Tightening Contention Delays While Scheduling Parallel Applications on Multi-core Architectures

Abstract

Multi-core systems are increasingly interesting candidates for executing parallel real-time applications, in avionic, space or automotive industries, as they provide both computing capabilities and power efficiency. However, ensuring that timing constraints are met on such platforms is challenging, because some hardware resources are shared between cores.Assuming worst-case contentions when analyzing the schedulability of applications may result in systems mistakenly declared unschedulable, although the worst-case level of contentions can never occur in practice. In this paper, we present two contention-aware scheduling strategies that produce a time-triggered schedule of the application's tasks. Based on knowledge of the application's structure, our scheduling strategies precisely estimate the effective contentions, in order to minimize the overall makespan of the schedule. An Integer Linear Programming (ILP) solution of the scheduling problem is presented, as well as a heuristic solution that generates schedules very close to ones of the ILP (5 % longer on average), with a much lower time complexity. Our heuristic improves by 19% the overall makespan of the resulting schedules compared to a worst-case contention baseline.

Accurate ILP-Based Contention Modeling on Statically Scheduled Multicore Systems

“…For example, a given task access cannot suffer contention from two accesses performed by two distinct tasks, mapped to the same contender core. These constraints are improving the tightness of the computed WCD, and are one of the aspects that differentiate our approach from related works [44], [36], [21]. In particular, Eq.…”

“…Other works explore the use of ILP for the characterization of contention effect, especially tailored to domain-specific platforms in the automotive [11], [21] and parallel computing [44], [36] application domains. The work in [11] uses an ILP formulation to schedule tasks with phased execution (as in [6]).…”

“…The ILP formulation presented in [21] models contention delays between pairs of tasks, without any consideration on whether the addressed tasks were overlapping, or on the system-level view of conflictive accesses. Authors in [44] propose contention-aware time-triggered scheduling strategies using some knowledge of the application's structure to find a schedule that minimizes contention under the readexecute-write semantics assumption. We differentiate from these approaches in that we do not seek for optimal schedules and, instead, we assume both static schedule and task-to-core mapping.…”

“…The static schedule is defined by a static allocation of tasks to MIFs that execute in sequence within the MAF. Accordingly, we are not interested in finding an optimal task schedule and/or assignment of tasks to cores [44]. The baseline notation w.r.t.…”

Timely Fine-Grained Interference-Sensitive Run-Time Adaptation of Time-Triggered Schedules

“…The majority of existing works on task scheduling and allocation using isWCET (interference-sensitive task scheduling and allocation), e.g. [6], [7], uses time-triggered execution so as to guarantee that the task scheduling and allocation solution remains unchanged during execution. In time-triggered executions, the tasks are executed exactly at their start times provided by a solution given by an offline scheduling and allocation algorithm.…”