Recently, the company was developing a complex, multi-die system-in-package with novel technologies and architecture. It was challenging to predict the system’s performance and power under realistic machine-learning (ML) workloads. The team needed a modern solution that could help define the overall system architecture executing the ML workload. Ideally, the solution would also effectively model performance of their custom SoC consisting of digital, analog, and optical components.
Today’s advanced SoCs contain many interdependencies and interconnects, all of which must be validated holistically to ensure that the system will perform as intended. A traditional method to accomplish this is manual, involving static spreadsheet calculations. The alternative is Synopsys Platform Architect, a dynamic simulation solution that provides SoC architecture analysis and optimization for performance and power.