Scaling Global Scheduling with Message Passing

Provided by: Max-Planck-Gesellschaft
Topic: Hardware
Format: PDF
Global real-time schedulers dispatch tasks to processors dynamically at runtime based on the state of all processors and all tasks. Such a global view of the system enables a wealth of attractive algorithmic properties - most famously optimality - that have spurred considerable interest in global scheduling in recent years. However, in practice, the Achilles' Heel of global scheduling is runtime overheads: maintaining the required consistent global state can be costly, and becomes increasingly problematic with rising core counts.

Find By Topic