A 1036 TOp/s/W, 12.2 mW, 2.72 μJ/Inference All Digital TNN Accelerator in 22 nm FDX Technology for TinyML Applications

Moritz Scherer, Alfio Di Mauro, Georg Rutishauser, Tim Fischer, Luca Benini

Abstract

Tiny Machine Learning (TinyML) applications impose μJ/Inference constraints, with maximum power consumption of a few tens of mW. It is extremely challenging to meet these requirement at a reasonable accuracy level. In this work, we address this challenge with a flexible, fully digital Ternary Neural Network (TNN) accelerator in a RISC-V-based SoC. The design achieves 2.72 μJ/Inference, 12.2 mW, 3200 Inferences/sec at 0.5 V for a non-trivial 9-layer, 96 channels-per-layer network with CIFAR-10 accuracy of 86 %. The peak energy efficiency is 1036 TOp/s/W, outperforming the state-of-the-art in silicon-proven TinyML accelerators by 1.67x.
BibTeX
@inproceedings{scherer2022top,
  author = {Moritz Scherer and Alfio Di Mauro and Georg Rutishauser and Tim Fischer and Luca Benini},
  title = {{A 1036 TOp/s/W, 12.2 mW, 2.72 μJ/Inference All Digital TNN Accelerator in 22 nm FDX Technology for TinyML Applications}},
  booktitle = {COOL CHIPS},
  year = {2022},
  doi = {10.1109/coolchips54332.2022.9772668},
}

Loading related content...

Figure