that optimal performance can be exploited by making the process scheduler aware of the multi-core topologies and the task characteristics. Multi-Core scheduling mechanisms and challenges are analyzed in an SMP environment. We looked at the impact of Intel® Dynamic Acceleration Technology on these workload scenarios. Some of the group-scheduling heuristics that can enhance optimal performance are presented. We looked at how some of these multi-core scheduling mechanisms are implemented in the Linux* operating system.

In future, one can expect the process scheduler to be micro-architectural aware for exploiting optimal performance. Similarly, one can expect that the research proposals and solutions in this area will drive future hardware and platform designs that will minimize the effects of shared resource contention and also assist the software in making and enforcing the right decisions.

Back to Top