Please use this identifier to cite or link to this item:
|Title:||A hybrid local-global approach for multi-core thermal management|
Dynamic thermal management (DTM)
|Citation:||Jayaseelan, R.,Mitra, T. (2009). A hybrid local-global approach for multi-core thermal management. IEEE/ACM International Conference on Computer-Aided Design, Digest of Technical Papers, ICCAD : 314-320. ScholarBank@NUS Repository.|
|Abstract:||Multi-core processors have become an integral part of mainstream high performance computer systems. In parallel, exponentially increasing power density and packaging costs have necessitated system level thermal management solutions for multi-core systems. Dynamic thermal management (DTM) techniques monitor on-chip temperature continuously and typically employs dynamic voltage and frequency scaling (DVFS) to lower the temperature when it exceeds a pre-defined threshold. State-of-the-art DTM solutions for multi-core systems include distributed DVFS (where each core can scale the voltage/ frequency individually) and global DVFS (where all cores scale voltage/frequency simultaneously). Distributed DVFS generally offers higher performance than global DVFS, but it is hard to implement and has major scalability issues. We propose a hybrid local-global thermal management approach for multi-core systems that offers better performance than distributed DVFS, while maintaining the simplicity of global DVFS. We employ global DVFS across all the cores but locally tune the performance of each core individually through architectural adaptations. We exploit easily reconfigurable micro-architecture parameters such as instruction window size, issue width, and fetch throttling in per-core thermal management. Our hybrid solution is easy to implement and highly effective towards temperature management. The key challenge is appropriate choice of configurations at runtime to provide optimal performance under thermal constraints. We formulate it as a configuration search problem and design an ef-ficient software-based solution that selects the appropriate configuration. Our hybrid method, though simpler to implement, achieves 5% better throughput compared to distributed DVFS. Copyright 2009 ACM.|
|Source Title:||IEEE/ACM International Conference on Computer-Aided Design, Digest of Technical Papers, ICCAD|
|Appears in Collections:||Staff Publications|
Show full item record
Files in This Item:
There are no files associated with this item.
checked on Jan 13, 2019
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.