Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

INSIGHT: Universal Neural Simulator for Analog Circuits Harnessing Autoregressive Transformers

Souradip Poddar, Youngmin Oh, Yao Lai, Hanqing Zhu, Bosun Hwang, David Z. Pan

TL;DR

INSIGHT addresses the cost and latency of SPICE-based analog circuit design by introducing a GPU-accelerated, technology-agnostic universal neural simulator built on a decoder-only Transformer for autoregressive performance prediction. It frames circuit prediction as sequential generation of performance metrics, yielding microsecond-scale inference and the ability to forecast costly transient specs from cheaper metrics. The paper also introduces INSIGHT-M, a model-based batch RL sizing framework that achieves remarkable sample efficiency, requiring fewer than 20 real-time simulations across diverse circuits and technologies. Collectively, INSIGHT provides a high-fidelity surrogate for SPICE with strong cross-technology generalization and practical impact on fast, reliable analog design optimization.

Abstract

Analog front-end design heavily relies on specialized human expertise and costly trial-and-error simulations, which motivated many prior works on analog design automation. However, efficient and effective exploration of the vast and complex design space remains constrained by the time-consuming nature of SPICE simulations, making effective design automation a challenging endeavor. In this paper, we introduce INSIGHT, a GPU-powered, technology-agnostic, effective universal neural simulator in the analog front-end design automation loop. INSIGHT accurately predicts the performance metrics of analog circuits across various technologies with just a few microseconds of inference time. Notably, its autoregressive capabilities enable INSIGHT to accurately predict simulation-costly critical transient specifications leveraging less expensive performance metric information. The low cost and high fidelity feature make INSIGHT a good substitute for standard simulators in analog front-end optimization frameworks. INSIGHT is compatible with any optimization framework, facilitating enhanced design space exploration for sample efficiency through sophisticated offline learning and adaptation techniques. Our experiments demonstrate that INSIGHT-M, a model-based batch reinforcement learning sizing framework with INSIGHT as the accurate surrogate, only requires < 20 real-time simulations with 100-1000x lower simulation costs and significant speedup over existing sizing methods.

INSIGHT: Universal Neural Simulator for Analog Circuits Harnessing Autoregressive Transformers

TL;DR

INSIGHT addresses the cost and latency of SPICE-based analog circuit design by introducing a GPU-accelerated, technology-agnostic universal neural simulator built on a decoder-only Transformer for autoregressive performance prediction. It frames circuit prediction as sequential generation of performance metrics, yielding microsecond-scale inference and the ability to forecast costly transient specs from cheaper metrics. The paper also introduces INSIGHT-M, a model-based batch RL sizing framework that achieves remarkable sample efficiency, requiring fewer than 20 real-time simulations across diverse circuits and technologies. Collectively, INSIGHT provides a high-fidelity surrogate for SPICE with strong cross-technology generalization and practical impact on fast, reliable analog design optimization.

Abstract

Analog front-end design heavily relies on specialized human expertise and costly trial-and-error simulations, which motivated many prior works on analog design automation. However, efficient and effective exploration of the vast and complex design space remains constrained by the time-consuming nature of SPICE simulations, making effective design automation a challenging endeavor. In this paper, we introduce INSIGHT, a GPU-powered, technology-agnostic, effective universal neural simulator in the analog front-end design automation loop. INSIGHT accurately predicts the performance metrics of analog circuits across various technologies with just a few microseconds of inference time. Notably, its autoregressive capabilities enable INSIGHT to accurately predict simulation-costly critical transient specifications leveraging less expensive performance metric information. The low cost and high fidelity feature make INSIGHT a good substitute for standard simulators in analog front-end optimization frameworks. INSIGHT is compatible with any optimization framework, facilitating enhanced design space exploration for sample efficiency through sophisticated offline learning and adaptation techniques. Our experiments demonstrate that INSIGHT-M, a model-based batch reinforcement learning sizing framework with INSIGHT as the accurate surrogate, only requires < 20 real-time simulations with 100-1000x lower simulation costs and significant speedup over existing sizing methods.
Paper Structure (17 sections, 3 equations, 3 figures, 6 tables)

This paper contains 17 sections, 3 equations, 3 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Analog Circuit Basics and Metrics
  4. Transformer Architectures
  5. Figure of Merit
  6. INSIGHT Framework and Algorithms
  7. INSIGHT for Modeling
  8. INSIGHT for Analog Sizing
  9. INSIGHT Pre-Training
  10. INSIGHT for Optimization
  11. Experiments
  12. Benchmark Circuits and Technologies
  13. INSIGHT Performance
  14. Prediction Accuracy
  15. Inference Speed
  16. ...and 2 more sections

Figures (3)

  • Figure 1: Autoregressive Performance Prediction in INSIGHT
  • Figure 2: Automatic Analog Sizing Framework
  • Figure 3: Benchmark Circuits: (a) 2-Stage Miller-Compensated OPAMP with NMOS Input Pair, (b) 2-Stage Miller-Compensated OPAMP with PMOS Input Pair, (c) 2-Stage TIA, (d) 3-Stage TIA, (e) Comparator, (f) Level Shifter.