Search

Suspension-based locks are widely used in real-time systems to coordinate simultaneous accesses to exclusive shared resources. Although suspension-based locks have been well studied for sequential real-time tasks, little work has been done on this topic for parallel real-time tasks. This article for the first time studies the problem of how to extend existing sequential-task locking protocols and their analysis techniques to the parallel task model. More specifically, we extend two locking protocols OMLP and OMIP, which were designed for clustered scheduling of sequential real-time tasks, to federated scheduling of parallel real-time tasks. We present corresponding blocking analysis techniques, and develop path-oriented techniques to analyze and count blocking time. Schedulability tests with different efficiency and accuracy are further developed. Experiments are conducted to evaluate the performance of our proposed approaches against the state-of-the-art. [ABSTRACT FROM AUTHOR]

As the power density of modern CPUs is gradually increasing, thermal management has become one of the primary concerns for multicore systems, where task scheduling and dynamic voltage/frequency scaling (DVFS) play a pivotal role in effectively managing the system temperature. In this article, we propose CARTAD, a new reinforcement learning (RL)-based task scheduling and DVFS method for temperature minimization and latency guarantee on multicore systems. The novelty of CARTAD framework is that we exploit the machine learning technique to analyze the applications’ intermediate representations (IRs) generated by a compiler and identify an important feature which is critical for predicting the application’s performance. With the newly explored feature, we construct an RL-based scheduler with the more effective state representation and reward function such that the system temperature can be minimized while guaranteeing applications’ latency. We implement and evaluate CARTAD on real platforms in comparison with the state-of-the-art approaches. Experimental results show CARTAD can reduce the maximum temperature by up to 16 °C and the average temperature by up to 10 °C. [ABSTRACT FROM AUTHOR]

The mixed-criticality system provides multiple real-time applications with different criticalities in a single system. Poor energy-saving performance of the previous studies on mixed-criticality sporadic tasks are mainly due to the fact that the slack time generated from the random arrival of sporadic tasks is not taken into account. In this article, we focus on scheduling energy aware mixed-criticality sporadic tasks and take the random arrival of sporadic tasks into account. We proposed a dynamically frequency updating mixed-criticality algorithm (DFU). DFU based on the earliest deadline first scheme can exploit the slack time generated from high criticality tasks in a low criticality mode to reduce processor frequency. In addition, it also can dynamically update the utilization of sporadic tasks set to further reduce processor frequency. The simulation experiments are conducted to evaluate the performance of DFU and experimental results show that DFU consumes 34.29% less energy than that of the existing algorithms. [ABSTRACT FROM AUTHOR]

Capacity augmentation bound (CAB) is a widely used quantitative metric in theoretical analysis for directed acyclic graph (DAG) parallel real-time tasks, which reveals the key factors the schedulability of DAG tasks heavily depending on: the normalized utilization (the ratio of the total utilization to the core numbers) and the tensity (the maximum ratio of task’s longest path length to task’s deadline). However, CAB requires both factors of a schedulable task system to be capped by the same threshold. A task system with a normalized utilization slightly larger than that threshold but very small tensity, or very smaller normalized utilization but slightly larger than that threshold has good chance to be scheduled are both denied by CAB. To this end, we propose a new concept called capacity augmentation function (CAF) to better characterize the schedulability of parallel real-time tasks, which provides a more loose and different threshold for both factors. In particular, we derive a CAF-based linear-time schedulability test for real-time constrained-deadline DAG tasks under global EDF, which entirely dominates the state-of-the-art CAB-based test for constrained-deadline settings. Finally, we conduct experiments to compare the acceptance ratio of our CAF-based test with the existing schedulability tests also having linear-time complexity. The results show that CAF-based test significantly outperforms the existing linear-time schedulability test under different parameter settings. [ABSTRACT FROM AUTHOR]

We consider the problem of statically scheduling a task graph of moldable streaming tasks (i.e., the actor network) to a multicore or many-core CPU with discrete dynamic voltage and frequency scaling (DVFS). We employ an integer linear programming (ILP) approach that combines allocating cores to tasks, mapping tasks to core subsets, selecting a DVFS level for each task, and considering all options for task fusion as provided by a cost model, given data throughput and latency requirements and targeting low energy consumption. We also propose a partly decoupled approach that applies greedy prefusion before running an ILP-based scheduler considering the other three subproblems together. We use microbenchmarking on an ARM big.LITTLE architecture to quantify the advantage of task fusion in the above setting, and evaluate the use of task fusion in terms of energy savings, latency improvement, and scheduling time for three real-world applications. We confirm the scheduling results by running the applications with and without task fusions on the ARM big.LITTLE. Results indicate that streaming applications can profit from task fusion, as we achieve a significant reduction of energy consumption in most cases, while scheduling time is only moderately increased. [ABSTRACT FROM AUTHOR]

Multicore processors have become mainstream computation platforms not only for general and high-performance computers but also for real-time embedded systems. To fully utilize the computation power of multicores, software must be parallelized. Recently, there has been a rapidly increasing interest in real-time scheduling of parallel real-time tasks, but the field is still much less mature than traditional real-time scheduling of sequential tasks. In this article, we study the real-time scheduling and techniques for parallel real-time tasks based on decomposition, where a task graph is transferred to a set of independent sporadic tasks. In particular, we propose new decomposition strategies that better explore the structure feature of each task to improve schedulability. We develop schedulability tests for the global earliest deadline first (EDF) scheduling algorithm based on decomposition and three types of its variants, with their own pros and cons in different aspects. We conduct experiments to evaluate the real-time performance of our proposed scheduling algorithms against the state-of-the-art scheduling and analysis methods of different types. [ABSTRACT FROM AUTHOR]