What is a 3D Network-on-Chip?

  • Authors:
  • Ciprian Seiculescu;Federico Angiolin;Giovanni De Micheli

  • Affiliations:
  • EPFL;iNOCs;EPFL

  • Venue:
  • ACM SIGDA Newsletter
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

While lithographic advances have made it possible to manufacture chips with ever smaller features (Moore's Law), the cost of the required machinery has been steadily increasing, and speculation is that finer geometries will become, if not technically infeasible, impractically expensive. At the same time, the "More than Moore" paradigm [1] has been advocating the creation of chip value through the integration of functionality beyond standard digital CMOS, for example sensing, radiofrequency, optoelectronics. These two trends meet in "3D" chip integration, whereby multiple dies, independently manufactured in possibly very heterogeneous technologies, are stacked together to provide a large amount of functionality in a small package footprint.