Abstract:Gate sizing plays an important role in timing optimization after physical design. Existing machine learning-based gate sizing works cannot optimize timing on multiple timing paths simultaneously and neglect the physical constraint on layouts. They cause sub-optimal sizing solutions and low-efficiency issues when compared with commercial gate sizing tools. In this work, we propose a learning-driven physically-aware gate sizing framework to optimize timing performance on large-scale circuits efficiently. In our gradient descent optimization-based work, for obtaining accurate gradients, a multi-modal gate sizing-aware timing model is achieved via learning timing information on multiple timing paths and physical information on multiple-scaled layouts jointly. Then, gradient generation based on the sizing-oriented estimator and adaptive back-propagation are developed to update gate sizes. Our results demonstrate that our work achieves higher timing performance improvements in a faster way compared with the commercial gate sizing tool.

What problem does this paper attempt to address?

This paper focuses on the gate-sizing optimization problem in integrated circuit design, which is a critical step in timing optimization. Existing machine learning methods have limitations in handling multiple timing paths and considering physical layout constraints simultaneously, leading to suboptimal solutions and low efficiency compared to commercial gate-sizing tools. The paper proposes a learning-based, physics-aware gate-sizing optimization framework aimed at efficiently improving the timing performance of large-scale circuits. The framework is implemented through gradient descent optimization. First, a multi-modal gate-sizing aware timing model is established, which accurately captures the timing and physical information through joint learning of multiple timing paths and multiple ratio layouts. Then, they develop a size-aware estimator and an adaptive backpropagation method to update gate sizes based on gradient generation. The research results show that this method outperforms commercial gate-sizing tools in terms of improving timing performance and speed. The challenges faced by the paper include constructing a timing model that considers the influence of gate sizes and efficiently generating and backpropagating gradients for discrete gate sizes in large-scale circuits. To address these issues, they propose multi-modal learning and a direct estimation algorithm to handle discreteness and avoid inconsistencies during the optimization process. In conclusion, this paper attempts to address how to effectively optimize gate sizes of large-scale circuits using machine learning and physics-aware techniques to improve timing performance, reduce iterations, and ultimately enhance design efficiency.

Learning-driven Physically-aware Large-scale Circuit Gate Sizing

Variation-aware gate sizing and clustering for post-silicon optimized circuits.

Design Time Body Bias Selection for Parametric Yield Improvement.

Optimization Algorithm of Evolutionary Design of Circuits Based on Genetic Algorithm

Accelerating Chip Design with Machine Learning: from Pre-Silicon to Post-Silicon

Improving the Process-Variation Tolerance of Digital Circuits Using Gate Sizing and Statistical Techniques

An Efficient Analog Circuit Sizing Method Based on Machine Learning Assisted Global Optimization

Machine Learning Driven Global Optimisation Framework for Analog Circuit Design

Fast Surrogate-Assisted Constrained Multiobjective Optimization for Analog Circuit Sizing Via Self-Adaptive Incremental Learning

Post-Layout Simulation Driven Analog Circuit Sizing

Circuit Optimization over Multiple Process Corners for Analog Electronic Design Automation.

CircuitVAE: Efficient and Scalable Latent Circuit Optimization

Timing-Driven Technology Mapping Approximation Based on Reinforcement Learning

Procedural- and Reinforcement-Learning-Based Automation Methods for Analog Integrated Circuit Sizing in the Electrical Design Space

GCN-RL Circuit Designer: Transferable Transistor Sizing with Graph Neural Networks and Reinforcement Learning

RLCkt: an Analog Circuit Automatic Sizing Sage Based on Reinforcement Learning

An efficient leakage power optimization framework based on reinforcement learning with graph neural network

Global approach for CMOS circuit optimization by transistor resizing

Learning to Design Circuits

Imbalanced Large Graph Learning Framework for FPGA Logic Elements Packing Prediction

ML-based AIG Timing Prediction to Enhance Logic Optimization