SciELO - Scientific Electronic Library Online

 
vol.28 número2On the Performance Assessment and Comparison of Features Selection ApproachesRocket Thrust Vectoring Attitude Control based on Convolutional Neural Networks índice de autoresíndice de assuntospesquisa de artigos
Home Pagelista alfabética de periódicos  

Serviços Personalizados

Journal

Artigo

Indicadores

Links relacionados

  • Não possue artigos similaresSimilares em SciELO

Compartilhar


Computación y Sistemas

versão On-line ISSN 2007-9737versão impressa ISSN 1405-5546

Resumo

REVILLA-DUARTE, Ulises; RAMIREZ-SALINAS, Marco A.; VILLA-VARGAS, Luis A.  e  TCHERNYKH, Andrei. Proactive Load Balancing to Reduce Unnecessary Thread Migrations on Chip Multi-Processor (CMP) Systems. Comp. y Sist. [online]. 2024, vol.28, n.2, pp.623-645.  Epub 31-Out-2024. ISSN 2007-9737.  https://doi.org/10.13053/cys-28-2-4403.

For a Linux operating system scheduler that is aware of Chip Multi-Processor (CMP) systems to carry out load balancing is extremely important and quite challenging. The scheduler is a vital component of the Linux kernel responsible for choosing the next thread to run and allocating to a processor core for execution. This process involves primarily a load-balancing procedure that provides the thread migration between the cores of a CMP system. A modern Linux scheduler is designed to obtain the best possible performance while ensuring a fair allocation of the processor cores’ time among the normal (non-real-time) threads, which is known as Completely Fair Scheduling (CFS) policy. However, this policy collaterally can cause a relentless execution of the load-balancing procedure, and therefore, an excessive number of thread migrations. According to the literature, an increased cache invalidation, scheduling latency, and power consumption are issues inherent to this. In this paper, we propose and evaluate a proactive load-balancing (PLB) algorithm to reduce unnecessary thread migrations on CMP systems. By comprehensive experimental analysis, we show that our PLB algorithm reduces the number of thread migrations by 43.8% on average without degradation of performance.

Palavras-chave : Linux CFS; load balancing; perf event tool; PMU counters; chip multi-processor.

        · texto em Inglês     · Inglês ( pdf )