Cache Aware Scheduling

Explore cache aware scheduling implementation in this 21-minute conference talk from the Linux Plumbers Conference presented by Intel engineers Tim Chen and Yu Chen. Learn about the proposed RFC patch series designed to optimize thread scheduling by keeping data-sharing threads within the same last level cache domain to minimize cache bouncing. Discover the primary use cases motivating this feature and examine current performance metrics demonstrating its effectiveness. Analyze the fundamental approach of the current patches and evaluate whether the feature should extend beyond single-process thread aggregation to include processes communicating through pipes, sockets, or shared memory. Investigate the potential integration of NUMA balancing memory scanning mechanisms to identify data-sharing tasks and estimate shared data extent. Review the load aggregation policy implementation and consider possible improvements to enhance scheduling efficiency in cache-aware environments.