← Back to the feed
Energy & materialsResearch findings

Neural Network-Based Transistor Model Accelerates 3nm Nanosheet SPICE Simulation

Original published 2026.10.26 05:14 KSTAdded to Blackdeck 2026.10.26 06:56 KST

What has changed?

Development of neural network-based transistor models that accurately predict drain current and charge in 3nm nanosheet devices

Key points

  1. Neural network models accurately predict drain current and charge in 3nm nanosheet transistors
  2. Achieve up to 8x speedup in standard cell library characterization
  3. Delay and power prediction errors remain below 0.11% and 0.76% respectively

Connect it to your work

NN-based models achieve sub-0.11% delay error and 0.76% power error, with up to 8x faster simulation speed during standard cell library characterization

What remains unproven

High precision required for complex circuit convergence; real-world EDA tool flow integration remains under evaluation

Originals used for this note

Links open the original publisher in a new tab.

  1. semiconductorsNeural Networks Speed 3nm Nanosheet SPICE Simulation (TU Munich)