Abstract:Placement and routing are two indispensable and challenging (NP-hard) tasks in modern chip design flows. Compared with traditional solvers using heuristics or expert-well-designed algorithms, machine learning has shown promising prospects by its data-driven nature, which can be of less reliance on knowledge and priors, and potentially more scalable by its advanced computational paradigms (e.g. deep networks with GPU acceleration). This survey starts with the introduction of basics of placement and routing, with a brief description on classic learning-free solvers. Then we present detailed review on recent advance in machine learning for placement and routing. Finally we discuss the challenges and opportunities for future research.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the challenges faced by the two crucial and time - consuming tasks, Placement and Routing, in the modern chip design flow. Specifically: 1. **Placement problem**: - The placement task involves allocating logical components (such as logic gates and functional blocks) to discrete positions in the physical layout of the chip. - The goal is to optimize power consumption, performance, and area (PPA) metrics while satisfying constraint conditions (such as placement density and routing congestion). - The placement problem can be regarded as a complex two - dimensional bin - packing problem with geometric constraints and is known to be NP - hard. - Modern placement needs to reduce routing congestion and improve routing capabilities in the early stage, because the placement quality cannot be accurately evaluated before the routing is completed, resulting in a long feedback loop in the design flow. 2. **Routing problem**: - The routing task is responsible for completing the physical connections between logical components. - The routing problem is also NP - hard, even in the simplest two - pin network case. - Given a placement solution, routing needs to use physical wires to connect the pins of millions of networks under limited routing resources and complex geometric design rules. - Routing is closely related to placement, and an excellent placement scheme can improve chip area utilization, timing performance, and routing capabilities. 3. **Limitations of traditional methods**: - Traditional placement and routing solvers rely on heuristic algorithms or expert - designed algorithms. Although these methods are effective, they face scalability problems when dealing with large - scale integrated circuits (IC). - As the circuit density increases, the feature size of modern VLSI design keeps decreasing, and on - chip interconnections become more and more complex, which leads to congestion problems and minimizing wire length becoming a crucial task in every design stage. 4. **Application prospects of machine learning**: - Machine learning, through its data - driven characteristics, can provide more scalable solutions with less reliance on prior knowledge. - The paper reviews the latest progress of machine learning in the fields of placement and routing in recent years and discusses the challenges and opportunities for future research. In conclusion, this paper aims to explore how to use machine learning techniques to overcome the limitations of traditional methods in the placement and routing tasks in modern chip design, so as to achieve a more efficient and optimized design flow.

Towards Machine Learning for Placement and Routing in Chip Design: a Methodological Overview

On Joint Learning for Solving Placement and Routing in Chip Design

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

A Survey on Machine Learning-Based Routing for VLSI Physical Design

Intelligent and kernelized placement: A survey

The Policy-gradient Placement and Generative Routing Neural Networks for Chip Design

Hierarchical Reinforcement Learning for Chip-Macro Placement in Integrated Circuit

Non-Overlapping Placement of Macro Cells based on Reinforcement Learning in Chip Design

Global Placement with Deep Learning-Enabled Explicit Routability Optimization

ML-based Routing Guidance Generation Data Pre-processing Trained ML Model Guided AMS Routing Guided Detailed Routing Post-processing ML Model Training AMS Circuit Layout Database Dashed : Training Solid : Inference Fr am ew or k Placement Routed Layout

Chip Placement with Deep Reinforcement Learning

Chiplet Placement Order Exploration Based on Learning to Rank with Graph Representation

A Machine Learning Mapping Algorithm for NoC Optimization

Global Routing Under a Congestion-Aware Reinforcement Learning Model

An Accurate Detailed Routing Routability Prediction Model in Placement

Implementation of Machine Learning In VLSI Global Routing

A Machine Learning Approach to Routing

RoutePlacer: An End-to-End Routability-Aware Placer with Graph Neural Network

Machine Learning for Yield Learning and Optimization

Detailed Placement in Advanced Technology Nodes: A Survey

Routing and Placement of Macros using Deep Reinforcement Learning