Papers by Patrick Meumeu Yomsi
River Publishers eBooks, Sep 1, 2022
This chapter surveys state-of-the-art manycore platforms. It discusses the historical evolution o... more This chapter surveys state-of-the-art manycore platforms. It discusses the historical evolution of computing platforms over the past decades and the technical hurdles that led to the manycore revolution, then presents in details several manycore platforms, outlining (i) the key architectural traits that enable scalability to several tens or hundreds of processing cores and (ii) the shared resources that are responsible for unpredictable timing.
On the Extension of Single-core Thermal Models to Multi-core Platforms in Critical Real-Time Domains
Reducing Peak Temperature by Redistributing Idle-Time in Modern MPSoCs
An Efficient Proactive Thermal-Aware Scheduler for DVFS-enabled Single-Core Processors
For decades now, thermal rise has been spotted as one of the major constraints of performance for... more For decades now, thermal rise has been spotted as one of the major constraints of performance for high-end safety-critical processors. In this context, Dynamic Voltage and Frequency Scaling () based solutions have proven to be effective to manage the chip temperature. In this paper, we consider the scheduling problem of non-preemptive periodic tasks on a single-core processor with DVFS-enabled capabilities under thermal-aware design. We assume that the tasks are scheduled by following any Fixed-Task-Priority (FTP) scheduler such as the traditional Rate Monotonic (RM) and Deadline Monotonic (DM). Then, we propose a new scheduling scheme, referred to as NP-COIN, which makes it possible to control both the processor activity and the triggering of the cooling mechanism with as little impact on performance as possible. We provide a thorough theoretical analysis of our solution, in terms of average temperature gain and timing penalty, against the classical schedule. Finally, we validate our theoretical results and assess the performance of our solution through a real-world use-case study from the avionics domain and through intensive simulations by using synthetic test cases.
Journal of Systems Architecture, Jun 1, 2021
To achieve low latency transmission of time-sensitive flows in Ethernet networks, the IEEE introd... more To achieve low latency transmission of time-sensitive flows in Ethernet networks, the IEEE introduced the IEEE 802.1Qbu Standard, which specifies a 1-level preemption scheme for IEEE 802.1 networks. The specification allows the suspension of a preemptable frame prior to its completion for the speedy transmission of an express frame; but any other preemptable frame cannot be transmitted before the completion of the already preempted frame. While this approach improves the performance of express frames, the performance is negatively impacted in scenarios where the number of express frames is high. Another limitation is the fact that preemptable frames with timing requirements can suffer long blocking periods due to the non-preemptive service of frames in the same category. This is irrespective of the individual priority level of each frame. Recently, a multi-level preemption scheme has been proposed to circumvent these limitations. The work focused on the feasibility and implementation requirement of such an approach, but a formal analysis of the worst-case performance guarantees under the proposed scheme was not provided. In this paper, we fill this gap by presenting the aforementioned analysis of the TSN IEEE 802.1Qbu networks under the multi-level preemption assumption. Evaluation is performed with a realistic automotive use-case and the results showcase an improvement of up to 53:07% for preemptable frames with firm timing requirements.
Energy consumption is a major concern in modern real-time embedded systems and leakage current is... more Energy consumption is a major concern in modern real-time embedded systems and leakage current is a main contributor to it. To deal with the leakage current, several procrastination approaches have been proposed in the past in order to reduce the energy consumption. These approaches approximate the procrastination interval for the ease of analysis and sub-optimally utilise the potential to reduce the energy consumption. This paper presents an optimal method to determine the procrastination interval of each task and generalise the task-model that also covers the constrained deadline tasks. Analytical and experimental results show the superiority of the proposed techniques. In the best case, the proposed technique extends the average sleep interval up to 75% and decrease the energy consumption in idle state up to 55% over the state-of-the-art.
River Publishers eBooks, Sep 1, 2022
This chapter focuses on the analysis of the timing behavior of software applications that expose ... more This chapter focuses on the analysis of the timing behavior of software applications that expose real-time (RT) requirements. The state-of-the-art methodologies to timing analysis of software programs are generally split into four categories, referred to as static, measurement-based, hybrid, and probabilistic analysis techniques. First, we present an overview of each of these methodologies and discuss their advantages and disadvantages. Next, we explain the choices made by our proposed methodology in Section 5.2 and present the details of the solution in Section 5.3. Finally, we conclude the chapter in Section 5.4 with a summary.
Multi-Level Preemption in TSN: Feasibility and Requirements Analysis
To overcome the limitation of strictly non-preemptive frame transmission in Ethernet networks, th... more To overcome the limitation of strictly non-preemptive frame transmission in Ethernet networks, the IEEE 802.1Qbu standard was introduced. This standard specifies a one-level frame preemption paradigm wherein, depending on their priority levels, frames are grouped into two categories: namely, the “express frames” and the “preemptable frames”. These two categories are given with the interpretation that (1) only express frames can preempt preemptable frames; and (2) two frames belonging to the same category cannot preempt each other. While this approach partially solves the problem, some preemptable frames can still suffer long blocking periods, irrespective of their individual priority levels. Indeed, there are frames that do not fall into the express frames category, but nevertheless have firm timing requirements that can only be met if they can benefit from preempting lower priority frames. To ameliorate the condition of such frames, we propose a multi-level preemption paradigm. Specifically, we expose the limitations of the one-level preemption approach experimentally; and we present the feasibility and implementation requirements of the multi-level preemption scheme in details.
SIGBED review, Jan 6, 2020
A carefully designed routing synthesis can help system designers achieve a better load balancing ... more A carefully designed routing synthesis can help system designers achieve a better load balancing in TSN-based networks and avoid congestion. To this end purpose, this work proposes two heuristics referred to as (1) LB-DRR, which aims at achieving a better load balancing and compute as much disjoint routing paths as possible for each replicated flow; and (2) CR-DRR, which recomputes paths for timesensitive flows in congestion situations. Extensive simulations demonstrate that the proposed approach outperforms the classical Shortest Path (SPA) and the weighted Equal Cost Multi-path (wt-ECMP) algorithms in terms of the maximum load transmitted on a link by more than 70% and 20%, respectively.
Euromicro Conference on Real-Time Systems, Jul 3, 2018
Ethernet is increasingly being considered as thesolution to high bandwidth requirements in the ne... more Ethernet is increasingly being considered as thesolution to high bandwidth requirements in the next generation oftiming critical applications that make their way in cars, planesor smart factories to mention a few examples. Until recently,ethernet frames used to be transmitted exclusively in a nonpreemptive manner. That is, once a frame starts transmitting ona switch output port, its transmission cannot be interrupted byany other frame until completion. This constraint may causetime critical frames to be blocked for long periods of timebecause of the transmission of non-critical frames. The IEEE802.3br standard addressed this issue by introducing a onelevelethernet frame preemption paradigm. In this approach, framestransmitted through a switch output port are classified as expressframes or preemptable frames, depending on their priority levels.Express frames can preempt preemptable frames and two framesbelonging to the same class cannot preempt each other. While thispartially solves the problem for express frames, all preemptableframes can still suffer blocking irrespective of their priority level.In this work, we investigate the feasibility and advantages ofmulti-level preemptions in timesensitive ethernet networks.
The advent of multi-core platforms in critical realtime domains such as the avionics, automotive ... more The advent of multi-core platforms in critical realtime domains such as the avionics, automotive and railways to achieve higher and higher computing performances has turned the view on thermal concerns of the underlying chip die while it is still mandatory to meet all the temporal constraints. As a matter of fact, high chip temperature may not only degrade system performance and reliability, but it may also damage the chip permanently. We propose a methodology to address this problem, based on fixed task-to-core mapping and per-core analysis to derive a sound system model without feedback loops. To this end, it is important to have a better and deeper understanding of the existing thermal models in the literature. This is the main contribution of this research.
Thermal management is a crucial aspect of the design and operation of safety-critical multi-core ... more Thermal management is a crucial aspect of the design and operation of safety-critical multi-core architectures, as their high power density can cause significant heat generation and risk of thermal overload. If not properly managed, thermal overload can lead to system failures and performance degradation, which is a major challenge for system designers. To address this challenge, advanced core mapping solutions have become increasingly popular in both industry and academia. In this paper, we present key insights, techniques and results on thermal management in multi-core architectures. We propose a new per-core power budget strategy called B-TSP that is scalable and enables system performance optimization while abstracting from mapping concerns. In addition, we present a new strategy called FBF that allows us to derive worst-case mappings as a function of the number of active cores from a power consumption perspective in a thermal-aware design. We demonstrate the effectiveness of our solution through intensive simulations with the homogeneous AMD EPYC 7351 16-cores platform. CCS CONCEPTS • Hardware → Thermal issues; • Computer systems organization → Embedded systems; Real-time systems.
A Thermal-Aware Approach for DVFS-enabled Multi-core Architectures
2022 IEEE 24th Int Conf on High Performance Computing & Communications; 8th Int Conf on Data Science & Systems; 20th Int Conf on Smart City; 8th Int Conf on Dependability in Sensor, Cloud & Big Data Systems & Application (HPCC/DSS/SmartCity/DependSys), Dec 1, 2022
River Publishers eBooks, Jul 1, 2018
This chapter focuses on the analysis of the timing behavior of software applications that expose ... more This chapter focuses on the analysis of the timing behavior of software applications that expose real-time (RT) requirements. The state-of-the-art methodologies to timing analysis of software programs are generally split into four categories, referred to as static, measurement-based, hybrid, and probabilistic analysis techniques. First, we present an overview of each of these methodologies and discuss their advantages and disadvantages. Next, we explain the choices made by our proposed methodology in Section 5.2 and present the details of the solution in Section 5.3. Finally, we conclude the chapter in Section 5.4 with a summary.
arXiv (Cornell University), Feb 10, 2011
Multi-mode real-time systems are those which support applications with different modes of operati... more Multi-mode real-time systems are those which support applications with different modes of operation, where each mode is characterized by a specific set of tasks. At run-time, such systems can, at any time, be requested to switch from its current operating mode to another mode (called "new mode") by replacing the current set of tasks with that of the new-mode. Thereby, ensuring that all the timing requirements are met not only requires that a schedulability test is performed on the tasks of each mode but also that (i) a protocol for transitioning from one mode to another is specified and (ii) a schedulability test for each transition is performed. We propose two distinct protocols that manage the mode transitions upon uniform and identical multiprocessor platforms at run-time, each specific to distinct task requirements. For each protocol, we formally establish schedulability analyses that indicate beforehand whether all the timing requirements will be met during any mode transition of the system. This is performed assuming both Fixed-Task-Priority and Fixed-Job-Priority schedulers.
2020 16th IEEE International Conference on Factory Communication Systems (WFCS)
The rising adoption of wireless technologies in the Industry 4.0, including the Industrial Intern... more The rising adoption of wireless technologies in the Industry 4.0, including the Industrial Internet of Things (IIoT), has stressed the need for traffic schedulability validation at system design-time. In this context, the demandbased schedulability tests have recently been proposed in the literature. This work revisits two well-established techniques borrowed from the multi-processor scheduling theory, namely the demand-bound-function (DBF) and the forced-forward-demand-bound-function (FFDBF), and evaluates their performances when adapted to the field of wireless sensor-actuator networks. Simulation experiments when varying network configurations confirm the equal or better accuracy of FFDBF over DBF to estimate both network demand and schedulability. In future work, we aim at building upon these promising results in order to design novel admission control and adaptation strategies that improve network schedulability under varying workload conditions.
A Configuration Framework for Multi-level Preemption Schemes in Time Sensitive Networking
Proceedings of the 30th International Conference on Real-Time Networks and Systems
Multi-Level Preemption in TSN: Feasibility and Requirements Analysis
2020 IEEE 23rd International Symposium on Real-Time Distributed Computing (ISORC), 2020
To overcome the limitation of strictly non-preemptive frame transmission in Ethernet networks, th... more To overcome the limitation of strictly non-preemptive frame transmission in Ethernet networks, the IEEE 802.1Qbu standard was introduced. This standard specifies a one-level frame preemption paradigm wherein, depending on their priority levels, frames are grouped into two categories: namely, the “express frames” and the “preemptable frames”. These two categories are given with the interpretation that (1) only express frames can preempt preemptable frames; and (2) two frames belonging to the same category cannot preempt each other. While this approach partially solves the problem, some preemptable frames can still suffer long blocking periods, irrespective of their individual priority levels. Indeed, there are frames that do not fall into the express frames category, but nevertheless have firm timing requirements that can only be met if they can benefit from preempting lower priority frames. To ameliorate the condition of such frames, we propose a multi-level preemption paradigm. Specifically, we expose the limitations of the one-level preemption approach experimentally; and we present the feasibility and implementation requirements of the multi-level preemption scheme in details.
New Results - Real-time scheduling
Uploads
Papers by Patrick Meumeu Yomsi