Affiliation:
1. Duke University, Durham, NC, USA
Abstract
Computational sprinting is a class of mechanisms that boost performance but dissipate additional power. We describe a sprinting architecture in which many, independent chip multiprocessors share a power supply and sprints are constrained by the chips' thermal limits and the rack's power limits. Moreover, we present the computational sprinting game, a multi-agent perspective on managing sprints. Strategic agents decide whether to sprint based on application phases and system conditions. The game produces an equilibrium that improves task throughput for data analytics workloads by 4-6× over prior greedy heuristics and performs within 90% of an upper bound on throughput from a globally optimized policy.
Funder
National Science Foundation
Publisher
Association for Computing Machinery (ACM)
Reference49 articles.
1. US census data (1990) data set. https://archive.ics.uci.edu/ml/datasets/US+Census+Data (1990). US census data (1990) data set. https://archive.ics.uci.edu/ml/datasets/US+Census+Data (1990).
2. Movielens dataset. http://grouplens.org/datasets/movielens/. Movielens dataset. http://grouplens.org/datasets/movielens/.
3. Web data commons: Hyperlink graphs. http://webdatacommons.org/hyperlinkgraph/index.html. Web data commons: Hyperlink graphs. http://webdatacommons.org/hyperlinkgraph/index.html.
4. Control-theoretic techniques and thermal-RC modeling for accurate and localized dynamic thermal management
Cited by
2 articles.
订阅此论文施引文献
订阅此论文施引文献，注册后可以免费订阅5篇论文的施引文献，订阅后可以查看论文全部施引文献