|
The goal of DFM can be seen as improving the design’s resilience to
systematic and random yield limiters by leveraging process knowledge beyond DRC in a stable and well characterized process. In contrast, the goal of design technology co-optimization (DTCO) is to optimize key elements of both the process and the design definition early in the technology node to maximize the design value of new technology features (e.g., local interconnect), minimize unnecessary design diversity to allow comprehensive assessment of layout-dependent process effects without handicapping designers, drive circuit-relevant yield learning by focusing silicon experiments on features that matter to the designers, and minimize the risk of major changes late in the technology node by facilitating continuous exchange of information during the process and design development.
DTCO is not a solution or a tool in itself; it is a process to help manage the complexity of ramping advanced-technology nodes by identifying scaling bottlenecks early and exploring means to overcome them without overburdening either the design or the process side of the technology. The biggest challenge for DTCO is that the most fundamental decisions are made with the least amount of data at the beginning of the technology development cycle. Mitigating the risk of making wrong decisions by maintaining multiple options and “downselecting” as data becomes available is cost and time prohibitive.
|
