Driven by Moore’s law, the chip design complexity is steadily increasing. Electronic Design Automation (EDA) has been able to cope with the challenging very large-scale integration process, assuring scalability, reliability, and proper time-to-market. However, EDA approaches are time and resource demanding, and they often do not guarantee optimal solutions. To alleviate these, Machine Learning (ML) has been incorporated into many stages of the design flow, such as in placement and routing. Many solutions employ Euclidean data and ML techniques without considering that many EDA objects are represented naturally as graphs. The trending Graph Neural Networks (GNNs) are an opportunity to solve EDA problems directly using graph structures for circuits, intermediate Register Transfer Levels, and netlists. In this article, we present a comprehensive review of the existing works linking the EDA flow for chip design and GNNs. We map those works to a design pipeline by defining graphs, tasks, and model types. Furthermore, we analyze their practical implications and outcomes. We conclude by summarizing challenges faced when applying GNNs within the EDA design flow.
Embedded systems are becoming increasingly complex, which has led to a productivity crisis in their design and verification. Although conventional design automation coupled with IP and platform reuse techniques have led to leaps in design productivity improvement, they face fundamental limits given that most design optimization and verification problems remain NP-hard and that reuse of pre-designed IP blocks and platforms inherently limits flexibility and optimality. At the same time, machine learning (ML) has recently made unprecedented advances and created phenomenal impact in various computing applications. In particular, application of ML techniques as a way to extract knowledge and learn from existing design, optimization and verification data has recently seen a lot of excitement and promise at lower physical and integrated circuit levels of abstraction. Using ML has the potential to similarly close the complexity gap in embedded system design, but corresponding ML-based approaches for embedded system optimization and verification at higher levels of abstraction are still at their infancy. This paper presents the current state of the art, along with opportunities and open challenges, in the application of ML methods for embedded system design and optimization. We discuss design and optimization at different levels of abstraction ranging from system-level modeling and optimization and high-level synthesis to RTL and micro-architecture design, bringing together perspectives from different communities in both academia and industry.
Data selection is essential for any data-based optimization technique, such as Reinforcement Learning. State-of-the-art sampling strategies for the experience replay buffer improve the performance of the Reinforcement Learning agent. However, they do not incorporate uncertainty in the Q-Value estimation. Consequently, they cannot adapt the sampling strategies, including exploration and exploitation of transitions, to the complexity of the task. To address this, this paper proposes a new sampling strategy that leverages the exploration-exploitation trade-off. This is enabled by the uncertainty estimation of the Q-Value function, which guides the sampling to explore more significant transitions and, thus, learn a more efficient policy. Experiments on classical control environments demonstrate stable results across various environments. They show that the proposed method outperforms state-of-the-art sampling strategies for dense rewards w.r.t. convergence and peak performance by 26% on average.