What is the problem?
While feature sizes are continuously scaled towards atomic dimensions, industry is increasingly confronted with unexpected physical artefacts to be considered at each new technology node. Among these, process variation and parameter degradation lead to reliability concerns impacting integrated circuit design at all abstraction levels. As variation and degradation may become a limiting factor for future scaled technologies, there has been a tremendous research effort in understanding these artefacts. Versatile tools, allowing consideration of these artefacts and their combined impact during the design of ICs are still in their infancy.
What are we doing about it?
Rather than developing yet another design support methodology, we aim to combine and refine existing reliability and variability prediction methodologies at the abstraction layers with highest industrial importance: Register transfer (RT) level – usual design entry, gate level – where most design for reliability (DfR) techniques are applied, and transistor level – where final sign-off is made.
MoRV will cover the strong relationship between variability and ageing, which are usually treated separately, fostering the idea of treating ageing as a form of time-dependent variability. Combined models from transistor, over gate, to RT level will be characterized directly from silicon measurement and all models will be able to interpret the same characterization data base from the silicon measurement.
The results will be introduced into a reference design flow combined with a multi-level multi-physics engine. Final goal of MoRV is to enable automated synthesis from specification to circuit. Each model layer will offer reliability and variation prediction for typical and worst case scenarios in order to assess the effectiveness of available design techniques.
This project has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstration under grant agreement no. 619234.