Packing, Spreading and Scheduling Latency

Explore the intricacies of task scheduling in CPUs through this 44-minute conference talk from the Linux Plumbers Conference. Delve into the complexities of spreading tasks on CPUs, examining how power domain topology and wakeup latency information can be leveraged to enhance power efficiency and scheduling latency. Analyze the results of an implementation to understand the benefits and potential drawbacks of task packing. Learn about current scheduling policies, system layout, idle statistics, and activity monitoring. Gain insights into when task packing is advantageous and how it can impact system power efficiency.