Safe and Efficient Deployment of Data-Parallelizable Applications on Many-Core Platforms: Theory and Practice

Abstract

The safe and, at the same time, efficient deployment of parallelisable applications on many-core platforms is a challenging task. Theoretical Models of Computation (MoC) require the realistic estimation of task Worst-Case Execution Time (WCET) to provide safe latency guarantees. Due to interferences on shared resources, task WCET estimations are often exceedingly pessimistic. In reality, though, rarely do all the tasks execute with their WCET, thus introducing an efficiency gap, which is of consequence in realizing safety-critical and mixed-criticality systems. In this paper, we outline the additional research efforts required to i) derive a safe deployment from a MoC reducing that efficiency gap and ii) adapt at runtime to further improve performance and still preserve safety. We also outline the impact of the level of data-parallelisation onto this efficiency gap and present experimental evidence of the performance improvements from accurate WCET estimation, level of data-parallelisation and runtime adaptation.

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

“…From the obtained results, we observed a significant performance gain compared with the global synchronisation mechanism [9] as the number of cores increases. When a single application is parallelized and executed on several cores, the observed gain is on average 1.64%, using only two cores, up to 37.32%, when eight cores are used.…”

“…The impact is illustrated in Figure 2b, where the unnecessary blocking (rectangle wait) is eliminated and the control phases are overlapping. The proposed isRA-FG improves over isRA-GLO [9] both in terms of i) the WCET of the controller and ii) run-time performance. • RTA-based formal proof that execution of time-triggered schedules with isRA-FG does not introduce additional interference and is free of timing anomalies.…”

“…The work in [9] allows run-time adaptation of interferencesensitive time-triggered schedules that preserves the timing guarantees. It inserts extra scheduling dependencies to the task scheduling and allocation solution to prohibit any additional task overlapping in case of run-time re-scheduling.…”

“…To further reduce the impact of the inherent pessimism in any kind of WCET estimations, several run-time mechanisms have been proposed, e.g. [15], [9]. In [9], the authors provide a run-time approach suitable for interference-sensitive WCET estimation.…”

“…[15], [9]. In [9], the authors provide a run-time approach suitable for interference-sensitive WCET estimation. However, a single global synchronization mechanism is considered that serializes the run-time mechanism having a negative impact in performance.…”