Using OS Observations to Improve Performance in Multicore Systems

Knauerhase, Rob; Brett, Paul; Hohlt, Barbara; Li, Tong; Hahn, Scott

doi:10.1109/mm.2008.48

Cited by 180 publications

(150 citation statements)

References 4 publications

Supporting

Mentioning

147

Contrasting

Unclassified

Order By: Relevance

“…System software also does not have control over hardware resources such as caches and memory bandwidth, which renders responding to contention and managing applications' QoS quite challenging. As a result, despite the amount of research attention given to contention problems on multicore platforms [28,45,10,19,44,51,52,27,32,15,29,24,63,23,24,7,4,60], mitigating the impact of contention on an application's performance and QoS, enforcing the relative QoS priorities of co-running applications, while maximizing machine utilization, remains key challenges in modern warehouse scale computers.…”

Section: Mitigating Contentionmentioning

confidence: 99%

“…[Contention Aware Scheduling] A research direction that recently attracted a wealth of attention is contention aware scheduling [15,29,24,63,23,24,7,4,60]. Contention aware scheduling techniques use predictors or models to decide what applications should be co-running together to minimize the performance degradation or to improve performance isolation.…”

Section: Software Runtime and Os Approaches To Mitigating Contentionmentioning

confidence: 99%

“…This is related to the dissertation because our profiling technique relies on indicators to identify contentious code regions. Knauerhase et al [29] use last level cache miss rate as an indicator of contentiousness of an application. Zhuravlev et al [63] demonstrate that cache contention is not the dominant cause for performance degradation of co-running applications on CMPs;…”

Section: Software Runtime and Os Approaches To Mitigating Contentionmentioning

confidence: 99%

“…Previous work [63,29,40] show that last level cache miss rate can be used as an indicator to estimate the footprint size of an application and to predict the potential performance degradation an application may cause to its co-runners. Based on an application's characteristics in terms of their resource usage when running alone, we can predict a good thread-to-core mapping that takes advantage of the memory sharing topology when applications are co-located.…”

Section: A Heuristic Approach To Ttc Mappingmentioning

confidence: 99%

“…Xie et al [59] on the other hand, argue the existence of applications that are not contentious but sensitive. Meanwhile, other recent works [63,29] argue that a contentious application that has high cache misses is likely to be very sensitive as well.…”

Section: Contentiousness and Sensitivitymentioning

confidence: 99%

See 4 more Smart Citations

Mitigating Memory Resource Contention in Warehouse Scale Computers

Tang¹

View full text Add to dashboard Cite

The class of modern datacenters hosting large-scale Internet services such as web-search, mail, and social networking has gained significant momentum in today's computing environments. However, these datacenters, recently coined as warehouse scale computers (WSCs), are extremely expensive to construct and operate. Improving software performance and server utilization is key to improving the efficiency and reducing the enormous cost in WSCs.Modern WSCs are constructed using commodity multicore processors, on which part of the memory subsystem is shared. When multiple applications are co-located on a multicore machine, contention for the shared memory resources, such as caches and memory bandwidth, may occur. This contention can cause severe cross-core performance interference and significantly degrade application performance. Mitigating resource contention is critical for improving application performance. However, despite the wealth of research effort on contention management, little is known about how emerging large-scale web-service applications interact with the shared memory resources on commodity processors and how this contention can be mitigated to improve the performance of these applications.In addition to performance, mitigating contention is also critical for improving the server utilization in WSCs. As multicore processors with expanding core counts continue to dominate the server market, the overall utilization of WSCs depends heavily on the consolidation of workloads to take advantage of the total computing potential provided by modern processors. However, many of the applications running in WSCs are user-facing, latency-sensitive applications with quality of service (QoS) requirements. These QoS requirements can be violated by the performance interference that can occur when multiple applications are consolidated on a single machine. As a result, the current common practice in WSCs is iii to disallow the co-location of latency-sensitive applications with other applications. This approach is undesirable as it results in low machine utilization in WSCs and millions of dollars wasted.This dissertation presents novel compilation and runtime approaches to significantly mitigate contention and thus improve performance, QoS and machine utilization in datacenters. Specifically, the main contributions of this dissertation include: 1) comprehensive investigation and characterization of the impact of memory resource sharing on industrystrength large-scale datacenter workloads, which expose new characteristics and insights contrary to recent literature; 2) the design of a heuristic based system and a runtime system to intelligently map application threads to cores to promote positive resource sharing and mitigate resource contention to improve application performance; and 3) the design of novel compilation techniques and runtime systems that statically and dynamically manipulate applications' contentious nature to enable the co-location of applications with varying QoS requirements and as a result, gr...

show abstract

Section: Mitigating Contentionmentioning

confidence: 99%

Section: Software Runtime and Os Approaches To Mitigating Contentionmentioning

confidence: 99%

Section: Software Runtime and Os Approaches To Mitigating Contentionmentioning

confidence: 99%

Section: A Heuristic Approach To Ttc Mappingmentioning

confidence: 99%

Section: Contentiousness and Sensitivitymentioning

confidence: 99%

See 3 more Smart Citations

Mitigating Memory Resource Contention in Warehouse Scale Computers

Tang¹

View full text Add to dashboard Cite

show abstract

Energy and Thermal Aware Scheduling in Data Centers

Zomaya

Lee

2012

Energy‐Efficient Distributed Computing Systems

View full text Add to dashboard Cite

The case for colocation of high performance computing workloads

Breslow

Porter

Tiwari

et al. 2013

Concurrency and Computation

View full text Add to dashboard Cite

The current state of practice in supercomputer resource allocation places jobs from different users on disjoint nodes both in terms of time and space. While this approach largely guarantees that jobs from different users do not degrade one another's performance, it does so at high cost to system throughput and energy efficiency. This focused study presents job striping, a technique that significantly increases performance over the current allocation mechanism by colocating pairs of jobs from different users on a shared set of nodes. To evaluate the potential of job striping in large-scale environments, the experiments are run at the scale of 128 nodes on the state-of-the-art Gordon supercomputer. Across all pairings of 1024 process network-attached storage parallel benchmarks, job striping increases mean throughput by 26% and mean energy efficiency by 22%. On pairings of the real applications Gyrokinetic Toroidal Code (GTC), Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS), and MIMD Lattice Computation (MILC) at equal scale, job striping improves average throughput by 12% and mean energy efficiency by 11%. In addition, the study provides a simple set of heuristics for avoiding low performing application pairs. 239 Figure 3. Increase in system throughput (STP) over compact when applying job spreading and striping to the NAS parallel benchmarks and GTC, LAMMPS, and MILC. Figure 3 shows the performance results for the first set of experiments with the NPBs and the second set of experiments with GTC, LAMMPS, and MILC. For the NPBs, the mean performance increase from job spreading is 50%. If one examines striped coschedules of non-identical NPBs, the average performance increase is 26%. If one selects the best running mate other than embarrassingly parallel (EP) for each benchmark, then the average increase in performance is 36%. We choose to exclude EP because EP is minimally contentious. Each EP task's working set fits entirely in the private levels of cache, and EP spends very little time in active communication. Because of these traits, EP universally causes every application that it stripes with to achieve its best striped performance. Thus, for the sake of fairness and realism, we exclude these results from the 'Best' average. For the NPBs, random striping yields about 50% of the performance benefit of job spreading, and striping each job with its best running mate provides 70% of the performance benefit of spreading. This trend continues for real applications as well. For GTC, LAMMPS, and MILC, job spreading increases throughput by 23% and the mean heterogeneous striping and the mean best striping improve performance by 12% and 16%, respectively. PERFORMANCE RESULTS Compact versus spreading versus striping Network-attached storage parallel benchmarksIn this section, we examine the increase in collective throughput and energy efficiency for pairs of striped NPBs. The results are presented in Figure 4. For completeness, we run all pairwise combinations. This includes both heterogeneous pairings ...

show abstract

Using OS Observations to Improve Performance in Multicore Systems

Cited by 180 publications

References 4 publications

Mitigating Memory Resource Contention in Warehouse Scale Computers

Mitigating Memory Resource Contention in Warehouse Scale Computers

Energy and Thermal Aware Scheduling in Data Centers

The case for colocation of high performance computing workloads

Contact Info

Product

Resources

About