Energy-Efficient Thread Assignment Optimization for Heterogeneous Multicore Systems

Petrucci, Vinícius; Loques, Orlando; Mossé, Daniel; Melhem, Rami; Gazala, Neven Abou; Gobriel, Sameh

doi:10.1145/2566618

Cited by 44 publications

(51 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Petrucci [12] also optimizes Integer Linear programming. Alhammad [13] presents Fork-join parallel model, making the most of divider and conquer.…”

Section: B Dynamic Modelingmentioning

confidence: 99%

Task-Aware Priority Scheduling for Multicore Processors

Shi¹

2016

Proceedings of the 2016 International Conference on Computer Engineering and Information Systems

View full text Add to dashboard Cite

Abstract-Task scheduling is a key issue in multicore processors, especially for the multicore processor in parallel framework. However, few state-of-art task scheduling strategies solve these important issues simultaneously: assignment of task to CPU core, task types, task order and resource conflict of task. To the best of our knowledge, the schedule problem is more complicated in the node of parallel framework. To settle this problem, we propose a method of task-aware priority scheduling for multicore processors. First, we introduce our scheduling framework, including four modules: Kernel Scheduler, Resource Management, Synchronization, Node Coordinator. Then follow two key algorithms: Task-to-core assignment algorithm and Resource distribution algorithm. Finally, we show that our method obtains good performance in contrast with Work-stealing algorithm.

show abstract

“…Petrucci [12] also optimizes Integer Linear programming. Alhammad [13] presents Fork-join parallel model, making the most of divider and conquer.…”

Section: B Dynamic Modelingmentioning

confidence: 99%

Task-Aware Priority Scheduling for Multicore Processors

Shi¹

2016

Proceedings of the 2016 International Conference on Computer Engineering and Information Systems

View full text Add to dashboard Cite

show abstract

“…Any modern Operating System (OS) encompasses instructions to implement parallel and distributed applications [2]. In every Embedded System (ES) platform, the use of threads is crucial due to the efficient handling of scarce resources allowing faster executions in a reasonable fashion [3]. The scheduling policy is a recurrent issue in OS implementation because it affects the execution efficiency directly [4].…”

Section: Introductionmentioning

confidence: 99%

A Framework Prototype for Multithreaded Implementation Over Micro-Controllers

Nagel¹,

Czekster

Webber³

et al. 2019

JICS

View full text Add to dashboard Cite

Multithreading is pervasive in embedded system applications development. The applications requirements are becoming more rigorous, demanding the execution of concurrent tasks that must also take into account modularity and flexibility. An important part of the operating systems development concerns the implementation of scheduling algorithms. In an embedded system context, it is essential to consider that the scheduling algorithm heavily influences application behavior. Due to restricted and finite hardware resources, it is important to evaluate the use of flexible algorithms to guarantee efficiency. Currently, projects for embedded operating systems do exist for microcontrollers’ devices that implement scheduling algorithms, however, the developer cannot change or add new scheduling policies without implementing kernel tweaks and modifications. The alternatives are not flexible when choosing the scheduling algorithm according to the application needs. This imposes restrictions to many systems, forcing them to run specific static scheduling algorithms because no other options are available. The objective of this work concerns the design and development of a framework that implements a microkernel with a modular scheduler unit, allowing the execution of tailored algorithms according to the application profile. The idea is to provide a flexible platform to conveniently select the most appropriated algorithm. We have employed low capacity hardware to implement multithreading patterns corresponding to sets of concurrent tasks, demonstrating the strengths of adopting our approach. Our results show that the use of modern techniques that combine modularity, multithreading, and scheduling methods for embedded systems yield best executions when compared to its sequential counterparts.

show abstract

“…With strict thermal and power budgets, the focus is shifting towards power-aware thread mapping on multicore platforms (e.g. [3]). Most of these approaches use linear programming or heuristics to generate an energy minimum thread mapping considering single thread execution on a core at any given time (i.e., no SMT).…”

Section: Introductionmentioning

confidence: 99%

The Slowdown or Race-to-idle Question: Workload-Aware Energy Optimization of SMT Multicore Platforms under Process Variation

Das

Merrett

Al-Hashimi

2016

Proceedings of the 2016 Design, Automation &Amp; Test in Europe Conference &Amp; Exhibition (DATE)

View full text Add to dashboard Cite

Abstract-Two widely used approaches for reducing energy consumption in multithreaded workloads are slowdown (using DVFS) and race-to-idle. In this paper, we first demonstrate that most energy-efficient choice is dependent on (1) workload (memory bound, CPU bound etc.), (2) process variation and (3) support for Simultaneous Multithreading (SMT). We then propose an approach for mapping application threads on SMT multicore systems at run-time, to minimize energy consumption. The proposed approach interfaces with the OS and hardware performance counters to characterize application threads. This characterization captures the effect of process variation on execution time and identifies the break-even operating point, where one strategy (slowdown or race-to-idle) outperforms the other. Thread mapping is performed using these characterized data by iteratively collapsing application threads (SMT) followed by binary programming-based thread mapping. Finally, performance slack is exploited at run-time to select between slowdown and race-to-idle, based upon the break-even operating point calculated for each individual thread. This end-to-end approach is implemented as a run-time manager for the Linux OS and is validated across a range of high performance applications. Results demonstrate up to 13% energy reduction over all state-of-the-art approaches, with an average of 18% improvement over Linux.

show abstract

Energy-Efficient Thread Assignment Optimization for Heterogeneous Multicore Systems

Cited by 44 publications

References 21 publications

Task-Aware Priority Scheduling for Multicore Processors

Task-Aware Priority Scheduling for Multicore Processors

A Framework Prototype for Multithreaded Implementation Over Micro-Controllers

The Slowdown or Race-to-idle Question: Workload-Aware Energy Optimization of SMT Multicore Platforms under Process Variation

Contact Info

Product

Resources

About