A power-measurement methodology for large-scale, high-performance computing

Scogland, Thomas R. W.; Steffen, Craig; Wilde, Torsten; Parent, Florent; Coghlan, Susan; Bates, Natalie; Feng, Wu-chun; Strohmaier, Erich

doi:10.1145/2568088.2576795

Cited by 31 publications

(18 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…PRO power meter, which reports instantaneous power consumption once per second. This sampling rate is typical in HPC environments [23], [37]. The accuracy of power measurement is ±1.5% plus 3 counts of the reported value.…”

Section: A Experimental Setupmentioning

confidence: 99%

“…Several studies have used this approach to identify patterns in applications' power consumption [15]- [17] or resource usage [18]- [21] or to detect anomalous behavior [22]. They build these signatures from traces of performance counter values rather than power consumption, and performance counters can be collected much more frequently than the standard 1 Hz for system-level power measurements in HPC [23]. However, to construct and validate the signatures, some of these studies use similar clustering and classification techniques to the ones we employ.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Power Signatures of High-Performance Computing Workloads

Combs

Nazor

Thysell

et al. 2014

2014 Energy Efficient Supercomputing Workshop

View full text Add to dashboard Cite

Workload-aware power management and scheduling techniques have the potential to save energy while minimizing negative impact on performance. The effectiveness of these techniques depends on the stability of a workload's power consumption pattern across different input data, resource allocations (e.g. number of cores), and hardware platforms.In this paper, we show that the power consumption behavior of HPC workloads can be accurately captured by concise signatures built from their power traces. We validate this approach using 255 traces collected from 13 high-performance computing workloads on 4 different hardware platforms. First, we use both feature-based and time-series-based distance metrics to cluster our traces, and we quantitatively show that feature-based clusterings segregate traces by workload just as effectively as the more compute-and space-intensive time-series-based clusterings. Second, we demonstrate that unlabeled traces can be classified by workload with over 85% accuracy, based only on these concise statistical signatures.

show abstract

Section: A Experimental Setupmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Power Signatures of High-Performance Computing Workloads

Combs

Nazor

Thysell

et al. 2014

2014 Energy Efficient Supercomputing Workshop

View full text Add to dashboard Cite

show abstract

“…Unlike traditional cloud infrastructure built on an identical processor architecture, heterogeneity assumes a cloud that makes use of different specialist processors that can accelerate the completion of specific tasks or can be turned off when not required, thus maximizing both performance and energy efficiency [14]. Another previous study [15] proposes a resource allocation strategy in a heterogeneous cluster (integration of core nodes and accelerator nodes) to realize a scheduling scheme that achieves high performance and fairness.…”

Section: B Heterogeneity In Cloudmentioning

confidence: 99%

CloudLightning: a Self-Organized Self-Managed Heterogeneous Cloud

Xiong

Dong

Filelis‐Papadopoulos

et al. 2017

Annals of Computer Science and Information Systems

View full text Add to dashboard Cite

Abstract-The increasing heterogeneity of cloud resources, and the increasing diversity of services being deployed in cloud environments are leading to significant increases in the complexities of cloud resource management. This paper presents an architecture to manage heterogeneous resources and to improve service delivery in cloud environments. A loosely-coupled, hierarchical, selfadapting management model, deployed across multiple layers, is used for heterogeneous resource management. Moreover, a service-specific coalition formation mechanism is employed to identify appropriate resources to support the process parallelism associated with high performance services. Finally, a proofof-concept of the proposed hierarchical cloud architecture, as realized in CloudLightning project, is presented.

show abstract

“…The Green500 [4] has more data, but its workload does not have any sense of graduated workload, making it impossible to measure energy proportionality. In addition, there are serious concerns about its measurement bias [8]. In contrast, the SPECpublished results may have less measurement bias.…”

Section: The Specpower_ssj2008 Benchmarkmentioning

confidence: 99%

Measuring Server Energy Proportionality

Hsu

Poole

2015

Proceedings of the 6th ACM/SPEC International Conference on Performance Engineering

View full text Add to dashboard Cite

In performance engineering, metrics are often used to track the progress over time. Concerning the potential bias of using a single metric, performance engineers tend to use multiple metrics for reasoning. However, this approach has its own challenges. In this work we study one of the challenges in the context of analyzing trends in server energy proportionality. We examine a wide range of metrics for measuring energy proportionality, trying to determine which metrics are essential and which are redundant. We do this by comparing the trend curves of the metrics for the published results of the SPECpower ssj2008 benchmark. While the context is specific, the proposed analysis method is quite general. We hope that this method would help us do performance engineering more effectively.

show abstract

A power-measurement methodology for large-scale, high-performance computing

Abstract: Improvement in the energy e ciency of supercomputers can be accelerated by improving the quality and comparability of e ciency measurements. The ability to generate accurate measurements at extreme scale are just now emerging.

Cited by 31 publications

References 9 publications

Power Signatures of High-Performance Computing Workloads

Power Signatures of High-Performance Computing Workloads

CloudLightning: a Self-Organized Self-Managed Heterogeneous Cloud

Measuring Server Energy Proportionality

Contact Info

Product

Resources

About