Adaptive Power Profiling for Many-Core HPC Architectures

Kelley, Jaimie; Stewart, Christopher; Tiwari, Devesh; Gupta, Saurabh

doi:10.1109/icac.2016.45

Cited by 8 publications

(2 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Additionnaly, peak power consumption hampers the size of these infrastructures, challenging the electricity provisioning [2]. Power capping techniques have been developed to control this power consumption [11], [12]. While these techniques dynamically limit the power consumption of running nodes, we argue here that a better understanding of HPC applications behavior could be more beneficial in finding an adequate trade-off between performance, power and energy consumption.…”

Section: Introductionmentioning

confidence: 99%

Energy Analysis of a Solver Stack for Frequency-Domain Electromagnetics

Agullo

Giraud

Lanteri

et al. 2019

2019 27th Euromicro International Conference on Parallel, Distributed and Network-Based Processing (PDP)

View full text Add to dashboard Cite

High-performance computing (HPC) aims at developing models and simulations for applications in numerous scientific fields. Yet, the energy consumption of these HPC facilities currently limits their size and performance, and consequently the size of the tackled problems. The complexity of the HPC software stacks and their various optimizations makes it difficult to finely understand the energy consumption of scientific applications. To highlight this difficulty on a concrete use-case, we perform an energy and power analysis of a software stack for the simulation of frequency-domain electromagnetic wave propagation. This solver stack combines a high order finite element discretization framework of the system of three-dimensional frequency-domain Maxwell equations with an algebraic hybrid iterative-direct sparse linear solver. This analysis is conducted on the KNLbased PRACE-PCP system. Our results illustrate the difficulty in predicting how to trade energy and runtime.

show abstract

Section: Introductionmentioning

confidence: 99%

Energy Analysis of a Solver Stack for Frequency-Domain Electromagnetics

Agullo

Giraud

Lanteri

et al. 2019

2019 27th Euromicro International Conference on Parallel, Distributed and Network-Based Processing (PDP)

View full text Add to dashboard Cite

show abstract

“…Profilo measures the energy consumption of the CPU with a goal to quantify the differences between different speed scaling strategies and uses RAPL as an effective measurement tool in their automated simulation process. Kelley et al [22] use RAPL as a power tracing tool to profile peak power under core scaling which reduces the profiling time significantly. Their approach shows that RAPL measurements are reliable enough to be used as a power measurement tool as they use RAPL to collect the power trace for a sampling period and then this trace is used to predict peak power under different settings and across different architectures.…”

Section: Introductionmentioning

confidence: 99%

RAPL in Action

Khan

Hirki

Niemi

et al. 2018

ACM Trans. Model. Perform. Eval. Comput. Syst.

136

View full text Add to dashboard Cite

To improve energy efficiency and comply with the power budgets, it is important to be able to measure the power consumption of cloud computing servers. Intel's Running Average Power Limit (RAPL) interface is a powerful tool for this purpose. RAPL provides power limiting features and accurate energy readings for CPUs and DRAM which are easily accessible through different interfaces on large distributed computing systems. Since its introduction, RAPL has been used extensively in power measurement and modeling. However, the advantages and disadvantages of RAPL have not been well investigated yet. To fill this gap, we conduct a series of experiments to disclose the underlying strengths and weaknesses of the RAPL interface by using both customized microbenchmarks and three well-known application level benchmarks:Stream, Stress-ng and ParFullCMS. Moreover, to make the analysis as realistic as possible, we leverage two production-level power measurement datasets from the Taito, a supercomputing cluster of the Finnish Center of Scientific Computing (CSC) and also replicate our experiments on Amazon EC2. Our results illustrate different aspects of RAPL and document the findings through comprehensive analysis. Our observations reveal that RAPL readings are highly correlated with plug power, promisingly accurate enough and have negligible performance overhead. Experimental results suggest RAPL can be a very useful tool to measure and monitor the energy consumption of servers without deploying any complex power meters. We also show that there are still some open issues such as driver support, non-atomicity of register updates and unpredictable timings that might weaken the usability of RAPL in certain scenarios. For such scenarios, we pinpoint solutions and workarounds. CCS Concepts: • Hardware → Energy metering; Enterprise level and data centers power issues; Emerging architectures; • Computer systems organization → Cloud computing;

show abstract

The Paradigm of Power Bounded High-Performance Computing

Feng

Zou

et al. 2023

J. Comput. Sci. Technol.

View full text Add to dashboard Cite

Adaptive Power Profiling for Many-Core HPC Architectures

Cited by 8 publications

References 19 publications

Energy Analysis of a Solver Stack for Frequency-Domain Electromagnetics

Energy Analysis of a Solver Stack for Frequency-Domain Electromagnetics

RAPL in Action

The Paradigm of Power Bounded High-Performance Computing

Contact Info

Product

Resources

About