Neural networks speed up transistor simulations with high precision
NN-Based Transistor Modeling for Ultra-Fast SPICE Simulation and Standard Cell Library Characterization
Emerging TechnologiesMachine Learning
Summary
Simulating how tiny transistors behave in computer chips usually takes a long time and needs very precise models. The authors developed neural network models that can predict transistor behavior very accurately but much faster than traditional methods. They tested these models on complex chip design tasks and found they deliver nearly the same accuracy but run up to eight times faster. This approach could help engineers design chips more quickly while keeping results reliable.
What this means in practice
- •For chip design engineers: Accelerate standard cell library characterization with faster transistor models that maintain high accuracy for timing and power analysis.
- •For eda tool developers: Integrate neural network-based transistor models into simulation software to improve circuit simulation speed and convergence reliability.
Authors
Rodion Novkin, Hussam Amrouch
Abstract
Neural network (NN)-based transistor models have proven to be a promising solution to accelerate device modeling. Although these models demonstrated remarkable speedup in circuit simulations, they were not rigorously tested in complex tasks within standard EDA tool flows for chip design, which require very high transistor model precision to provide accurate results and converge properly. To investigate their full potential and promise, we develop NN-based transistor models that accurately predict the drain current and various charges for a 3nm nanosheet, and employ the models for a standard cell library characterization. This enables us to evaluate the accuracy and speedup under a very wide range of SPICE simulations, benchmarking our models against an industry-standard implementation. Since standard cell library characterization involves thousands of SPICE simulations, it serves as a great benchmark for transistor model quality and consistency. To further evaluate model accuracy, we also perform analog- and digital-level simulations for various complex circuits. Our experiments reveal that NN-based models are a valid alternative to existing transistor compact models, capable of producing accurate delay and power estimates with a much higher speed and convergence time compared to the conventional Verilog-A-based industry-standard (BSIM-CMG) transistor compact model. Our NNs achieve sub-0.11% and 0.76% errors for delay and total power evaluation on a wide range of circuits, while taking up to 8 times less time during the standard cell library characterization. Additionally, we provide comprehensive information and guidance on NN accuracy, unveiling the relationship between NN size and its effect on circuit simulation precision.