Nvidia is building an AI system called AutoDPM that helps engineers determine the layout of microchips. This is intended to speed up the development process and lead to better optimized chips.

Nvidia shows AutoDPM. This stands for Automated DreamPlace-based Macro Placement. It is an Electronic Design Automation (EDA) system based on machine learning and AI. AutoDPM is designed to help chip designers determine the optimal layout of a microchip: the so-called floor plan. This is a complex and time-consuming problem with a major impact on the final performance of a new chip.

optimization problem

Chip designers develop new circuits and components, but these have to be connected to each other on the finished microchip via interconnects. How this happens determines how powerful and efficient a chip is. For humans, this complex optimization problem is difficult to solve. By using AI, Nvidia hopes for a system with AutoDPM that can deliver the ideal solution in a few hours.

Nvidia isn’t the first to see the potential of AI in chip development. Cadence AI already proposed a solution in the same design with Cerebrus in the middle of last year. Google has already examined the potential of artificial intelligence for floor plan creation, and Nvidia is doing its part with AutoDPM.

As an AI specialist and chip designer, Nvidia is deeply committed to improving microchip development processes. The company previously rolled out a solution to quickly turn designs into photomasks for use in production.