BagNet: Berkeley Analog Generator with Layout Optimizer Boosted with Deep Neural Networks
This work addresses the problem of time-consuming post-layout simulations for analog circuit designers, offering a method to integrate layout effects more efficiently, though it appears incremental as it builds on existing evolutionary optimizers with a DNN discriminator.
The paper tackles the widening discrepancy between post-layout and schematic simulations in analog design by introducing a learning framework that uses a deep neural network to reduce simulation counts in evolutionary-based optimizers, achieving at least two orders of magnitude improvement in sample efficiency for large circuits like an optical link receiver layout.
The discrepancy between post-layout and schematic simulation results continues to widen in analog design due in part to the domination of layout parasitics. This paradigm shift is forcing designers to adopt design methodologies that seamlessly integrate layout effects into the standard design flow. Hence, any simulation-based optimization framework should take into account time-consuming post-layout simulation results. This work presents a learning framework that learns to reduce the number of simulations of evolutionary-based combinatorial optimizers, using a DNN that discriminates against generated samples, before running simulations. Using this approach, the discriminator achieves at least two orders of magnitude improvement on sample efficiency for several large circuit examples including an optical link receiver layout.