Graphic processing units （GPUs） have been widely recognized as cost-efficient co-processors with acceptable size, weight, and power consumption. However, adopting GPUs in real-time systems is still challenging, due ...Graphic processing units （GPUs） have been widely recognized as cost-efficient co-processors with acceptable size, weight, and power consumption. However, adopting GPUs in real-time systems is still challenging, due to the lack in framework for real-time analysis. In order to guarantee real-time requirements while maintaining system utilization ~in modern heterogeneous systems, such as multicore multi-GPU systems, a novel suspension-based k-exclusion real-time locking protocol and the associated suspension-aware schedulability analysis are proposed. The proposed protocol provides a synchronization framework that enables multiple GPUs to be efficiently integrated in multicore real-time systems. Comparative evaluations show that the proposed methods improve upon the existing work in terms of schedulability.展开更多
In the real-time scheduling theory,schedulability and synchronization analyses are used to evaluate scheduling algorithms and real-time locking protocols,respectively,and the empirical synthesis experiment is one of t...In the real-time scheduling theory,schedulability and synchronization analyses are used to evaluate scheduling algorithms and real-time locking protocols,respectively,and the empirical synthesis experiment is one of the major methods to compare the performance of such analyses.However,since many sophisticated techniques have been adopted to improve the analytical accuracy,the implementation of such analyses and experiments is often time-consuming.This paper proposes a schedulability experiment toolkit for multiprocessor real-time systems(SET-MRTS),which provides a framework with infrastructures to implement the schedulability and synchronization analyses and the deployment of empirical synthesis experiments.Besides,with well-designed peripheral components for the input and output,experiments can be conducted easily and flexibly on SET-MRTS.This demonstration further proves the effectiveness of SET-MRTS in both functionality and availability.展开更多
Coordinated partitioning and resource sharing have attracted considerable research interest in the field of real-time multiprocessor systems.However,finding an optimal partition is widely known as NP-hard,even for ind...Coordinated partitioning and resource sharing have attracted considerable research interest in the field of real-time multiprocessor systems.However,finding an optimal partition is widely known as NP-hard,even for independent tasks.A recently proposed resource-oriented partitioned(ROP)fixed-priority scheduling that partitions tasks and shared resources respectively has been shown to achieve a non-trivial speedup factor guarantee,which promotes the research of coordinated scheduling to a new level.Despite the theoretical elegance,the schedulability performance of ROP scheduling is restricted by the heuristic partitioning methods used in the original study.In this paper,we address the partitioning problem for tasks and shared resources under the ROP scheduling.A unified schedulability analysis framework for the ROP scheduling is proposed in the first place.A sophisticated partitioning approach based on integer linear programming(ILP)is then proposed based on the unified analysis.Empirical results show that the proposed methods improve the schedulability of ROP scheduling significantly,and the runtime complexity for searching a solution is reduced prominently compared with other ILP-based approaches as well.展开更多
基金supported by the National Natural Science Foundation of China under Grant No.61003032/F020207
文摘Graphic processing units （GPUs） have been widely recognized as cost-efficient co-processors with acceptable size, weight, and power consumption. However, adopting GPUs in real-time systems is still challenging, due to the lack in framework for real-time analysis. In order to guarantee real-time requirements while maintaining system utilization ~in modern heterogeneous systems, such as multicore multi-GPU systems, a novel suspension-based k-exclusion real-time locking protocol and the associated suspension-aware schedulability analysis are proposed. The proposed protocol provides a synchronization framework that enables multiple GPUs to be efficiently integrated in multicore real-time systems. Comparative evaluations show that the proposed methods improve upon the existing work in terms of schedulability.
基金supported by the National Natural Science Foundation of China under Grant No.61802052the Fundamental Research Funds for the Central Universities under Grant No.A030202063008085the China Postdoctoral Science Foundation Funded Project under Grant No.2017M612947。
文摘In the real-time scheduling theory,schedulability and synchronization analyses are used to evaluate scheduling algorithms and real-time locking protocols,respectively,and the empirical synthesis experiment is one of the major methods to compare the performance of such analyses.However,since many sophisticated techniques have been adopted to improve the analytical accuracy,the implementation of such analyses and experiments is often time-consuming.This paper proposes a schedulability experiment toolkit for multiprocessor real-time systems(SET-MRTS),which provides a framework with infrastructures to implement the schedulability and synchronization analyses and the deployment of empirical synthesis experiments.Besides,with well-designed peripheral components for the input and output,experiments can be conducted easily and flexibly on SET-MRTS.This demonstration further proves the effectiveness of SET-MRTS in both functionality and availability.
基金the National Natural Science Foundation of China under Grant Nos.61572112 and 61802052the Applied Basic Research Programs of Science and Technology Department in Sichuan Province of China under Grant No.2019YJ0185.
文摘Coordinated partitioning and resource sharing have attracted considerable research interest in the field of real-time multiprocessor systems.However,finding an optimal partition is widely known as NP-hard,even for independent tasks.A recently proposed resource-oriented partitioned(ROP)fixed-priority scheduling that partitions tasks and shared resources respectively has been shown to achieve a non-trivial speedup factor guarantee,which promotes the research of coordinated scheduling to a new level.Despite the theoretical elegance,the schedulability performance of ROP scheduling is restricted by the heuristic partitioning methods used in the original study.In this paper,we address the partitioning problem for tasks and shared resources under the ROP scheduling.A unified schedulability analysis framework for the ROP scheduling is proposed in the first place.A sophisticated partitioning approach based on integer linear programming(ILP)is then proposed based on the unified analysis.Empirical results show that the proposed methods improve the schedulability of ROP scheduling significantly,and the runtime complexity for searching a solution is reduced prominently compared with other ILP-based approaches as well.
