This paper proposes a statistical link design methodology for variability-tolerant current-mode interconnect applied forNetworks-on-Chip links. The model takes into considerations the systematic and random effects of process variability. The model calculates the resistive, capacitive and device variations then uses it to calculate current variations of each NoC link in a floor-plan. Statistical link design proposes a current safe guard to keep signal integrity versus existing process variability sources. The proposed technique is tested using test cases of 4x4 meshes at 65 nm, 45 nm, 32 nm, 22nm, and 16 nm technologies. Results show that the received current variations at 16nm approach 30% of the total current at the link receiver. The current variations are increased by 100% as NoC mesh size scales from 4x4 to 16x16 at 45 nm. Comparing our statistical design to worst-case at 65 nm, we save up to 33 % of the total power cost compared to worstcase. The link failure probability is modeled to calculate the average NoC link failure rate. NoC with links designed to have statistical guard achieves low failure rate that is up to 3.7 % for 4x4 mesh.
KeywordsFloor-plan; Networks-on-Chip; Current-mode interconnect; process variability.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.