Improving energy consumption of NoC based architectures through approximate communication

This paper presents an approximate communication technique aimed at improving the energy efficiency of Network-on-Chip (NoC) based architectures. The basic idea is a selective adjustment, at run-time, of the voltage swing of the links of the NoC to obtain a trade-off between the communication energy consumption and the accuracy of the running application. The proposed technique is applied to a case study, namely, a JPEG encoder resulting in an energy saving up to 70% with a negligible impact on the quality of the compressed image.

[1]  Salvatore Monteleone,et al.  Distributed topology discovery in self-assembled nano network-on-chip , 2014, Comput. Electr. Eng..

[2]  André Rossi,et al.  Crosstalk-aware link power model for Networks-on-Chip , 2016, 2016 Conference on Design and Architectures for Signal and Image Processing (DASIP).

[3]  裕幸 飯田,et al.  International Technology Roadmap for Semiconductors 2003の要求清浄度について - シリコンウエハ表面と雰囲気環境に要求される清浄度, 分析方法の現状について - , 2004 .

[4]  Hadi Esmaeilzadeh,et al.  AxBench: A Multiplatform Benchmark Suite for Approximate Computing , 2017, IEEE Design & Test.

[5]  Gerald E. Sobelman,et al.  Network-on-chip link analysis under power and performance constraints , 2006, 2006 IEEE International Symposium on Circuits and Systems.

[6]  Jie Han,et al.  Approximate computing: An emerging paradigm for energy-efficient design , 2013, 2013 18th IEEE European Test Symposium (ETS).

[7]  Salvatore Monteleone,et al.  Energy efficient transceiver in wireless Network on Chip architectures , 2016, 2016 Design, Automation & Test in Europe Conference & Exhibition (DATE).

[8]  Ahmad Khademzadeh,et al.  Data Encoding Techniques for Reducing Energy Consumption in Network-on-Chip , 2014, IEEE Transactions on Very Large Scale Integration (VLSI) Systems.

[9]  Sparsh Mittal,et al.  A Survey of Techniques for Approximate Computing , 2016, ACM Comput. Surv..