DPS: Adaptive Power Management for Overprovisioned Systems
Open Access
- 11 November 2023
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 1-14
Abstract
Maximizing performance under a power budget is essential for HPC systems and has inspired the development of many power management frameworks. These can be broadly characterized into two groups: model-based and stateless. Model-based frameworks use machine learning to achieve good performance under a power budget but are highly dependent on the quality of the learned model and the data used to train it. Stateless frameworks are more robust and require no training, but are generally lower performance. In this paper, we propose a new framework that does not require a model, but does track a small amount of state in the form of recent power dynamics. We implement this idea and test it on a public cloud running both Spark and HPC jobs. We find when total power demand is low, our framework achieves equivalent performance to prior work, but when power demand is high it achieves mean 8% performance improvement (with no reliance on a learned model).Keywords
Funding Information
- Army Research Office (W911NF22C0082)
- National Science Foundation CCF (CCF-1822949)
- National Science Foundation PPoSS (CCF-2119184)
- National Science Foundation CNS (CNS-1764039)
This publication has 41 references indexed in Scilit:
- Dynamic power sharing for higher job throughput2015
- JouleGuard2015
- POW2015
- A Probabilistic Graphical Model-based Approach for Minimizing Energy Under Performance ConstraintsACM SIGPLAN Notices, 2015
- Maximizing Throughput of Overprovisioned HPC Data Centers Under a Strict Power Budget2014
- Optimizing power allocation to CPU and memory subsystems in overprovisioned HPC systems2013
- Power containers2013
- Koala2009
- GRACE-2: integrating fine-grained application adaptation with global adaptation for saving energyInternational Journal of Embedded Systems, 2009
- The Case for Energy-Proportional ComputingComputer, 2007