The research introduces a novel approach in circuit design by using artificial intelligence technologies. The most time-consuming parts of the physical design process are placement and routing. This research proposes a new perspective by treating circuit components as point clouds and using Transformer-based point cloud perception methods to extract features from the circuit. This approach allows for direct feature extraction from raw data, end-to-end training, and results in high performance. The method shows state-of-the-art performance in congestion prediction tasks on both the CircuitNet and ISPD2015 datasets, as well as in design rule check (DRC) violation prediction tasks on the CircuitNet dataset.
Publication date: 26 Oct 2023
Project Page: https://github.com/hustvl/circuitformer
Paper: https://arxiv.org/pdf/2310.17418
