Real-Time Systems for Multicore Architectures

Real-time systems on multicore architectures aim to ensure that critical tasks meet stringent timing constraints in the presence of shared hardware resources and concurrent execution. As application domains such as autonomous vehicles, industrial automation and avionics demand ever higher performance, designers have turned to multicore processors to deliver throughput, energy efficiency and functional integration. However, multiple cores sharing caches, memory buses and interconnects introduce timing interference that can violate deadline guarantees. Modern solutions combine precise worst-case execution time estimation with resource-partitioning techniques, predictable arbitration policies and hardware monitoring to bound contention. Scheduling algorithms—both static and dynamic—are tailored to heterogeneous core clusters and mixed-criticality workloads, often supported by hypervisors or real-time operating systems that enforce temporal isolation. Memory controllers with predictable command scheduling and I/O virtualization frameworks further mitigate interference in SoCs. Advances in formal analysis, simulation and on-chip tracing now enable developers to tune platform configurations for guaranteed latency, while balancing energy consumption and average-case performance. This confluence of hardware mechanisms and analytical methods underpins the safe deployment of time-sensitive applications in safety-critical and real-time computing environments worldwide.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Research from all publishers

Researchers have proposed an extensible framework for multicore response-time analysis that moves beyond fixed worst-case execution times by modelling resource demands directly. This approach instantiates arbitration policies—such as TDMA, FIFO and fixed priority—for shared interconnects and caches, enabling timing verification across diverse hardware configurations. Evaluations demonstrate that predictable architectures designed with this framework yield substantially tighter latency bounds than those optimised for average-case performance. A comprehensive survey of interference-reduction techniques categorises methods for partitioning main memory, cache, and bus resources, and assesses their integration into schedulability analysis. It highlights trade-offs between hardware modifications, software isolation and analytical complexity, guiding the selection of approaches for mixed-criticality systems. Closer to industrial deployment, an analysis of Arm’s Memory System Resource Partitioning and Monitoring (MPAM) specification presents detailed instantiations of memory-access regulation mechanisms. By modelling worst-case memory contention across partitions, the study quantifies the performance impact of different QoS regulator configurations, offering insights for predictable memory architectures on commercial multicore platforms.

Real-Time Systems for Multicore Architectures publication trend

The graph below shows the total number of articles in real-time systems for multicore architectures across all publications each year (not limited to Nature Index journals).

Technical terms

Worst-case execution time (WCET): The maximum duration required for a task to complete under all possible execution scenarios, used to guarantee timing bounds.

Schedulability analysis: A formal method for determining whether a set of real-time tasks can all meet their deadlines under a given scheduling policy and resource model.

Timing interference: Delay introduced when concurrent tasks contend for shared hardware resources, such as caches or memory buses, affecting each other’s execution time.

Mixed-criticality: A system design paradigm where tasks of differing safety or performance importance co-exist on shared hardware, requiring temporal isolation guarantees.

Time-division multiplexing (TDMA): A predictable arbitration policy that allocates fixed time slots to masters for accessing shared resources, eliminating contention variability.

Hypervisor: A software layer that virtualises hardware resources and enforces scheduling and isolation policies for multiple operating systems or real-time partitions.

References

1. An extensible framework for multicore response time analysis. Real-Time Systems (2017).
2. A Survey of Techniques for Reducing Interference in Real-Time Applications on Multicore Platforms. IEEE Access (2022).
3. Analyzing Arm's MPAM From the Perspective of Time Predictability. IEEE Transactions on Computers (2022).
4. Architecture and analysis of a dynamically-scheduled real-time memory controller. Real-Time Systems (2015).
5. An I/O Virtualization Framework With I/O-Related Memory Contention Control for Real-Time Systems. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems (2022).
6. Energy-Efficient Task Scheduling in Design of Multithread Time Predictable Real-Time Systems. IEEE Access (2021).