Generative AI technology identifies and address EM-IR violations
Generative AI technology that automatically identifies the root cause of EM-IR drop violations early in the design process has been introduced by Cadence. Voltus InsightAI selects and implements the most efficient fixes to improve power, performance and area (PPA). Using it, designers can fix up to 95 per cent of violations prior to signoff, leading to doubling productivity in EM-IR closure, said Cadence.
EM-IR violations need to be addressed early in the design phase but a major bottleneck for analysis is that it is computationally very expensive due to the size and coupled nature of the power integrity network. The AI-driven Voltus InsightAI uses machine learning methods for very fast incremental IR analysis. Customers can use the in-design analysis to enhance on-chip and chiplet power integrity to uncover issues early.
Productivity enhancing features include the fast IR inferencing engine which uses proprietary neural networks to build models of the power grid. It can also perform extremely fast incremental IR analysis to provide instant feedback on the impact of design changes.
Voltus InsightAI uses deep learning to discover the root cause of IR drop problems, identify aggressors, victims and resistance bottlenecks. The diagnosis function uses electrical, spatial and timing factors to predict IR drop issues during design.
There is also multi-method fixing using decision-tree methods to perform timing and design rule check (DRC)-aware fixes of IR drop. It uses multiple methods such as placement, grid reinforcement, routing and engineering change orders (ECOs). Voltus InsightAI selects precise fixing methods based on the root cause of the problem to improve utilisation and PPA.
Voltus InsightAI is fully integrated with the Cadence Innovus implementation system, the Cadence Tempus timing solution, the Cadence Voltus IC power integrity solution and the Cadence Pegasus verification system for complete IR design closure from implementation to sign off that is timing- and DRC-aware.
Ben Gu, corporate vice president of R&D, Multiphysics Systems Analysis business unit, Cadence, commented: “With this capability, designers don’t need to over-design the power grid, thereby enabling far better PPA”.
