Machine Learning Is Revolutionizing IBIS-AMI Optimization in High-Speed Design

The complexity of IBIS-AMI models used in simulating serial links has increased to keep pace with serial link speeds. In the past, an exhaustive search method was used to find the best set of parameters for a given channel. As the number of model parameters and ranges has increased, this approach has quickly become computationally expensive, even with parallel execution. To check as many channels as possible, a better method has been developed that more efficiently provides an optimal set of parameter values.

The DesignCon 2024 proceedings included a paper that presented the use of a machine learning optimization algorithm to find the best set of parameters in a set of IBIS-AMI models and compared them to the exhaustive traditional search method. The paper, titled "The Optimization of IBIS-AMI Model Parameters with Machine Learning Algorithms” and authored by Jared James and Ambrish Varma of Cadence, introduces an approach to optimizing input/output buffer information specification - algorithmic modeling interface (IBIS-AMI) model parameters using machine learning techniques and proposes an AI/ML focused solution to tackle inefficiencies in the simulation processes.

For engineers and designers working with high-speed serial links, this paper represents more than just an academic advancement—it’s a direct response to challenges they face daily. By combining simulation methods with machine learning, James and Varma are paving the way for a more efficient design methodology. To view their highly regarded paper, click here.

The Cadence Sigrity X Platform provides a full array of signal Integrity and IBIS modeling solutions.