Lorecast: Layout-Aware Performance and Power Forecasting from Natural Language

  • 2025-04-22 20:01:42
  • Runzhi Wang, Prianka Sengupta, Cristhian Roman-Vicharra, Yiran Chen, Jiang Hu
  • 0

Abstract

In chip design planning, obtaining reliable performance and power forecastsfor various design options is of critical importance. Traditionally, thisinvolves using system-level models, which often lack accuracy, or trialsynthesis, which is both labor-intensive and time-consuming. We introduce a newmethodology, called Lorecast, which accepts English prompts as input to rapidlygenerate layout-aware performance and power estimates. This approach bypassesthe need for HDL code development and synthesis, making it both fast anduser-friendly. Experimental results demonstrate that Lorecast achieves accuracywithin a few percent of error compared to post-layout analysis, whilesignificantly reducing turnaround time.

 

Quick Read (beta)

loading the full paper ...