Symbiotic job scheduling on the IBM POWER8

Feliu, Josué; Eyerman, Stijn; Sahuquillo, Julio; Petit, Salvador

doi:10.1109/hpca.2016.7446103

Cited by 15 publications

(27 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Other works have studied job symbiosis as a way to rise performance by co-scheduling jobs with compatible demands on the shared resources. In recent approaches, Eyerman et al [22], [23] and Feliu et al [24] predict how the interference among processes affects the performance of different co-schedules without actually running them, which is used to select the co-schedule with highest performance on each quantum. With the same goal, but focused on L1 bandwidth, Feliu et al [13] propose a process allocation policy to balance the overall L1 requests across all the L1 caches of the processor.…”

Section: Related Workmentioning

confidence: 99%

Perf&Fair: A Progress-Aware Scheduler to Enhance Performance and Fairness in SMT Multicores

Feliu

Sahuquillo

Petit

et al. 2017

IEEE Trans. Comput.

Self Cite

View full text Add to dashboard Cite

Nowadays, high performance multicore processors implement multithreading capabilities. The processes running concurrently on these processors are continuously competing for the shared resources, not only among cores, but also within the core. While resource sharing increases the resource utilization, the interference among processes accessing the shared resources can strongly affect the performance of individual processes and its predictability. In this scenario, process scheduling plays a key role to deal with performance and fairness. In this work we present a process scheduler for SMT multicores that simultaneously addresses both performance and fairness. This is a major design issue since scheduling for only one of the two targets tends to damage the other. To address performance, the scheduler tackles bandwidth contention at the L1 cache and main memory. To deal with fairness, the scheduler estimates the progress experienced by the processes, and gives priority to the processes with lower accumulated progress. Experimental results on an Intel Xeon E5645 featuring six dual-threaded SMT cores show that the proposed scheduler improves both performance and fairness over two state-of-the-art schedulers and the Linux OS scheduler. Compared to Linux, unfairness is reduced to a half while still improving performance by 5.6%.

show abstract

Section: Related Workmentioning

confidence: 99%

Perf&Fair: A Progress-Aware Scheduler to Enhance Performance and Fairness in SMT Multicores

Feliu

Sahuquillo

Petit

et al. 2017

IEEE Trans. Comput.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Previous work on symbiotic job scheduling for SMT uses sampling to explore the space of possible schedules [24], relies on novel hardware support [8], or performs an offline analysis to predict the interference between applications on an SMT core [27]. In contrast to these works, we propose an online modelbased scheduler [10] that does not require sampling and can be used on an existing commercial processor. The devised scheduler leverages the existing CPI stack accounting mechanism on the IBM POWER8 to build a model that predicts the interference among threads on an SMT core.…”

Section: Introductionmentioning

confidence: 99%

“…In this work we extend our symbiotic job scheduler [10], making the following contributions. 1) We redefine the construction of the model such that the interference of applications running on the same SMT core is predicted more accurately.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Improving IBM POWER8 Performance Through Symbiotic Job Scheduling

Feliu

Eyerman

Sahuquillo

et al. 2017

IEEE Trans. Parallel Distrib. Syst.

Self Cite

View full text Add to dashboard Cite

Symbiotic job scheduling, i.e., scheduling applications that co-run well together on a core, can have a considerable impact on the performance of processors with simultaneous multithreading (SMT) cores. SMT cores share most of their microarchitectural components among the co-running applications, which causes performance interference between them. Therefore, scheduling applications with complementary resource requirements on the same core can greatly improve the throughput of the system. This paper enhances symbiotic job scheduling for the IBM POWER8 processor. We leverage the existing cycle accounting mechanism to build an interference model that predicts symbiosis between applications. The proposed models achieve higher accuracy than previous models by predicting job symbiosis from throttled CPI stacks, i.e., CPI stacks of the applications when running in the same SMT mode to consider the statically partitioned resources, but without interference from other applications. The symbiotic scheduler uses these interference models to decide, at run-time, which applications should run on the same core or on separate cores. We prototype the symbiotic scheduler as a user-level scheduler in the Linux operating system and evaluate it on an IBM POWER8 server running multiprogram workloads. The symbiotic job scheduler significantly improves performance compared to both an agnostic random scheduler and the default Linux scheduler. Across all evaluated workloads in SMT4 mode, throughput improves by 12.4% and 5.1% on average over the random and Linux schedulers, respectively.

show abstract

“…The work discussed in this chapter has been published in [75]. In this chapter, the main contributions of these proposals are summarized, followed by a discussion about future work, and an enumeration of the scientific publications related with this dissertation.…”

Section: Discussionmentioning

confidence: 99%

Contention-Aware Scheduling for SMT Multicore Processors

Pérez¹

View full text Add to dashboard Cite

iii AgraïmentsXicotetes decisions, aparentment intranscendents, acaben marcant una vida. En la meua, una d'aquestes decisions fou entrar al despatx de Julio un 15 de juliol de 2010.Simplement buscava informació per a fer un projecte final de carrera i poc més que vaig eixir amb un doctorat planificat.En primer lloc aquesta tesi té molt que agrair a Julio i Salva. Per confiar en mi en aquelles primeres reunions de 2010 quan no era més que un estudiant d'enginyeria informàtica.Per tot el que m'han ensenyat sobre l'arquitectura de computadors i la investigació. I per tot el treball intensíssim que ha portat aquesta tesi i, principalment, les publicacions en les que es basa. Moltes hores extres de treball repartides entre nits, caps de setmana i vacances.És una obvietat que sense ells aquesta tesi no existiria.En segon lloc vuic agrair la col·laboració a la resta de professors que han contribuït al desenvolupament d'aquesta tesi. A José Duato, per la seva col·laboració en part dels articles, així com per ajudar-nos a obtenir la beca amb la que he realitzat aquest doctorat. A Stijn Eyerman, per l'acollida tan bona que vaig tenir en la Ghent University, així com per la seva enorme contribució en la publicació més important d'aquesta tesi. I a Lieven Eeckhout, pels seus precisos comentaris i col·laboració en part del treball realitzat.A continuació vuic agrair a tots els companys que he tingut durant el desenvolupament de la tesi al grup d'arquitectures paral·leles, destacant aquells amb els que he compartit més dinars i debats. Amb un ambient de treball tan bo es rendeix molt millor. També agrair als companys de la Ghent University que em feren sentir com a casa.Perúltim, però no menys important, a la meua família pel suport que m'han donat. En especial a ma mare, per exigir sempre arribar un poquet més lluny, i a la qui segurament també he d'agrair-li totes les hores que va dedicar als meus estudis quan jo era ben menut. A la meua novia Amparo, per recolzar-me sempre (incloent les dos estades a Ghent), així com per aguantar-me en els moment de major estrés, on ha sigut complicat separar recerca i vida. I finalment als meus amics, que sense saber-ho han contribuït a que puga desconnectar el cap de setmana i començar cada dilluns amb forces.v

show abstract

Symbiotic job scheduling on the IBM POWER8

Cited by 15 publications

References 19 publications

Perf&Fair: A Progress-Aware Scheduler to Enhance Performance and Fairness in SMT Multicores

Perf&Fair: A Progress-Aware Scheduler to Enhance Performance and Fairness in SMT Multicores

Improving IBM POWER8 Performance Through Symbiotic Job Scheduling

Contention-Aware Scheduling for SMT Multicore Processors

Contact Info

Product

Resources

About