A Learning-Based Approach to the Automated Design of MPSoC Networks
Many embedded computing platforms are highly specialised towards a given task and, consequently, sacrifice generality for high performance and energy efficiency at low cost. It is commonly accepted that integrating multiple processor cores on the same chip is the most promising way of delivering a high level of processing power under tight energy and cost constraints. Whereas the customisation of individual processing elements to particular tasks such as DSP or multimedia functions is a well-studied problem, the specialisation of application-specific on-chip and off-chip interconnects between processing elements has been largely neglected.
In this paper we explore the design space of a tree-based network on chip of a synthesisable application-specific MPSoC. We empirically deduce the optimal network configurations, in terms of runtime and energy consumption, for a range of benchmark workloads. We present a machine learning approach that is able to predict optimal, or near-optimal, network-on-chip configurations for a new and as-yet-unseen workload. This new approach to automated NoC design yields designs that are, on average, within 9% of optimal design for the given workload. Moreover, the model predicts network configurations based on sample data from a single profiling run of the new application on a reference platform, providing the answer up to 280 times faster than an exhaustive search.
KeywordsDesign Space Dynamic Energy Machine Learning Model Core Clock Slave Device
Unable to display preview. Download preview PDF.
- 1.Angiolini, F., Meloni, P., Carta, S., Benini, L., Raffo, L.: Contrasting a NoC and a traditional interconnect fabric with layout awareness. In: Proceedings of Design, Automation and Test in Europe Conference and Exhibition 2004, pp. 124–129 (2006)Google Scholar
- 2.ARM: AMBA AXI Protocol Specification (March 2004)Google Scholar
- 3.Chen, X., Peh, L.S.: Leakage power modeling and optimization in interconnection networks. In: ISLPED 2003: Proceedings of the 2003 International Symposium on Low Power Electronics and Design, pp. 90–95. ACM, New York (2003)Google Scholar
- 6.EEMBC: Coremark (May 2010), http://www.coremark.org
- 7.Gilabert, F., Ludovici, D., Medardoni, S., Bertozzi, D., Benini, L., Gaydadjiev, G.: Designing regular network-on-chip topologies under technology, architecture and software constraints. In: International Conference on Complex, Intelligent and Software Intensive Systems, CISIS 2009, 16-19, pp. 681–687 (2009)Google Scholar
- 8.Ho, R., Mai, K., Horowitz, M.: Managing wire scaling: a circuit perspective, pp. 177–179 (June 2003)Google Scholar
- 10.Kim, J., Park, D., Nicopoulos, C., Vijaykrishnan, N., Das, C.R.: Design and analysis of an noc architecture from performance, reliability and energy perspective. In: ANCS 2005: Proceedings of the 2005 ACM Symposium on Architecture for Networking and Communications Systems, pp. 173–182. ACM, New York (2005)CrossRefGoogle Scholar
- 14.Ludovici, D., Gilabert, F., Medardoni, S., Gómez, C., Gómez, M., López, P., Gaydadjiev, G., Bertozzi, D.: Assessing fat-tree topologies for regular network-on-chip design under nanoscale technology constraints. In: Proceedings of Design, Automation and Test in Europe Conference and Exhibition 2009, pp. 562–565 (2009)Google Scholar
- 15.van Meeuwen, T., Vandecappelle, A., van Zelst, A., Catthoor, F., Verkest, D.: System-level interconnect architecture exploration for custom memory organizations. In: Proceedings of the 14th International Symposium on System Synthesis, pp. 13–18 (2001)Google Scholar
- 16.Micheli, G.D., Benini, L.: Networks on Chips: Technology and Tools (Systems on Silicon). Morgan Kaufmann Publishers Inc., San Francisco (2006)Google Scholar
- 17.Rodriguez, S., Jacob, B.: Energy/power breakdown of pipelined nanometer caches (90nm/65nm/45nm/32nm). In: ISLPED 2006: Proceedings of the 2006 International Symposium on Low Power Electronics and Design, pp. 25–30. ACM, New York (2006)Google Scholar
- 18.Scott, S., Abts, D., Kim, J., Dally, W.: The blackwidow high-radix clos network. In: 33rd International Symposium on Computer Architecture, ISCA 2006, pp. 16–28 (2006)Google Scholar
- 19.Towles, B., Dally, W.J.: Route packets, not wires: On-chip inteconnectoin networks. In: Design Automation Conference, pp. 684–689 (2001)Google Scholar