Supercomputing Centers and Electricity Service Providers: A Geographically Distributed Perspective on Demand Management in Europe and the United States

Patki, Tapasya; Bates, Natalie; Ghatikar, Girish; Clausen, Anders; Klingert, Sonja; Abdulla, Ghaleb; Sheikhalishahi, Mehdi

doi:10.1007/978-3-319-41321-1_13

Cited by 11 publications

(3 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hence, this work could be extended to become more dynamic: If the energy budget followed electricity price, we could control and thus reduce a significant part of the cluster's costs. This would provide an improved solution to the remaining use cases described in [3] whom electricity cost vary since it may partially depend on renewable sources.…”

Section: Discussionmentioning

confidence: 99%

“…Nevertheless, they lack in adaptability, given that power is controlled 1 https://www.top500.org/system/178764 independently of the instant load. A recent study [3], implicating a larger group of supercomputers and electricity providers in both the US and Europe, showed that while the upper power-bound is an important parameter, power variations do not affect the final energy cost in most use cases. In this paper, we show that adopting flexible power adaptive scheduling techniques, by setting restriction on energy consumption instead of power, can optimize system utilization, slowdown and even energy efficiency when compared to rigid powercapping strategies.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Towards Energy Budget Control in HPC

Dutot

Georgiou²,

Glesser

et al. 2017

2017 17th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGRID)

View full text Add to dashboard Cite

Energy consumption has become one of the most critical issues in the evolution of High Performance Computing systems (HPC). Controlling the energy consumption of HPC platforms is not only a way to control the cost but also a step forward on the road towards exaflops. Powercapping is a widely studied technique that guarantees that the platform will not exceed a certain power threshold instantaneously but it gives no flexibility to adapt job scheduling to a longer term energy budget control. We propose a job scheduling mechanism that extends the backfilling algorithm to become energy-aware. Simultaneously, we adapt resource management with a node shutdown technique to minimize energy consumption whenever needed. This combination enables an efficient energy consumption budget control on a cluster during a period of time. The technique is experimented, validated and compared with various alternatives through extensive simulations. Experimentation results show high system utilization and limited bounded slowdown along with interesting outcomes in energy efficiency while respecting an energy budget during a particular time period.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Towards Energy Budget Control in HPC

Dutot

Georgiou²,

Glesser

et al. 2017

2017 17th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGRID)

View full text Add to dashboard Cite

show abstract

“…However, the high power consumption of GPUs significantly impacts their reliability, economic viability and operational cost. This is a particular issue for GPU clouds and high-performance computing (HPC) systems, where the GPU power and cooling infrastructures contribute to a large part of the operational cost [4]. As a Yiming Wang, Meng Hao, Hui He and Weizhe Zhang are with the School of Cyberspace Science, Harbin Institute of Technology, Harbin 150001, China.…”

Section: Introductionmentioning

confidence: 99%

DRLCAP: Runtime GPU Frequency Capping With Deep Reinforcement Learning

Wang,

Hao,

et al. 2024

IEEE Trans. Sustain. Comput.

View full text Add to dashboard Cite

Power and energy consumption is the limiting factor of modern computing systems. As the GPU becomes a mainstream computing device, power management for GPUs becomes increasingly important. Current works focus on GPU kernel-level power management, with challenges in portability due to architecture-specific considerations. We present DRLCAP, a general runtime power management framework intended to support power management across various GPU architectures. It periodically monitors system-level information to dynamically detect program phase changes and model the workload and GPU system behavior. This elimination from kernel-specific constraints enhances adaptability and responsiveness. The framework leverages dynamic GPU frequency capping, which is the most widely used power knob, to control the power consumption. DRLCAP employs deep reinforcement learning (DRL) to adapt to the changing of program phases by automatically adjusting its power policy through online learning, aiming to reduce the GPU power consumption without significantly compromising the application performance. We evaluate DRLCAP on three NVIDIA and one AMD GPU architectures. Experimental results show that DRLCAP improves prior GPU power optimization strategies by a large margin. On average, it reduces the GPU energy consumption by 22% with less than 3% performance slowdown on NVIDIA GPUs. This translates to a 20% improvement in the energy efficiency measured by the energy-delay product (EDP) over the NVIDIA default GPU power management strategy. For the AMD GPU architecture, DRLCAP saves energy consumption by 10%, on average, with a 4% percentage loss, and improves energy efficiency by 8%.

show abstract