Abstract
The rise of deep neural network (DNN) accelerators demands physical design tools that recognize spatial regularity patterns. Traditional placers, unaware of the regularity of spatial arrays, produce suboptimal solutions. This work proposes RegPlace, a regularity-aware placement algorithm for full-system DNN accelerators that automatically identifies processing element using graph convolutional networks and employs a variance-based soft regularity loss to guide optimization. Compared to state-of-the-art methods, our approach achieves up to 6% wirelength reduction while maintaining comparable runtime, with post-placement metrics further confirming its effectiveness.
Funding Information
  • Research Grants Council of Hong Kong SAR (CUHK14210723)
  • Research Grants Council of Hong Kong SAR (CUHK14211824)