Towards Holistic Analysis for Fork-Join Parallel/Distributed Real-Time Tasks

Parallel/distributed processing is a solution for providing scaling computing power for computational-intensive applications. Parallel/distributed applications are commonly based on the fork-join model, where precedence constraints have to be considered on the development of an adequate timing analysis. Moreover, as the main difference with multicore architectures, distributed systems entail the transfer of messages upon a communication network that should be integrated in the timing analysis. In this context, this paper presents the current status of the work towards holistic analysis for fixed priority fork-join parallel/distributed tasks. This analysis takes into consideration the interactions between parallel threads and their respective messages. These considerations will be helpful for the improvement of the combination of existing results for computing the worstcase response time and the specific case of fork-join parallel/distributed real-time tasks. Towards Holistic Analysis for Fork-Join Parallel/Distributed Real-Time Tasks Ricardo Garibay-Martinez 1 , Luis Lino Ferreira 1 , Geoffrey Nelissen 1 , Paulo Pedreiras 2 , Luis Miguel Pinho 1 1 CISTER/INESC-TEC, ISEP, Porto, Portugal 2 DETI/IT/University of Aveiro, Aveiro, Portugal 1 {rgmaz, llf, grrpn, lmp}@isep.ipp.pt; 2 pbrp@ua.pt Abstract—Parallel/distributed processing is a solution for providing scaling computing power for computational-intensive applications. Parallel/distributed applications are commonly based on the fork-join model, where precedence constraints have to be considered on the development of an adequate timing analysis. Moreover, as the main difference with multicore architectures, distributed systems entail the transfer of messages upon a communication network that should be integrated in the timing analysis. In this context, this paper presents the current status of the work towards holistic analysis for fixed priority fork-join parallel/distributed tasks. This analysis takes into consideration the interactions between parallel threads and their respective messages. These considerations will be helpful for the improvement of the combination of existing results for computing the worst-case response time and the specific case of fork-join parallel/distributed real-time tasks.Parallel/distributed processing is a solution for providing scaling computing power for computational-intensive applications. Parallel/distributed applications are commonly based on the fork-join model, where precedence constraints have to be considered on the development of an adequate timing analysis. Moreover, as the main difference with multicore architectures, distributed systems entail the transfer of messages upon a communication network that should be integrated in the timing analysis. In this context, this paper presents the current status of the work towards holistic analysis for fixed priority fork-join parallel/distributed tasks. This analysis takes into consideration the interactions between parallel threads and their respective messages. These considerations will be helpful for the improvement of the combination of existing results for computing the worst-case response time and the specific case of fork-join parallel/distributed real-time tasks. Keywords—Real-time; parallel execution; distributed systems; holistic analysis.