Abstract:Process variations and device aging impose profound challenges for circuit designers. Without a precise understanding of the impact of variations on the delay of circuit paths, guardbands, which keep timing violations at bay, cannot be correctly estimated. This problem is exacerbated for advanced technology nodes, where transistor dimensions reach atomic levels and established margins are severely constrained. Hence, traditional worst-case analysis becomes impractical, resulting in intolerable performance overheads. Contrarily, process-variation/aging-aware static timing analysis (STA) equips designers with accurate statistical delay distributions. Timing guardbands that are small, yet sufficient, can then be effectively estimated. However, such analysis is costly as it requires intensive Monte-Carlo simulations. Further, it necessitates access to confidential physics-based aging models to generate the standard-cell libraries required for STA. In this work, we employ graph neural networks (GNNs) to accurately estimate the impact of process variations and device aging on the delay of any path within a circuit. Our proposed GNN4REL framework empowers designers to perform rapid and accurate reliability estimations without accessing transistor models, standard-cell libraries, or even STA; these components are all incorporated into the GNN model via training by the foundry. Specifically, GNN4REL is trained on a FinFET technology model that is calibrated against industrial 14nm measurement data. Through our extensive experiments on EPFL and ITC-99 benchmarks, as well as RISC-V processors, we successfully estimate delay degradations of all paths -- notably within seconds -- with a mean absolute error down to 0.01 percentage points.

Aging-Aware Logic Restructure Acceleration Based on Heterogeneous Graph Learning

Fast and Accurate Aging-Aware Cell Timing Model Via Graph Learning

Aging-Aware Critical Path Selection Via Graph Attention Networks

Fast Aging-Aware Timing Analysis Framework with Temporal-Spatial Graph Neural Network

Aging-Aware Gate-Level Modeling for Circuit Reliability Analysis

Multi-View Graph Learning for Path-Level Aging-Aware Timing Prediction

Aging Aware Retraining for Memristor-based Neuromorphic Computing

Analog IC Aging-induced Degradation Estimation Via Heterogeneous Graph Convolutional Networks

Novel Critical Gate-Based Circuit Path-Level NBTI-Aware Aging Circuit Degradation Prediction.

Deep H-GCN: Fast Analog IC Aging-Induced Degradation Estimation

GNN4REL: Graph Neural Networks for Predicting Circuit Reliability Degradation

An Aging-Aware Routing Algorithm for Improving Reliability in NoC-Based Multicore Systems.

Reliability-Enhanced Circuit Design Flow Based on Approximate Logic Synthesis

Aging-Aware Task Scheduling for Mesh-Based Network-on-Chips under Aging Effect

A Machine Learning Based Reliability Analysis of Negative Bias Temperature Instability (NBTI) Compliant Design for Ultra Large Scale Digital Integrated Circuit

Statistical Reliability Analysis under Process Variation and Aging Effects

LOSTIN: Logic Optimization Via Spatio-Temporal Information with Hybrid Graph Models

ML-based AIG Timing Prediction to Enhance Logic Optimization

Timing-Driven Technology Mapping Approximation Based on Reinforcement Learning

Modeling and Predicting Transistor Aging under Workload Dependency using Machine Learning

Simulation Study of Aging in Cmos Binary Adders