This post-doc position stands in the context of RAKES project. The overall objective of RAKES project is to study how a mix electrical/radiofrequency (RF) network-on-chip could improve performance (both in terms of execution time and execution time per watt) of future large scale coherent shared-memory many-cores. Keeping the coherence of data in such a shared-memory architecture requires a deluge of data movement and coherence traffic, sometimes targeting many (if not all) the cores. By nature, radio communications provide broadcast capabilities at negligible latency. Therefore, they have the potential to disseminate information very quickly at the scale of a circuit and thus to be an opening for solving these issues. In this project, we intend to study how wireless communications can solve the scalability of the above-mentioned problems, by using mixed wired/wireless Network on Chip.
The main goal of this mission consists in adapting and refining the NoC model and the memory mapping to take full benefit from cache and radio proposals.