The Journal of Supercomputing

BALANCER: Bandwidth Allocation and Cache Partitioning for Multicore Processors

Agustín Navarro Torres1 Jesús Alastruey-Benedé1
Pablo Ibáñez1 Víctor Viñals1
1Dept. Informática e Ingeniería de Sistemas - I3A, Universidad de Zaragoza, Spain

News

Abstract

The management of shared resources in multicore processors is an open problem due to the continuous evolution of these systems. The trend towards increasing the number of cores and organizing them in clusters sets out new challenges not considered in previous works. In this paper, we characterize the use of the shared cache and memory bandwidth of an AMD Rome processor executing multiprogrammed workloads and propose several mechanisms that control the use of these resources to improve the system performance and fairness. Our control mechanisms require no hardware or operating system modifications. We evaluate Balancer on a real system running SPEC CPU2006 and CPU2017 applications. Balancer tuned for performance shows an average increase of 7.1% in system performance and an unfairness reduction of 18.6% with respect to a system without any control mechanism. Balancer tuned for fairness decreases the performance by 1.3% in exchange for a 64.5% reduction of unfairness.

Downloads

Bibtex

@Article{Navarro-Torres2023,
	author={Navarro-Torres, Agust{\'i}n and Alastruey-Bened{\'e}, Jes{\'u}s and Ib{\'a}{\~{n}}ez, Pablo and Vi{\~{n}}als-Y{\'u}fera, V{\'i}ctor},
	title={BALANCER: bandwidth allocation and cache partitioning for multicore processors},
	journal={The Journal of Supercomputing},
	year={2023},
	month={Jun},
	day={01},
	volume={79},
	number={9},
	pages={10252-10276},
	abstract={The management of shared resources in multicore processors is an open problem due to the continuous evolution of these systems. The trend toward increasing the number of cores and organizing them in clusters sets out new challenges not considered in previous works. In this paper, we characterize the use of the shared cache and memory bandwidth of an AMD Rome processor executing multiprogrammed workloads and propose several mechanisms that control the use of these resources to improve the system performance and fairness. Our control mechanisms require no hardware or operating system modifications. We evaluate Balancer on a real system running SPEC CPU2006 and CPU2017 applications. Balancer tuned for performance shows an average increase of 7.1{\%} in system performance and an unfairness reduction of 18.6{\%} with respect to a system without any control mechanism. Balancer tuned for fairness decreases the performance by 1.3{\%} in exchange for a 64.5{\%} reduction of unfairness.},
	issn={1573-0484},
	doi={10.1007/s11227-023-05070-0},
	url={https://doi.org/10.1007/s11227-023-05070-0}
}

Acknowledgements

This work was supported by MCIN/AEI/10.13039/501100011033 and by “ERDF A way of making Europe” (grant PID2019-105660RB-C21), and by Government of Aragón (T5820R research group).