A Deep Learning Toolbox for Analog Integrated Circuit Placement

Inhalt:
This paper presents a deep learning toolbox, DEEPPLACER, to assist designers during the layout design of analog integrated circuits. DEEPPLACER relies on a simple pair-wise device interaction circuit description, i.e., the circuits’ topological constraints, to propose valid floorplan solutions for block-level structures, including topologies and deep technology nodes not used for its training, at push-button speed. Despite its automatic functionalities, the toolbox is focused on explainable artificial intelligence, involving the designer in the synthesis flow via filtering and editing options over the candidate floorplan solutions. This constant state of human-machine feedback environment turns the designer aware of the impact of each device’s position change and inherent tradeoffs while suggesting subsequent moves, ultimately increasing the designers’ productivity in this time-consuming and iterative task. Finally, DEEPPLACER is shown to instantly generate a floorplan with 61% better constraint fulfilment than a human designed solution.