A Holistic Approach to Meet Multicore-System Timing and Interference Requirements (Download)
Multicore processors (MCPs) address limitations in single-core computing by overcoming the physical limitations of clock speeds and temperature control to support larger central processing unit (CPU) workloads. For hard real-time applications, such systems pose challenges to developers in knowing how to meet strict timing requirements and managing interference among heterogenous cores.
Hard real-time applications demand deterministic execution times. While the average execution time for a given set of tasks running on an MCP platform tends to be lower than for the same set of tasks on a single-core processor (SCP) setup, the distribution of those times is often spread out. This variability makes it difficult to guarantee timing requirements for critical tasks. It also causes significant issues for applications where every individual execution time matters—not just the average.