Impact of AI in Architecture Exploration & System Simulation

The term “AI” has long carried diverse connotations in the public consciousness, preceding its practical applications. Now, as AI begins to demonstrate its value, it is pertinent to explore its implications in electronics, networking and semiconductors, particularly within the realm of simulation. 

In the context of simulation, AI predominantly involves the utilization of machine learning and neural networks. Below, we examine where AI can be harnessed for simulation through the lens of discrete-event simulation and VisualSim.

System level modeling primarily applies to architecture trade-offs using rapid simulations. AI enhances this approach through deep reinforcement learning, termed “system level learning.” By interacting with the discrete-event simulation model (environment), AI-driven systems can execute hundreds of thousands or even millions of quick timed simulation sequences, culminating in the development of an optimal control mechanism. Widely utilized in electronics, applications of system level modeling span diverse domains including bandwidth allocation, system sizing, hardware-software partitioning and system validation. The application spreads across SoC, automotive, aerospace and defense sectors. Furthermore, its minimal compute footprint renders it suitable for execution on laptops, enabling real-time or near real-time responses.